publishedAt
timestamp[ns] | title
string | thumbnail
string | numComments
int64 | submittedBy
dict | isAuthorParticipating
bool | mediaUrls
list | paper_id
string | paper_authors
list | paper_publishedAt
timestamp[ns] | paper_title
string | paper_summary
string | paper_upvotes
int64 | paper_discussionId
string | paper_projectPage
string | paper_githubRepo
string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2025-02-20T22:33:22.039000 | SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features | 7 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | true | null | 2502.14786 | [
{
"_id": "67b7ed0d58f6b70b18dda7b4",
"hidden": false,
"name": "Michael Tschannen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:23:44.125Z",
"user": {
"_id": "6489893e1ec8356ba5bb9777",
"avatarUrl": "/avatars/54354c1e5774cadd1d83d42054e9d96b.svg",
"fullname": "Michael Tschannen",
"isPro": false,
"type": "user",
"user": "mitsch"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7b5",
"hidden": false,
"name": "Alexey Gritsenko",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:23:50.440Z",
"user": {
"_id": "62d8f9887b8dc0ba17271415",
"avatarUrl": "/avatars/12ec78d34fd849bad44217b212f31e98.svg",
"fullname": "Alexey Gritsenko",
"isPro": false,
"type": "user",
"user": "AlexeyG"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7b6",
"hidden": false,
"name": "Xiao Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ed0d58f6b70b18dda7b7",
"hidden": false,
"name": "Muhammad Ferjad Naeem",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-26T15:37:42.802Z",
"user": {
"_id": "67bf33a512368ec2fad4fe29",
"avatarUrl": "/avatars/ea5c03744ec1c2bcc0e6c13efc8f7ddc.svg",
"fullname": "Muhammad Ferjad Naeem",
"isPro": false,
"type": "user",
"user": "ferjad"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7b8",
"hidden": false,
"name": "Ibrahim Alabdulmohsin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:24:03.702Z",
"user": {
"_id": "630545da20668afe24860235",
"avatarUrl": "/avatars/5d82be2e7412bff1af15cc5eafa60b7d.svg",
"fullname": "Ibrahim Alabdulmohsin",
"isPro": false,
"type": "user",
"user": "ibomohsin"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7b9",
"hidden": false,
"name": "Nikhil Parthasarathy",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ed0d58f6b70b18dda7ba",
"hidden": false,
"name": "Talfan Evans",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:24:14.273Z",
"user": {
"_id": "66be3074dac7d71326f613cf",
"avatarUrl": "/avatars/012a531aad3eb1a2751bb3c31a619bf5.svg",
"fullname": "Talfan Evans",
"isPro": false,
"type": "user",
"user": "talfanevans"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7bb",
"hidden": false,
"name": "Lucas Beyer",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:44:51.778Z",
"user": {
"_id": "642d334ff65714b4585f2de4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/642d334ff65714b4585f2de4/gxBynq5KyoUP0VlAQD3-w.jpeg",
"fullname": "Lucas Beyer",
"isPro": false,
"type": "user",
"user": "giffmana"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7bc",
"hidden": false,
"name": "Ye Xia",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ed0d58f6b70b18dda7bd",
"hidden": false,
"name": "Basil Mustafa",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:44:44.933Z",
"user": {
"_id": "63cfcad6e23b90128c66685c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674562206311-noauth.png",
"fullname": "Basil Mustafa",
"isPro": false,
"type": "user",
"user": "BasilMustafa"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7be",
"hidden": false,
"name": "Olivier Hénaff",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:44:38.156Z",
"user": {
"_id": "64f30d8ceb5f2982081db604",
"avatarUrl": "/avatars/eedf65a104d099d8a60bbffe69bc2571.svg",
"fullname": "Olivier Henaff",
"isPro": false,
"type": "user",
"user": "olivierhenaff"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7bf",
"hidden": false,
"name": "Jeremiah Harmsen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:44:32.302Z",
"user": {
"_id": "65d77c0e1e7686c460255fda",
"avatarUrl": "/avatars/1d26a7a7ffdc5ca9e67b97030f21b098.svg",
"fullname": "Jeremiah Harmsen",
"isPro": false,
"type": "user",
"user": "jharmsen"
}
},
{
"_id": "67b7ed0d58f6b70b18dda7c0",
"hidden": false,
"name": "Andreas Steiner",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ed0d58f6b70b18dda7c1",
"hidden": false,
"name": "Xiaohua Zhai",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:44:07.595Z",
"user": {
"_id": "65dcd90082bddd501f68174b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/M2bc9PyKeFs1cCXjTfGGq.jpeg",
"fullname": "Xiaohua Zhai",
"isPro": false,
"type": "user",
"user": "xiaohuazhai"
}
}
]
| 2025-02-20T18:08:29 | SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic
Understanding, Localization, and Dense Features | We introduce SigLIP 2, a family of new multilingual vision-language encoders
that build on the success of the original SigLIP. In this second iteration, we
extend the original image-text training objective with several prior,
independently developed techniques into a unified recipe -- this includes
captioning-based pretraining, self-supervised losses (self-distillation, masked
prediction) and online data curation. With these changes, SigLIP 2 models
outperform their SigLIP counterparts at all model scales in core capabilities,
including zero-shot classification, image-text retrieval, and transfer
performance when extracting visual representations for Vision-Language Models
(VLMs). Furthermore, the new training recipe leads to significant improvements
on localization and dense prediction tasks. We also train variants which
support multiple resolutions and preserve the input's native aspect ratio.
Finally, we train on a more diverse data-mixture that includes de-biasing
techniques, leading to much better multilingual understanding and improved
fairness. To allow users to trade off inference cost with performance, we
release model checkpoints at four sizes: ViT-B (86M), L (303M), So400m (400M),
and g (1B). | 124 | 67b7ed0e58f6b70b18dda7f4 | null | null |
|
2025-02-20T22:30:51.542000 | RelaCtrl: Relevance-Guided Efficient Control for Diffusion Transformers | 2 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | false | null | 2502.14377 | [
{
"_id": "67b7f350357c2729ac216494",
"hidden": false,
"name": "Ke Cao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T15:08:00.737Z",
"user": {
"_id": "66e4077369d1083dd97c7cd8",
"avatarUrl": "/avatars/0dad41e3e2f38f89b7b21c12d673f432.svg",
"fullname": "Ke Cao",
"isPro": false,
"type": "user",
"user": "kecao"
}
},
{
"_id": "67b7f350357c2729ac216495",
"hidden": false,
"name": "Jing Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f350357c2729ac216496",
"hidden": false,
"name": "Ao Ma",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T15:07:53.347Z",
"user": {
"_id": "65292034825f4bba46c78581",
"avatarUrl": "/avatars/7f212daaa20ab0d7405e9d6351ec308c.svg",
"fullname": "Ao Ma",
"isPro": false,
"type": "user",
"user": "AoMa"
}
},
{
"_id": "67b7f350357c2729ac216497",
"hidden": false,
"name": "Jiasong Feng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f350357c2729ac216498",
"hidden": false,
"name": "Zhanjie Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T15:07:37.282Z",
"user": {
"_id": "640496240ab5e22719f31719",
"avatarUrl": "/avatars/be957a1b0e1c037f03f1439d6142e9ce.svg",
"fullname": "Zhanjie Zhang",
"isPro": false,
"type": "user",
"user": "zhangzhanjay"
}
},
{
"_id": "67b7f350357c2729ac216499",
"hidden": false,
"name": "Xuanhua He",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T15:07:32.005Z",
"user": {
"_id": "64375035946fb080c6fc4551",
"avatarUrl": "/avatars/8dba2c8911726656d4088862a2b8fe7c.svg",
"fullname": "Xuanhua He",
"isPro": false,
"type": "user",
"user": "Alexhe101"
}
},
{
"_id": "67b7f350357c2729ac21649a",
"hidden": false,
"name": "Shanyuan Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T15:07:26.104Z",
"user": {
"_id": "666197e302a5e5f4377a545c",
"avatarUrl": "/avatars/6d6ef51cc403ffccc81265ad8adf43bc.svg",
"fullname": "liu",
"isPro": false,
"type": "user",
"user": "shanyuanLiu"
}
},
{
"_id": "67b7f350357c2729ac21649b",
"hidden": false,
"name": "Bo Cheng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f350357c2729ac21649c",
"hidden": false,
"name": "Dawei Leng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f350357c2729ac21649d",
"hidden": false,
"name": "Yuhui Yin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f350357c2729ac21649e",
"hidden": false,
"name": "Jie Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-20T09:10:05 | RelaCtrl: Relevance-Guided Efficient Control for Diffusion Transformers | The Diffusion Transformer plays a pivotal role in advancing text-to-image and
text-to-video generation, owing primarily to its inherent scalability. However,
existing controlled diffusion transformer methods incur significant parameter
and computational overheads and suffer from inefficient resource allocation due
to their failure to account for the varying relevance of control information
across different transformer layers. To address this, we propose the
Relevance-Guided Efficient Controllable Generation framework, RelaCtrl,
enabling efficient and resource-optimized integration of control signals into
the Diffusion Transformer. First, we evaluate the relevance of each layer in
the Diffusion Transformer to the control information by assessing the
"ControlNet Relevance Score"-i.e., the impact of skipping each control layer on
both the quality of generation and the control effectiveness during inference.
Based on the strength of the relevance, we then tailor the positioning,
parameter scale, and modeling capacity of the control layers to reduce
unnecessary parameters and redundant computations. Additionally, to further
improve efficiency, we replace the self-attention and FFN in the commonly used
copy block with the carefully designed Two-Dimensional Shuffle Mixer (TDSM),
enabling efficient implementation of both the token mixer and channel mixer.
Both qualitative and quantitative experimental results demonstrate that our
approach achieves superior performance with only 15% of the parameters and
computational complexity compared to PixArt-delta. More examples are available
at https://relactrl.github.io/RelaCtrl/. | 12 | 67b7f354357c2729ac216582 | null | null |
|
2025-02-20T22:19:05.902000 | Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning | 5 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | false | null | 2502.14768 | [
{
"_id": "67b7f08c357c2729ac20a81b",
"hidden": false,
"name": "Tian Xie",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f08c357c2729ac20a81c",
"hidden": false,
"name": "Zitian Gao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:50:06.783Z",
"user": {
"_id": "641ddac5be3bd3a5a06ed4a4",
"avatarUrl": "/avatars/14969dff861d53b0a75305606495eca7.svg",
"fullname": "zitian gao",
"isPro": false,
"type": "user",
"user": "zgao3186"
}
},
{
"_id": "67b7f08c357c2729ac20a81d",
"hidden": false,
"name": "Qingnan Ren",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f08c357c2729ac20a81e",
"hidden": false,
"name": "Haoming Luo",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:23:21.219Z",
"user": {
"_id": "6501535970b6b05c5af84383",
"avatarUrl": "/avatars/a827dfa11589cabd6868c617eeecbbba.svg",
"fullname": "Haoming Luo",
"isPro": false,
"type": "user",
"user": "Resnet-340"
}
},
{
"_id": "67b7f08c357c2729ac20a81f",
"hidden": false,
"name": "Yuqian Hong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f08c357c2729ac20a820",
"hidden": false,
"name": "Bryan Dai",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:49:20.858Z",
"user": {
"_id": "67090a395f4ed2ff1f0b3658",
"avatarUrl": "/avatars/03744a2fcbcb0e8074c04bad83b3e34c.svg",
"fullname": "Zhenbang Dai",
"isPro": false,
"type": "user",
"user": "BryanDai"
}
},
{
"_id": "67b7f08c357c2729ac20a821",
"hidden": false,
"name": "Joey Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f08c357c2729ac20a822",
"hidden": false,
"name": "Kai Qiu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7f08c357c2729ac20a823",
"hidden": false,
"name": "Zhirong Wu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:48:55.865Z",
"user": {
"_id": "67034414f2b11c7dd251e232",
"avatarUrl": "/avatars/6b741ac2eab48c6f72185342f9af7d1f.svg",
"fullname": "wzr",
"isPro": false,
"type": "user",
"user": "wuzhirong"
}
},
{
"_id": "67b7f08c357c2729ac20a824",
"hidden": false,
"name": "Chong Luo",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:48:44.370Z",
"user": {
"_id": "676a328148d749b7086782d0",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/Tt7u8l8f_1oVBWmBp7tkm.png",
"fullname": "Chong Luo",
"isPro": false,
"type": "user",
"user": "cluo-ms"
}
}
]
| 2025-02-20T17:49:26 | Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement
Learning | Inspired by the success of DeepSeek-R1, we explore the potential of
rule-based reinforcement learning (RL) in large reasoning models. To analyze
reasoning dynamics, we use synthetic logic puzzles as training data due to
their controllable complexity and straightforward answer verification. We make
some key technical contributions that lead to effective and stable RL training:
a system prompt that emphasizes the thinking and answering process, a stringent
format reward function that penalizes outputs for taking shortcuts, and a
straightforward training recipe that achieves stable convergence. Our 7B model
develops advanced reasoning skills-such as reflection, verification, and
summarization-that are absent from the logic corpus. Remarkably, after training
on just 5K logic problems, it demonstrates generalization abilities to the
challenging math benchmarks AIME and AMC. | 44 | 67b7f08e357c2729ac20a88f | null | null |
|
2025-02-20T22:15:33.133000 | SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines | 10 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | true | null | 2502.14739 | [
{
"_id": "67b7efc26348a1df80a8ae53",
"hidden": false,
"name": "M-A-P Team",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae54",
"hidden": false,
"name": "Xinrun Du",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T14:42:53.525Z",
"user": {
"_id": "654907a4a1faff97850c4eff",
"avatarUrl": "/avatars/458c90151614bc7f116943b6e67d6b8a.svg",
"fullname": "du",
"isPro": false,
"type": "user",
"user": "dododododo"
}
},
{
"_id": "67b7efc26348a1df80a8ae55",
"hidden": false,
"name": "Yifan Yao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae56",
"hidden": false,
"name": "Kaijing Ma",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:20:18.121Z",
"user": {
"_id": "65eb65722fbf6807134a636c",
"avatarUrl": "/avatars/282920ff99c8d83cdac5fd6ee507096a.svg",
"fullname": "Kaijing Ma",
"isPro": false,
"type": "user",
"user": "mkj69"
}
},
{
"_id": "67b7efc26348a1df80a8ae57",
"hidden": false,
"name": "Bingli Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:20:24.792Z",
"user": {
"_id": "658d0a228cff48d3a4612689",
"avatarUrl": "/avatars/70e297c6cb12d1bdde6d91c23f590b63.svg",
"fullname": "Bingli Wang",
"isPro": false,
"type": "user",
"user": "BingliW"
}
},
{
"_id": "67b7efc26348a1df80a8ae58",
"hidden": false,
"name": "Tianyu Zheng",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:24.002Z",
"user": {
"_id": "64ab99dcb76bfd863eba64c1",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64ab99dcb76bfd863eba64c1/UBXwDPx17X-gl-SzBPvrc.jpeg",
"fullname": "TY.Zheng",
"isPro": false,
"type": "user",
"user": "aaabiao"
}
},
{
"_id": "67b7efc26348a1df80a8ae59",
"hidden": false,
"name": "Kang Zhu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:20:45.485Z",
"user": {
"_id": "6578265ddea7e2122d02f6ba",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6578265ddea7e2122d02f6ba/Bh6JjoVF5ceLSjV7Z7nTk.jpeg",
"fullname": "kang zhu",
"isPro": false,
"type": "user",
"user": "kangz"
}
},
{
"_id": "67b7efc26348a1df80a8ae5a",
"hidden": false,
"name": "Minghao Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:25.894Z",
"user": {
"_id": "6417d9ea8f689506e7148417",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6417d9ea8f689506e7148417/bAYcruWNw4WvmuQcGgcwC.jpeg",
"fullname": "minghao",
"isPro": false,
"type": "user",
"user": "Liam-Liu"
}
},
{
"_id": "67b7efc26348a1df80a8ae5b",
"hidden": false,
"name": "Yiming Liang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:21:08.553Z",
"user": {
"_id": "6555e8d8a0c34cd61a6b9ce3",
"avatarUrl": "/avatars/71dc562cef4bd42f6b762f036357c800.svg",
"fullname": "yimingliang",
"isPro": false,
"type": "user",
"user": "yimingliang"
}
},
{
"_id": "67b7efc26348a1df80a8ae5c",
"hidden": false,
"name": "Xiaolong Jin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae5d",
"hidden": true,
"name": "Zhenlin Wei",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:21:31.565Z",
"user": {
"_id": "67375a6ae6b1d15ff5359a54",
"avatarUrl": "/avatars/9d32d9e3bfb43b8d001c6ddeae720ec5.svg",
"fullname": "weizhenlin",
"isPro": false,
"type": "user",
"user": "vzl123"
}
},
{
"_id": "67b7efc26348a1df80a8ae5e",
"hidden": false,
"name": "Chujie Zheng",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:34.124Z",
"user": {
"_id": "610b70452719facd4ea85e28",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/610b70452719facd4ea85e28/S7nMy7D0Rxq0VIVblhYDG.jpeg",
"fullname": "Chujie Zheng",
"isPro": false,
"type": "user",
"user": "chujiezheng"
}
},
{
"_id": "67b7efc26348a1df80a8ae5f",
"hidden": false,
"name": "Kaixing Deng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae60",
"hidden": false,
"name": "Shuyue Guo",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae61",
"hidden": false,
"name": "Shian Jia",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae62",
"hidden": false,
"name": "Sichao Jiang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:22:09.140Z",
"user": {
"_id": "675085408119fa5fac3cd7cf",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/0Mrrkzhv0wBggP5kGKtSt.png",
"fullname": "jiangsichao",
"isPro": false,
"type": "user",
"user": "jsc137"
}
},
{
"_id": "67b7efc26348a1df80a8ae63",
"hidden": false,
"name": "Yiyan Liao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:22:15.619Z",
"user": {
"_id": "67a9d186b22571659007a43d",
"avatarUrl": "/avatars/79b06cf0983083b6161374e66a8c51b2.svg",
"fullname": "Yiyan Liao",
"isPro": false,
"type": "user",
"user": "yiyanliao"
}
},
{
"_id": "67b7efc26348a1df80a8ae64",
"hidden": false,
"name": "Rui Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae65",
"hidden": false,
"name": "Qinrui Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae66",
"hidden": false,
"name": "Sirun Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:22:30.478Z",
"user": {
"_id": "67ab7826ab5ebf181a7f78d7",
"avatarUrl": "/avatars/d6baf414011d6df659da4eb58e9d8958.svg",
"fullname": "Sirun Li",
"isPro": false,
"type": "user",
"user": "inorganicwriter"
}
},
{
"_id": "67b7efc26348a1df80a8ae67",
"hidden": false,
"name": "Yizhi Li",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T14:42:51.449Z",
"user": {
"_id": "6382252f54421460665ec501",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6382252f54421460665ec501/gW9fev3T5QPcNq4f9hqB1.jpeg",
"fullname": "Yizhi Li",
"isPro": false,
"type": "user",
"user": "yizhilll"
}
},
{
"_id": "67b7efc26348a1df80a8ae68",
"hidden": false,
"name": "Yunwen Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae69",
"hidden": false,
"name": "Dehua Ma",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae6a",
"hidden": false,
"name": "Yuansheng Ni",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:30.371Z",
"user": {
"_id": "64de37ee5e192985054be575",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64de37ee5e192985054be575/fVV7JQMtp_J3uFqszJJHH.jpeg",
"fullname": "Yuansheng Ni",
"isPro": false,
"type": "user",
"user": "yuanshengni"
}
},
{
"_id": "67b7efc26348a1df80a8ae6b",
"hidden": false,
"name": "Haoran Que",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae6c",
"hidden": false,
"name": "Qiyao Wang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:28.639Z",
"user": {
"_id": "64560618bfdf9c63ce2d658a",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64560618bfdf9c63ce2d658a/GVBWU4yNzRsjdyzKT3z3B.jpeg",
"fullname": "Mathsion Wong",
"isPro": false,
"type": "user",
"user": "QiYao-Wang"
}
},
{
"_id": "67b7efc26348a1df80a8ae6d",
"hidden": false,
"name": "Zhoufutu Wen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae6e",
"hidden": false,
"name": "Siwei Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae6f",
"hidden": false,
"name": "Tianshun Xing",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae70",
"hidden": false,
"name": "Ming Xu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae71",
"hidden": false,
"name": "Zhenzhu Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae72",
"hidden": false,
"name": "Zekun Moore Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae73",
"hidden": false,
"name": "Junting Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae74",
"hidden": false,
"name": "Yuelin Bai",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae75",
"hidden": false,
"name": "Xingyuan Bu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae76",
"hidden": false,
"name": "Chenglin Cai",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:23:17.731Z",
"user": {
"_id": "64f9c21b681224dbe49a2280",
"avatarUrl": "/avatars/df26cc4b4c6105af2c77392db61e3a27.svg",
"fullname": "caichenglin",
"isPro": false,
"type": "user",
"user": "easy4mego"
}
},
{
"_id": "67b7efc26348a1df80a8ae77",
"hidden": false,
"name": "Liang Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae78",
"hidden": false,
"name": "Yifan Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae79",
"hidden": false,
"name": "Chengtuo Cheng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae7a",
"hidden": false,
"name": "Tianhao Cheng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae7b",
"hidden": false,
"name": "Keyi Ding",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae7c",
"hidden": false,
"name": "Siming Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae7d",
"hidden": false,
"name": "Yun Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae7e",
"hidden": false,
"name": "Yaoru Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae7f",
"hidden": false,
"name": "Yizhe Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae80",
"hidden": false,
"name": "Zhaoqun Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae81",
"hidden": false,
"name": "Tianhao Liang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae82",
"hidden": false,
"name": "Chengdong Lin",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T15:15:33.213Z",
"user": {
"_id": "66751c722b487c2e015a1f60",
"avatarUrl": "/avatars/d66a98b625451ccea1b4dfcdaf623304.svg",
"fullname": "lin",
"isPro": false,
"type": "user",
"user": "adams6435"
}
},
{
"_id": "67b7efc26348a1df80a8ae83",
"hidden": false,
"name": "Hongquan Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae84",
"hidden": false,
"name": "Yinghao Ma",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae85",
"hidden": false,
"name": "Zhongyuan Peng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae86",
"hidden": false,
"name": "Zifan Peng",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:20.429Z",
"user": {
"_id": "65adda5299c3bd19c74d6a8d",
"avatarUrl": "/avatars/1ce504b64ab60f375b235ebaf81cafd6.svg",
"fullname": "PENG ZIFAN",
"isPro": false,
"type": "user",
"user": "Ziffer"
}
},
{
"_id": "67b7efc26348a1df80a8ae87",
"hidden": false,
"name": "Qige Qi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae88",
"hidden": false,
"name": "Shi Qiu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae89",
"hidden": false,
"name": "Xingwei Qu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae8a",
"hidden": false,
"name": "Yizhou Tan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae8b",
"hidden": false,
"name": "Zili Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae8c",
"hidden": false,
"name": "Chenqing Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae8d",
"hidden": false,
"name": "Hao Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae8e",
"hidden": false,
"name": "Yiya Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae8f",
"hidden": false,
"name": "Yubo Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae90",
"hidden": false,
"name": "Jiajun Xu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae91",
"hidden": false,
"name": "Kexin Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae92",
"hidden": false,
"name": "Ruibin Yuan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae93",
"hidden": false,
"name": "Yuanhao Yue",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae94",
"hidden": false,
"name": "Tianyang Zhan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae95",
"hidden": false,
"name": "Chun Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae96",
"hidden": false,
"name": "Jingyang Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae97",
"hidden": false,
"name": "Xiyue Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae98",
"hidden": false,
"name": "Xingjian Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae99",
"hidden": false,
"name": "Yue Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae9a",
"hidden": false,
"name": "Yongchi Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae9b",
"hidden": false,
"name": "Xiangyu Zheng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae9c",
"hidden": false,
"name": "Chenghua Zhong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae9d",
"hidden": false,
"name": "Yang Gao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae9e",
"hidden": false,
"name": "Zhoujun Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8ae9f",
"hidden": false,
"name": "Dayiheng Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea0",
"hidden": false,
"name": "Qian Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:32.399Z",
"user": {
"_id": "612ee6a7b960e78c6d2319d4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/612ee6a7b960e78c6d2319d4/2Hu9BaAyXbyh1vt0v1Qui.jpeg",
"fullname": "Qian Liu",
"isPro": false,
"type": "user",
"user": "SivilTaram"
}
},
{
"_id": "67b7efc26348a1df80a8aea1",
"hidden": false,
"name": "Tianyu Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea2",
"hidden": false,
"name": "Shiwen Ni",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea3",
"hidden": false,
"name": "Junran Peng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea4",
"hidden": false,
"name": "Yujia Qin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea5",
"hidden": false,
"name": "Wenbo Su",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea6",
"hidden": false,
"name": "Guoyin Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T10:23:25.946Z",
"user": {
"_id": "6490d4ba1afdee3acd1147f6",
"avatarUrl": "/avatars/ae13c7b21fe9ced7541dcd664d1b94ed.svg",
"fullname": "Guoyin Wang",
"isPro": false,
"type": "user",
"user": "guoyinwang"
}
},
{
"_id": "67b7efc26348a1df80a8aea7",
"hidden": false,
"name": "Shi Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea8",
"hidden": false,
"name": "Jian Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aea9",
"hidden": false,
"name": "Min Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aeaa",
"hidden": false,
"name": "Meng Cao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aeab",
"hidden": false,
"name": "Xiang Yue",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aeac",
"hidden": false,
"name": "Zhaoxiang Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aead",
"hidden": false,
"name": "Wangchunshu Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aeae",
"hidden": false,
"name": "Jiaheng Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:22.185Z",
"user": {
"_id": "65377c30e48353201e6fdda0",
"avatarUrl": "/avatars/a8f803b6f2e598eaee9c52c0d2ddfc16.svg",
"fullname": "Jiaheng Liu",
"isPro": false,
"type": "user",
"user": "CheeryLJH"
}
},
{
"_id": "67b7efc26348a1df80a8aeaf",
"hidden": false,
"name": "Qunshu Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aeb0",
"hidden": false,
"name": "Wenhao Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7efc26348a1df80a8aeb1",
"hidden": false,
"name": "Ge Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-27T09:17:53.865Z",
"user": {
"_id": "638efcf4c67af472d316d424",
"avatarUrl": "/avatars/97a57859d7d87a3a8f1bb41d32a72bc2.svg",
"fullname": "Ge Zhang",
"isPro": false,
"type": "user",
"user": "zhangysk"
}
}
]
| 2025-02-20T17:05:58 | SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines | Large language models (LLMs) have demonstrated remarkable proficiency in
mainstream academic disciplines such as mathematics, physics, and computer
science. However, human knowledge encompasses over 200 specialized disciplines,
far exceeding the scope of existing benchmarks. The capabilities of LLMs in
many of these specialized fields-particularly in light industry, agriculture,
and service-oriented disciplines-remain inadequately evaluated. To address this
gap, we present SuperGPQA, a comprehensive benchmark that evaluates
graduate-level knowledge and reasoning capabilities across 285 disciplines. Our
benchmark employs a novel Human-LLM collaborative filtering mechanism to
eliminate trivial or ambiguous questions through iterative refinement based on
both LLM responses and expert feedback. Our experimental results reveal
significant room for improvement in the performance of current state-of-the-art
LLMs across diverse knowledge domains (e.g., the reasoning-focused model
DeepSeek-R1 achieved the highest accuracy of 61.82% on SuperGPQA), highlighting
the considerable gap between current model capabilities and artificial general
intelligence. Additionally, we present comprehensive insights from our
management of a large-scale annotation process, involving over 80 expert
annotators and an interactive Human-LLM collaborative system, offering valuable
methodological guidance for future research initiatives of comparable scope. | 94 | 67b7efc66348a1df80a8afc8 | null | null |
|
2025-02-20T22:11:45.130000 | AlphaMaze: Enhancing Large Language Models' Spatial Intelligence via GRPO | 2 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | false | null | 2502.14669 | [
{
"_id": "67b7eeddaf9f1b1bd95b878b",
"hidden": false,
"name": "Alan Dao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T15:03:59.165Z",
"user": {
"_id": "62d7b2339b629105a5d6888a",
"avatarUrl": "/avatars/c3f164fde6b8f9a671890e08ce8a3e75.svg",
"fullname": "Alan Dao",
"isPro": false,
"type": "user",
"user": "alandao"
}
},
{
"_id": "67b7eeddaf9f1b1bd95b878c",
"hidden": false,
"name": "Dinh Bach Vu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-20T16:05:18 | AlphaMaze: Enhancing Large Language Models' Spatial Intelligence via
GRPO | Large Language Models (LLMs) have demonstrated impressive capabilities in
language processing, yet they often struggle with tasks requiring genuine
visual spatial reasoning. In this paper, we introduce a novel two-stage
training framework designed to equip standard LLMs with visual reasoning
abilities for maze navigation. First, we leverage Supervised Fine Tuning (SFT)
on a curated dataset of tokenized maze representations to teach the model to
predict step-by-step movement commands. Next, we apply Group Relative Policy
Optimization (GRPO)-a technique used in DeepSeekR1-with a carefully crafted
reward function to refine the model's sequential decision-making and encourage
emergent chain-of-thought behaviors. Experimental results on synthetically
generated mazes show that while a baseline model fails to navigate the maze,
the SFT-trained model achieves 86% accuracy, and further GRPO fine-tuning
boosts accuracy to 93%. Qualitative analyses reveal that GRPO fosters more
robust and self-corrective reasoning, highlighting the potential of our
approach to bridge the gap between language models and visual spatial tasks.
These findings offer promising implications for applications in robotics,
autonomous navigation, and other domains that require integrated visual and
sequential reasoning. | 11 | 67b7eeddaf9f1b1bd95b87c8 | null | null |
|
2025-02-20T22:08:38.225000 | MLGym: A New Framework and Benchmark for Advancing AI Research Agents | 3 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | false | null | 2502.14499 | [
{
"_id": "67b7ee1dfedfe971271dcca0",
"hidden": false,
"name": "Deepak Nathani",
"status": "extracted_confirmed",
"statusLastChangedAt": "2025-02-21T07:20:46.836Z",
"user": {
"_id": "6114c9fae7a2566ae7d1a1a7",
"avatarUrl": "/avatars/c71ab1850322fcf5ef239cb8d31cb137.svg",
"fullname": "Deepak Nathani",
"isPro": false,
"type": "user",
"user": "dnathani"
}
},
{
"_id": "67b7ee1dfedfe971271dcca1",
"hidden": false,
"name": "Lovish Madaan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dcca2",
"hidden": false,
"name": "Nicholas Roberts",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dcca3",
"hidden": false,
"name": "Nikolay Bashlykov",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T14:42:58.738Z",
"user": {
"_id": "633476edc3cb9eda9328e556",
"avatarUrl": "/avatars/a127e270a606c18623fe00cd723313f6.svg",
"fullname": "Nikolay B",
"isPro": false,
"type": "user",
"user": "bashnick"
}
},
{
"_id": "67b7ee1dfedfe971271dcca4",
"hidden": false,
"name": "Ajay Menon",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dcca5",
"hidden": false,
"name": "Vincent Moens",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dcca6",
"hidden": false,
"name": "Amar Budhiraja",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dcca7",
"hidden": false,
"name": "Despoina Magka",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dcca8",
"hidden": false,
"name": "Vladislav Vorotilov",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dcca9",
"hidden": false,
"name": "Gaurav Chaurasia",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dccaa",
"hidden": false,
"name": "Dieuwke Hupkes",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dccab",
"hidden": false,
"name": "Ricardo Silveira Cabral",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T14:42:55.449Z",
"user": {
"_id": "67b8749ffa8442592bce008e",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/ctbzupAxRhcNXDka75ANi.png",
"fullname": "Ricardo Silveira Cabral",
"isPro": false,
"type": "user",
"user": "rscabral"
}
},
{
"_id": "67b7ee1dfedfe971271dccac",
"hidden": false,
"name": "Tatiana Shavrina",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T14:43:03.777Z",
"user": {
"_id": "60dc2eb60037b630c5df57aa",
"avatarUrl": "/avatars/fbe707b1231a3d9dc6e87ec011e0e738.svg",
"fullname": "Tatiana",
"isPro": false,
"type": "user",
"user": "Shavrina"
}
},
{
"_id": "67b7ee1dfedfe971271dccad",
"hidden": false,
"name": "Jakob Foerster",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dccae",
"hidden": false,
"name": "Yoram Bachrach",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T14:43:01.304Z",
"user": {
"_id": "671b9e3ba54e59639d597fcb",
"avatarUrl": "/avatars/95201b98d1cf2ffa68d23f3b74e387fb.svg",
"fullname": "Yoram Bachrach",
"isPro": false,
"type": "user",
"user": "yorambac"
}
},
{
"_id": "67b7ee1dfedfe971271dccaf",
"hidden": false,
"name": "William Yang Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ee1dfedfe971271dccb0",
"hidden": false,
"name": "Roberta Raileanu",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-21T03:08:15.471Z",
"user": {
"_id": "633e94793a17ab61de8e2b9c",
"avatarUrl": "/avatars/5f2f58ddeed211393660ada6b135f0d5.svg",
"fullname": "Roberta Raileanu",
"isPro": false,
"type": "user",
"user": "rraileanu"
}
}
]
| 2025-02-20T12:28:23 | MLGym: A New Framework and Benchmark for Advancing AI Research Agents | We introduce Meta MLGym and MLGym-Bench, a new framework and benchmark for
evaluating and developing LLM agents on AI research tasks. This is the first
Gym environment for machine learning (ML) tasks, enabling research on
reinforcement learning (RL) algorithms for training such agents. MLGym-bench
consists of 13 diverse and open-ended AI research tasks from diverse domains
such as computer vision, natural language processing, reinforcement learning,
and game theory. Solving these tasks requires real-world AI research skills
such as generating new ideas and hypotheses, creating and processing data,
implementing ML methods, training models, running experiments, analyzing the
results, and iterating through this process to improve on a given task. We
evaluate a number of frontier large language models (LLMs) on our benchmarks
such as Claude-3.5-Sonnet, Llama-3.1 405B, GPT-4o, o1-preview, and Gemini-1.5
Pro. Our MLGym framework makes it easy to add new tasks, integrate and evaluate
models or agents, generate synthetic data at scale, as well as develop new
learning algorithms for training agents on AI research tasks. We find that
current frontier models can improve on the given baselines, usually by finding
better hyperparameters, but do not generate novel hypotheses, algorithms,
architectures, or substantial improvements. We open-source our framework and
benchmark to facilitate future research in advancing the AI research
capabilities of LLM agents. | 171 | 67b7ee1ffedfe971271dcd3a | null | null |
|
2025-02-20T22:04:42.635000 | S*: Test Time Scaling for Code Generation | 3 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | true | null | 2502.14382 | [
{
"_id": "67b7ed3e58f6b70b18ddb4bc",
"hidden": false,
"name": "Dacheng Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:45:13.558Z",
"user": {
"_id": "63715b25ffc0489ed7d1f415",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63715b25ffc0489ed7d1f415/xZJepbs0LRqFbW1knnBKR.jpeg",
"fullname": "Dacheng Li",
"isPro": false,
"type": "user",
"user": "DachengLi"
}
},
{
"_id": "67b7ed3e58f6b70b18ddb4bd",
"hidden": false,
"name": "Shiyi Cao",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:43.358Z",
"user": {
"_id": "64ebbae6895a36ab28de811a",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64ebbae6895a36ab28de811a/gBiaQP4paS4L13eu-yRm7.jpeg",
"fullname": "Shiyi Cao",
"isPro": false,
"type": "user",
"user": "eva98"
}
},
{
"_id": "67b7ed3e58f6b70b18ddb4be",
"hidden": false,
"name": "Chengkun Cao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7ed3e58f6b70b18ddb4bf",
"hidden": false,
"name": "Xiuyu Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:45:28.150Z",
"user": {
"_id": "644570ba2d91b15b4c7f6311",
"avatarUrl": "/avatars/d5e66012066d0c330b8f23718b1499d8.svg",
"fullname": "Xiuyu Li",
"isPro": false,
"type": "user",
"user": "xiuyul"
}
},
{
"_id": "67b7ed3e58f6b70b18ddb4c0",
"hidden": false,
"name": "Shangyin Tan",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:45:33.730Z",
"user": {
"_id": "663cfbd9b0f659de3db65c1a",
"avatarUrl": "/avatars/b7d82d281026ee04a9932b44a770b840.svg",
"fullname": "Shangyin Tan",
"isPro": false,
"type": "user",
"user": "shangyint"
}
},
{
"_id": "67b7ed3e58f6b70b18ddb4c1",
"hidden": false,
"name": "Kurt Keutzer",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:45:39.902Z",
"user": {
"_id": "6251bf4b183aa4266924ad91",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1678041834400-6251bf4b183aa4266924ad91.jpeg",
"fullname": "Kurt Keutzer",
"isPro": true,
"type": "user",
"user": "kurtkeutzer"
}
},
{
"_id": "67b7ed3e58f6b70b18ddb4c2",
"hidden": false,
"name": "Jiarong Xing",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:45:45.774Z",
"user": {
"_id": "66ff8aa43a31c499dc48fdd6",
"avatarUrl": "/avatars/060dc90fb13991bd013ce8173f12ae3e.svg",
"fullname": "Jiarong Xing",
"isPro": false,
"type": "user",
"user": "JerryPotter"
}
},
{
"_id": "67b7ed3e58f6b70b18ddb4c3",
"hidden": false,
"name": "Joseph E. Gonzalez",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-21T14:45:52.578Z",
"user": {
"_id": "645d2e8401f4eaab2a0878ce",
"avatarUrl": "/avatars/1273c5fb607b4b622a746a42692fa632.svg",
"fullname": "Joseph E. Gonzalez",
"isPro": false,
"type": "user",
"user": "ProfJoeyG"
}
},
{
"_id": "67b7ed3e58f6b70b18ddb4c4",
"hidden": false,
"name": "Ion Stoica",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-20T09:18:53 | S*: Test Time Scaling for Code Generation | Increasing test-time compute for LLMs shows promise across domains but
remains underexplored in code generation, despite extensive study in math. In
this paper, we propose S*, the first hybrid test-time scaling framework that
substantially improves the coverage and selection accuracy of generated code.
S* extends the existing parallel scaling paradigm with sequential scaling to
push performance boundaries. It further leverages a novel selection mechanism
that adaptively generates distinguishing inputs for pairwise comparison,
combined with execution-grounded information to robustly identify correct
solutions. We evaluate across 12 Large Language Models and Large Reasoning
Model and show: (1) S* consistently improves performance across model families
and sizes, enabling a 3B model to outperform GPT-4o-mini; (2) S* enables
non-reasoning models to surpass reasoning models - GPT-4o-mini with S*
outperforms o1-preview by 3.7% on LiveCodeBench; (3) S* further boosts
state-of-the-art reasoning models - DeepSeek-R1-Distill-Qwen-32B with S*
achieves 85.7% on LiveCodeBench, approaching o1 (high) at 88.5%. Code will be
available under https://github.com/NovaSky-AI/SkyThought. | 59 | 67b7ed3f58f6b70b18ddb510 | null | null |
|
2025-02-20T21:25:09.725000 | On the Trustworthiness of Generative Foundation Models: Guideline, Assessment, and Perspective | 2 | {
"_id": "639d94ab7145123e0d44e48a",
"avatarUrl": "/avatars/5bb6a65b306d1383c4a8bcd9334b470a.svg",
"followerCount": 2,
"fullname": "Yue Huang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "HowieHwong",
"type": "user"
} | true | null | 2502.14296 | [
{
"_id": "67b7e371f17ca6989faa9884",
"hidden": false,
"name": "Yue Huang",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-21T02:22:45.907Z",
"user": {
"_id": "639d94ab7145123e0d44e48a",
"avatarUrl": "/avatars/5bb6a65b306d1383c4a8bcd9334b470a.svg",
"fullname": "Yue Huang",
"isPro": false,
"type": "user",
"user": "HowieHwong"
}
},
{
"_id": "67b7e371f17ca6989faa9885",
"hidden": false,
"name": "Chujie Gao",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:51.964Z",
"user": {
"_id": "65a13cb1c5770b27aef2a2bc",
"avatarUrl": "/avatars/88ec5b988f10ad9fd4d469ae2fa34680.svg",
"fullname": "Chujie Gao",
"isPro": false,
"type": "user",
"user": "Flossie"
}
},
{
"_id": "67b7e371f17ca6989faa9886",
"hidden": false,
"name": "Siyuan Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9887",
"hidden": false,
"name": "Haoran Wang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:23:23.186Z",
"user": {
"_id": "64b82c659ebb69a79f0073f6",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64b82c659ebb69a79f0073f6/INEHG7kijEHOhFZMjdBIM.png",
"fullname": "Haoran Wang",
"isPro": false,
"type": "user",
"user": "wang2226"
}
},
{
"_id": "67b7e371f17ca6989faa9888",
"hidden": false,
"name": "Xiangqi Wang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:45.842Z",
"user": {
"_id": "66e4aa8d4926518abbf5cae2",
"avatarUrl": "/avatars/dcff2521e0292b602f86c76fc4b5bbae.svg",
"fullname": "XiangqiWang",
"isPro": false,
"type": "user",
"user": "qisein"
}
},
{
"_id": "67b7e371f17ca6989faa9889",
"hidden": false,
"name": "Yujun Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa988a",
"hidden": false,
"name": "Yanbo Wang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:54.838Z",
"user": {
"_id": "6512a4322f0aa026dd6dc9f8",
"avatarUrl": "/avatars/7cc88d2d8061a83a24bb4458d7cbb242.svg",
"fullname": "wyf",
"isPro": false,
"type": "user",
"user": "wyf23187"
}
},
{
"_id": "67b7e371f17ca6989faa988b",
"hidden": false,
"name": "Jiayi Ye",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa988c",
"hidden": false,
"name": "Jiawen Shi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa988d",
"hidden": false,
"name": "Qihui Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa988e",
"hidden": false,
"name": "Yuan Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa988f",
"hidden": false,
"name": "Han Bao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9890",
"hidden": false,
"name": "Zhaoyi Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9891",
"hidden": false,
"name": "Tianrui Guan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9892",
"hidden": false,
"name": "Dongping Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9893",
"hidden": false,
"name": "Ruoxi Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9894",
"hidden": false,
"name": "Kehan Guo",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9895",
"hidden": false,
"name": "Andy Zou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9896",
"hidden": false,
"name": "Bryan Hooi Kuen-Yew",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9897",
"hidden": false,
"name": "Caiming Xiong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9898",
"hidden": false,
"name": "Elias Stengel-Eskin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa9899",
"hidden": false,
"name": "Hongyang Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa989a",
"hidden": false,
"name": "Hongzhi Yin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa989b",
"hidden": false,
"name": "Huan Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa989c",
"hidden": false,
"name": "Huaxiu Yao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa989d",
"hidden": false,
"name": "Jaehong Yoon",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:50.021Z",
"user": {
"_id": "652066649004117947e46ed6",
"avatarUrl": "/avatars/972c97df6f26d2c3d6ce71ec579984bb.svg",
"fullname": "Jaehong Yoon",
"isPro": false,
"type": "user",
"user": "jaehong31"
}
},
{
"_id": "67b7e371f17ca6989faa989e",
"hidden": false,
"name": "Jieyu Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa989f",
"hidden": false,
"name": "Kai Shu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a0",
"hidden": false,
"name": "Kaijie Zhu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a1",
"hidden": false,
"name": "Ranjay Krishna",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a2",
"hidden": false,
"name": "Swabha Swayamdipta",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a3",
"hidden": false,
"name": "Taiwei Shi",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:56.645Z",
"user": {
"_id": "62e1b3cb3eb0730f621a83f6",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1658958764563-noauth.jpeg",
"fullname": "Taiwei Shi",
"isPro": false,
"type": "user",
"user": "MaksimSTW"
}
},
{
"_id": "67b7e371f17ca6989faa98a4",
"hidden": false,
"name": "Weijia Shi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a5",
"hidden": false,
"name": "Xiang Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a6",
"hidden": false,
"name": "Yiwei Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a7",
"hidden": false,
"name": "Yuexing Hao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a8",
"hidden": false,
"name": "Yuexing Hao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98a9",
"hidden": false,
"name": "Zhihao Jia",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98aa",
"hidden": false,
"name": "Zhize Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98ab",
"hidden": false,
"name": "Xiuying Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98ac",
"hidden": false,
"name": "Zhengzhong Tu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98ad",
"hidden": false,
"name": "Xiyang Hu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98ae",
"hidden": false,
"name": "Tianyi Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98af",
"hidden": false,
"name": "Jieyu Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b0",
"hidden": false,
"name": "Lichao Sun",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b1",
"hidden": false,
"name": "Furong Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b2",
"hidden": false,
"name": "Or Cohen Sasson",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:58:47.678Z",
"user": {
"_id": "67b7eccd10a9714460d767fc",
"avatarUrl": "/avatars/93b03f5a9df7bab777295d811520454f.svg",
"fullname": "Or Cohen-Sasson",
"isPro": false,
"type": "user",
"user": "orcs-prime"
}
},
{
"_id": "67b7e371f17ca6989faa98b3",
"hidden": false,
"name": "Prasanna Sattigeri",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b4",
"hidden": false,
"name": "Anka Reuel",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b5",
"hidden": false,
"name": "Max Lamparth",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b6",
"hidden": false,
"name": "Yue Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b7",
"hidden": false,
"name": "Nouha Dziri",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b8",
"hidden": false,
"name": "Yu Su",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98b9",
"hidden": false,
"name": "Huan Sun",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98ba",
"hidden": false,
"name": "Heng Ji",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98bb",
"hidden": false,
"name": "Chaowei Xiao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98bc",
"hidden": false,
"name": "Mohit Bansal",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98bd",
"hidden": false,
"name": "Nitesh V. Chawla",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98be",
"hidden": false,
"name": "Jian Pei",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98bf",
"hidden": false,
"name": "Jianfeng Gao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98c0",
"hidden": false,
"name": "Michael Backes",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98c1",
"hidden": false,
"name": "Philip S. Yu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98c2",
"hidden": false,
"name": "Neil Zhenqiang Gong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98c3",
"hidden": false,
"name": "Pin-Yu Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98c4",
"hidden": false,
"name": "Bo Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e371f17ca6989faa98c5",
"hidden": false,
"name": "Xiangliang Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-20T06:20:36 | On the Trustworthiness of Generative Foundation Models: Guideline,
Assessment, and Perspective | Generative Foundation Models (GenFMs) have emerged as transformative tools.
However, their widespread adoption raises critical concerns regarding
trustworthiness across dimensions. This paper presents a comprehensive
framework to address these challenges through three key contributions. First,
we systematically review global AI governance laws and policies from
governments and regulatory bodies, as well as industry practices and standards.
Based on this analysis, we propose a set of guiding principles for GenFMs,
developed through extensive multidisciplinary collaboration that integrates
technical, ethical, legal, and societal perspectives. Second, we introduce
TrustGen, the first dynamic benchmarking platform designed to evaluate
trustworthiness across multiple dimensions and model types, including
text-to-image, large language, and vision-language models. TrustGen leverages
modular components--metadata curation, test case generation, and contextual
variation--to enable adaptive and iterative assessments, overcoming the
limitations of static evaluation methods. Using TrustGen, we reveal significant
progress in trustworthiness while identifying persistent challenges. Finally,
we provide an in-depth discussion of the challenges and future directions for
trustworthy GenFMs, which reveals the complex, evolving nature of
trustworthiness, highlighting the nuanced trade-offs between utility and
trustworthiness, and consideration for various downstream applications,
identifying persistent challenges and providing a strategic roadmap for future
research. This work establishes a holistic framework for advancing
trustworthiness in GenAI, paving the way for safer and more responsible
integration of GenFMs into critical applications. To facilitate advancement in
the community, we release the toolkit for dynamic evaluation. | 45 | 67b7e375f17ca6989faa9a28 | null | null |
|
2025-02-20T21:13:28.792000 | Which of These Best Describes Multiple Choice Evaluation with LLMs? A) Forced B) Flawed C) Fixable D) All of the Above | 2 | {
"_id": "62a3f93fe2b7740fe2a94c86",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/62a3f93fe2b7740fe2a94c86/ZiaPqiVqXI2ANIyWQY_hT.png",
"followerCount": 6,
"fullname": "Nishant Balepur",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "nbalepur",
"type": "user"
} | true | null | 2502.14127 | [
{
"_id": "67b7e12b92b9b5b8184c6580",
"hidden": false,
"name": "Nishant Balepur",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:02.019Z",
"user": {
"_id": "62a3f93fe2b7740fe2a94c86",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/62a3f93fe2b7740fe2a94c86/ZiaPqiVqXI2ANIyWQY_hT.png",
"fullname": "Nishant Balepur",
"isPro": false,
"type": "user",
"user": "nbalepur"
}
},
{
"_id": "67b7e12b92b9b5b8184c6581",
"hidden": false,
"name": "Rachel Rudinger",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7e12b92b9b5b8184c6582",
"hidden": false,
"name": "Jordan Lee Boyd-Graber",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T22:11:52 | Which of These Best Describes Multiple Choice Evaluation with LLMs? A)
Forced B) Flawed C) Fixable D) All of the Above | Multiple choice question answering (MCQA) is popular for LLM evaluation due
to its simplicity and human-like testing, but we argue for its reform. We first
reveal flaws in MCQA's format, as it struggles to: 1) test
generation/subjectivity; 2) match LLM use cases; and 3) fully test knowledge.
We instead advocate for generative formats based on human testing-where LLMs
construct and explain answers-better capturing user needs and knowledge while
remaining easy to score. We then show even when MCQA is a useful format, its
datasets suffer from: leakage; unanswerability; shortcuts; and saturation. In
each issue, we give fixes from education, like rubrics to guide MCQ writing;
scoring methods to bridle guessing; and Item Response Theory to build harder
MCQs. Lastly, we discuss LLM errors in MCQA-robustness, biases, and unfaithful
explanations-showing how our prior solutions better measure or address these
issues. While we do not need to desert MCQA, we encourage more efforts in
refining the task based on educational testing, advancing evaluations. | 2 | 67b7e12c92b9b5b8184c65a5 | null | null |
|
2025-02-20T16:00:25.426000 | REALTALK: A 21-Day Real-World Dataset for Long-Term Conversation | 2 | {
"_id": "6142ec5a7215c6d505bafd4e",
"avatarUrl": "/avatars/ae0387b672435c5a4cf16ff6764ce597.svg",
"followerCount": null,
"fullname": "Dong-Ho Lee",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "danny911kr",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/6142ec5a7215c6d505bafd4e/8ZXPnL7UdgpvHkiP0HHDI.png"
]
| 2502.13270 | [
{
"_id": "67b7975d10a9714460c03882",
"hidden": false,
"name": "Dong-Ho Lee",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:30.731Z",
"user": {
"_id": "6142ec5a7215c6d505bafd4e",
"avatarUrl": "/avatars/ae0387b672435c5a4cf16ff6764ce597.svg",
"fullname": "Dong-Ho Lee",
"isPro": false,
"type": "user",
"user": "danny911kr"
}
},
{
"_id": "67b7975d10a9714460c03883",
"hidden": false,
"name": "Adyasha Maharana",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7975d10a9714460c03884",
"hidden": false,
"name": "Jay Pujara",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7975d10a9714460c03885",
"hidden": false,
"name": "Xiang Ren",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7975d10a9714460c03886",
"hidden": false,
"name": "Francesco Barbieri",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T20:29:01 | REALTALK: A 21-Day Real-World Dataset for Long-Term Conversation | Long-term, open-domain dialogue capabilities are essential for chatbots
aiming to recall past interactions and demonstrate emotional intelligence (EI).
Yet, most existing research relies on synthetic, LLM-generated data, leaving
open questions about real-world conversational patterns. To address this gap,
we introduce REALTALK, a 21-day corpus of authentic messaging app dialogues,
providing a direct benchmark against genuine human interactions.
We first conduct a dataset analysis, focusing on EI attributes and persona
consistency to understand the unique challenges posed by real-world dialogues.
By comparing with LLM-generated conversations, we highlight key differences,
including diverse emotional expressions and variations in persona stability
that synthetic dialogues often fail to capture.
Building on these insights, we introduce two benchmark tasks: (1) persona
simulation where a model continues a conversation on behalf of a specific user
given prior dialogue context; and (2) memory probing where a model answers
targeted questions requiring long-term memory of past interactions.
Our findings reveal that models struggle to simulate a user solely from
dialogue history, while fine-tuning on specific user chats improves persona
emulation. Additionally, existing models face significant challenges in
recalling and leveraging long-term context within real-world conversations. | 6 | 67b7975e10a9714460c038bb | null | null |
|
2025-02-20T14:34:52.849000 | From Tools to Teammates: Evaluating LLMs in Multi-Session Coding Interactions | 3 | {
"_id": "62645f88c39850dc093d6105",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1650745211725-noauth.png",
"followerCount": 51,
"fullname": "Mohammed Hamdy",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "mmhamdy",
"type": "user"
} | true | null | 2502.13791 | [
{
"_id": "67b7838bb41e5f760f8bd1b0",
"hidden": false,
"name": "Nathanaël Carraz Rakotonirina",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T15:51:38.471Z",
"user": {
"_id": "6195d3199b7166aedc74247f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6195d3199b7166aedc74247f/8N5mUpSf3q5hRfMd8x70U.jpeg",
"fullname": "Nathanaël Carraz Rakotonirina",
"isPro": false,
"type": "user",
"user": "nathanaelc"
}
},
{
"_id": "67b7838bb41e5f760f8bd1b1",
"hidden": false,
"name": "Mohammed Hamdy",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:34.800Z",
"user": {
"_id": "62645f88c39850dc093d6105",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1650745211725-noauth.png",
"fullname": "Mohammed Hamdy",
"isPro": false,
"type": "user",
"user": "mmhamdy"
}
},
{
"_id": "67b7838bb41e5f760f8bd1b2",
"hidden": false,
"name": "Jon Ander Campos",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7838bb41e5f760f8bd1b3",
"hidden": false,
"name": "Lucas Weber",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7838bb41e5f760f8bd1b4",
"hidden": false,
"name": "Alberto Testoni",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7838bb41e5f760f8bd1b5",
"hidden": false,
"name": "Marzieh Fadaee",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7838bb41e5f760f8bd1b6",
"hidden": false,
"name": "Sandro Pezzelle",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b7838bb41e5f760f8bd1b7",
"hidden": false,
"name": "Marco Del Tredici",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T14:58:04 | From Tools to Teammates: Evaluating LLMs in Multi-Session Coding
Interactions | Large Language Models (LLMs) are increasingly used in working environments
for a wide range of tasks, excelling at solving individual problems in
isolation. However, are they also able to effectively collaborate over
long-term interactions? To investigate this, we introduce MemoryCode, a
synthetic multi-session dataset designed to test LLMs' ability to track and
execute simple coding instructions amid irrelevant information, simulating a
realistic setting. While all the models we tested handle isolated instructions
well, even the performance of state-of-the-art models like GPT-4o deteriorates
when instructions are spread across sessions. Our analysis suggests this is due
to their failure to retrieve and integrate information over long instruction
chains. Our results highlight a fundamental limitation of current LLMs,
restricting their ability to collaborate effectively in long interactions. | 5 | 67b7838cb41e5f760f8bd209 | null | null |
|
2025-02-20T13:47:47.134000 | Judging the Judges: A Collection of LLM-Generated Relevance Judgements | 2 | {
"_id": "64108fc514215c0775e13f5e",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64108fc514215c0775e13f5e/pHWr8TlBnYrulo2owIrrv.jpeg",
"followerCount": null,
"fullname": "Hossein A. (Saeed) Rahmani",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "rahmanidashti",
"type": "user"
} | true | null | 2502.13908 | [
{
"_id": "67b75ce1fedef65ff99cf5f8",
"hidden": false,
"name": "Hossein A. Rahmani",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T16:57:36.417Z",
"user": {
"_id": "64108fc514215c0775e13f5e",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64108fc514215c0775e13f5e/pHWr8TlBnYrulo2owIrrv.jpeg",
"fullname": "Hossein A. (Saeed) Rahmani",
"isPro": false,
"type": "user",
"user": "rahmanidashti"
}
},
{
"_id": "67b75ce1fedef65ff99cf5f9",
"hidden": false,
"name": "Clemencia Siro",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b75ce1fedef65ff99cf5fa",
"hidden": false,
"name": "Mohammad Aliannejadi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b75ce1fedef65ff99cf5fb",
"hidden": false,
"name": "Nick Craswell",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b75ce1fedef65ff99cf5fc",
"hidden": false,
"name": "Charles L. A. Clarke",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b75ce1fedef65ff99cf5fd",
"hidden": false,
"name": "Guglielmo Faggioli",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b75ce1fedef65ff99cf5fe",
"hidden": false,
"name": "Bhaskar Mitra",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b75ce1fedef65ff99cf5ff",
"hidden": false,
"name": "Paul Thomas",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b75ce1fedef65ff99cf600",
"hidden": false,
"name": "Emine Yilmaz",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T17:40:32 | Judging the Judges: A Collection of LLM-Generated Relevance Judgements | Using Large Language Models (LLMs) for relevance assessments offers promising
opportunities to improve Information Retrieval (IR), Natural Language
Processing (NLP), and related fields. Indeed, LLMs hold the promise of allowing
IR experimenters to build evaluation collections with a fraction of the manual
human labor currently required. This could help with fresh topics on which
there is still limited knowledge and could mitigate the challenges of
evaluating ranking systems in low-resource scenarios, where it is challenging
to find human annotators. Given the fast-paced recent developments in the
domain, many questions concerning LLMs as assessors are yet to be answered.
Among the aspects that require further investigation, we can list the impact of
various components in a relevance judgment generation pipeline, such as the
prompt used or the LLM chosen.
This paper benchmarks and reports on the results of a large-scale automatic
relevance judgment evaluation, the LLMJudge challenge at SIGIR 2024, where
different relevance assessment approaches were proposed. In detail, we release
and benchmark 42 LLM-generated labels of the TREC 2023 Deep Learning track
relevance judgments produced by eight international teams who participated in
the challenge. Given their diverse nature, these automatically generated
relevance judgments can help the community not only investigate systematic
biases caused by LLMs but also explore the effectiveness of ensemble models,
analyze the trade-offs between different models and human assessors, and
advance methodologies for improving automated evaluation techniques. The
released resource is available at the following link:
https://llm4eval.github.io/LLMJudge-benchmark/ | 4 | 67b75ce2fedef65ff99cf623 | null | null |
|
2025-02-20T12:26:53.898000 | MMTEB: Massive Multilingual Text Embedding Benchmark | 3 | {
"_id": "5f1eb362eec0ad2a071ad6e2",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/5f1eb362eec0ad2a071ad6e2/IXMYkYKuTwn6kBdWnQeeY.png",
"followerCount": 120,
"fullname": "Niklas Muennighoff",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Muennighoff",
"type": "user"
} | true | null | 2502.13595 | [
{
"_id": "67b6fa9cb544aa153178a60b",
"hidden": false,
"name": "Kenneth Enevoldsen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:52:45.751Z",
"user": {
"_id": "5ff5943752c26e9bc240bada",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/5ff5943752c26e9bc240bada/Exyzf3C_gJ2KdsL4K5_cq.png",
"fullname": "Kenneth C. Enevoldsen",
"isPro": false,
"type": "user",
"user": "KennethEnevoldsen"
}
},
{
"_id": "67b6fa9cb544aa153178a60c",
"hidden": false,
"name": "Isaac Chung",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:42.744Z",
"user": {
"_id": "64cc0e80a257a3212c0c4b24",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64cc0e80a257a3212c0c4b24/wqs6WZN8-3OQthcnQXgN7.png",
"fullname": "Isaac Chung",
"isPro": false,
"type": "user",
"user": "isaacchung"
}
},
{
"_id": "67b6fa9cb544aa153178a60d",
"hidden": false,
"name": "Imene Kerboua",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T10:18:18.756Z",
"user": {
"_id": "62610f8040e04009e81047e9",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/62610f8040e04009e81047e9/iqGTYB7OMmS0jkFEB7cEF.jpeg",
"fullname": "Imene Kerboua",
"isPro": false,
"type": "user",
"user": "imenelydiaker"
}
},
{
"_id": "67b6fa9cb544aa153178a60e",
"hidden": false,
"name": "Márton Kardos",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:40.740Z",
"user": {
"_id": "62696cd3d1ac0cde59280dcf",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1656940964610-62696cd3d1ac0cde59280dcf.jpeg",
"fullname": "Márton Kardos",
"isPro": false,
"type": "user",
"user": "kardosdrur"
}
},
{
"_id": "67b6fa9cb544aa153178a60f",
"hidden": false,
"name": "Ashwin Mathur",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a610",
"hidden": false,
"name": "David Stap",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a611",
"hidden": false,
"name": "Jay Gala",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a612",
"hidden": false,
"name": "Wissam Siblini",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a613",
"hidden": false,
"name": "Dominik Krzemiński",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T10:58:20.283Z",
"user": {
"_id": "662cc845616128914a3c9817",
"avatarUrl": "/avatars/cd186de00f28bb866abc1ab6c4465663.svg",
"fullname": "DomKrz",
"isPro": false,
"type": "user",
"user": "dokato"
}
},
{
"_id": "67b6fa9cb544aa153178a614",
"hidden": false,
"name": "Genta Indra Winata",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a615",
"hidden": false,
"name": "Saba Sturua",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a616",
"hidden": false,
"name": "Saiteja Utpala",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a617",
"hidden": false,
"name": "Mathieu Ciancone",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a618",
"hidden": false,
"name": "Marion Schaeffer",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T16:39:04.878Z",
"user": {
"_id": "6527f0678808d80ccff9c230",
"avatarUrl": "/avatars/3f7edd63770d42472947252c36ffbf5e.svg",
"fullname": "Marion Schaeffer",
"isPro": false,
"type": "user",
"user": "mschaeffer"
}
},
{
"_id": "67b6fa9cb544aa153178a619",
"hidden": false,
"name": "Gabriel Sequeira",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a61a",
"hidden": false,
"name": "Diganta Misra",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a61b",
"hidden": false,
"name": "Shreeya Dhakal",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a61c",
"hidden": false,
"name": "Jonathan Rystrøm",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a61d",
"hidden": false,
"name": "Roman Solomatin",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:52:48.202Z",
"user": {
"_id": "61af4544d691b3aadd1f62b6",
"avatarUrl": "/avatars/7a4067accdd1005f78c3c4adad3ee0a5.svg",
"fullname": "Solomatin Roman",
"isPro": false,
"type": "user",
"user": "Samoed"
}
},
{
"_id": "67b6fa9cb544aa153178a61e",
"hidden": false,
"name": "Ömer Çağatan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a61f",
"hidden": false,
"name": "Akash Kundu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a620",
"hidden": false,
"name": "Martin Bernstorff",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a621",
"hidden": false,
"name": "Shitao Xiao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a622",
"hidden": false,
"name": "Akshita Sukhlecha",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a623",
"hidden": false,
"name": "Bhavish Pahwa",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a624",
"hidden": false,
"name": "Rafał Poświata",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-25T09:40:57.852Z",
"user": {
"_id": "63933543f8b4767ae646e8a1",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670591762482-noauth.png",
"fullname": "Rafał Poświata",
"isPro": false,
"type": "user",
"user": "rafalposwiata"
}
},
{
"_id": "67b6fa9cb544aa153178a625",
"hidden": false,
"name": "Kranthi Kiran GV",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a626",
"hidden": false,
"name": "Shawon Ashraf",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a627",
"hidden": false,
"name": "Daniel Auras",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:24:56.683Z",
"user": {
"_id": "642c54a5b09c70b36de03071",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/642c54a5b09c70b36de03071/EwQyQmust01dgkdRMtYFt.jpeg",
"fullname": "rasdani",
"isPro": false,
"type": "user",
"user": "rasdani"
}
},
{
"_id": "67b6fa9cb544aa153178a628",
"hidden": false,
"name": "Björn Plüster",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a629",
"hidden": false,
"name": "Jan Philipp Harries",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a62a",
"hidden": false,
"name": "Loïc Magne",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a62b",
"hidden": false,
"name": "Isabelle Mohr",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a62c",
"hidden": false,
"name": "Mariya Hendriksen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:23:49.198Z",
"user": {
"_id": "63960f35339bf68bc775b468",
"avatarUrl": "/avatars/d82a914df001b830378246904634b756.svg",
"fullname": "Mariya Hendrikse",
"isPro": false,
"type": "user",
"user": "mariya-he"
}
},
{
"_id": "67b6fa9cb544aa153178a62d",
"hidden": false,
"name": "Dawei Zhu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a62e",
"hidden": false,
"name": "Hippolyte Gisserot-Boukhlef",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a62f",
"hidden": false,
"name": "Tom Aarsen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T15:15:34.699Z",
"user": {
"_id": "6317233cc92fd6fee317e030",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6317233cc92fd6fee317e030/cJHSvvimr1kqgQfHOjO5n.png",
"fullname": "Tom Aarsen",
"isPro": false,
"type": "user",
"user": "tomaarsen"
}
},
{
"_id": "67b6fa9cb544aa153178a630",
"hidden": false,
"name": "Jan Kostkan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a631",
"hidden": false,
"name": "Konrad Wojtasik",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a632",
"hidden": false,
"name": "Taemin Lee",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a633",
"hidden": false,
"name": "Marek Šuppa",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a634",
"hidden": false,
"name": "Crystina Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a635",
"hidden": false,
"name": "Roberta Rocca",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a636",
"hidden": false,
"name": "Mohammed Hamdy",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:46.087Z",
"user": {
"_id": "62645f88c39850dc093d6105",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1650745211725-noauth.png",
"fullname": "Mohammed Hamdy",
"isPro": false,
"type": "user",
"user": "mmhamdy"
}
},
{
"_id": "67b6fa9cb544aa153178a637",
"hidden": false,
"name": "Andrianos Michail",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-28T13:07:57.365Z",
"user": {
"_id": "60e1bc418479fac0bd1daa0e",
"avatarUrl": "/avatars/61f4f8ac0714aae7d5cbb7d4e1038020.svg",
"fullname": "Andrianos Michail",
"isPro": false,
"type": "user",
"user": "Andrianos"
}
},
{
"_id": "67b6fa9cb544aa153178a638",
"hidden": false,
"name": "John Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a639",
"hidden": false,
"name": "Manuel Faysse",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a63a",
"hidden": false,
"name": "Aleksei Vatolin",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:25:10.077Z",
"user": {
"_id": "6454e5ac273f64983024ba5d",
"avatarUrl": "/avatars/ca6f861ed830a79f7a1eba04ebe84afc.svg",
"fullname": "Vatolin Alexey",
"isPro": false,
"type": "user",
"user": "vatolinalex"
}
},
{
"_id": "67b6fa9cb544aa153178a63b",
"hidden": false,
"name": "Nandan Thakur",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:24:59.423Z",
"user": {
"_id": "60196690dd31fde3c1062960",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1612277330660-noauth.jpeg",
"fullname": "Nandan Thakur",
"isPro": false,
"type": "user",
"user": "nthakur"
}
},
{
"_id": "67b6fa9cb544aa153178a63c",
"hidden": false,
"name": "Manan Dey",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a63d",
"hidden": false,
"name": "Dipam Vasani",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a63e",
"hidden": false,
"name": "Pranjal Chitale",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a63f",
"hidden": false,
"name": "Simone Tedeschi",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:23:51.243Z",
"user": {
"_id": "61b85aa99ba538c73a7dc78b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/61b85aa99ba538c73a7dc78b/gWxtQAvOYn7cXgE_nAy0p.jpeg",
"fullname": "Simone Tedeschi",
"isPro": false,
"type": "user",
"user": "sted97"
}
},
{
"_id": "67b6fa9cb544aa153178a640",
"hidden": false,
"name": "Nguyen Tai",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a641",
"hidden": false,
"name": "Artem Snegirev",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a642",
"hidden": false,
"name": "Michael Günther",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a643",
"hidden": false,
"name": "Mengzhou Xia",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a644",
"hidden": false,
"name": "Weijia Shi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a645",
"hidden": false,
"name": "Xing Han Lù",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T16:15:18.716Z",
"user": {
"_id": "5fa9ff3ea13e063b8b2b60cb",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1633380224986-5fa9ff3ea13e063b8b2b60cb.jpeg",
"fullname": "Xing Han Lu",
"isPro": false,
"type": "user",
"user": "xhluca"
}
},
{
"_id": "67b6fa9cb544aa153178a646",
"hidden": false,
"name": "Jordan Clive",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a647",
"hidden": false,
"name": "Gayatri Krishnakumar",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a648",
"hidden": false,
"name": "Anna Maksimova",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a649",
"hidden": false,
"name": "Silvan Wehrli",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a64a",
"hidden": false,
"name": "Maria Tikhonova",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a64b",
"hidden": false,
"name": "Henil Panchal",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a64c",
"hidden": false,
"name": "Aleksandr Abramov",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a64d",
"hidden": false,
"name": "Malte Ostendorff",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T16:15:16.649Z",
"user": {
"_id": "5efda656ff69163f6f59e5d2",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/5efda656ff69163f6f59e5d2/ru2nfhaNjB9-Ls_vbMq92.jpeg",
"fullname": "malteos",
"isPro": false,
"type": "user",
"user": "malteos"
}
},
{
"_id": "67b6fa9cb544aa153178a64e",
"hidden": false,
"name": "Zheng Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a64f",
"hidden": false,
"name": "Simon Clematide",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a650",
"hidden": false,
"name": "Lester James Miranda",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a651",
"hidden": false,
"name": "Alena Fenogenova",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a652",
"hidden": false,
"name": "Guangyu Song",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a653",
"hidden": false,
"name": "Ruqiya Bin Safi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a654",
"hidden": false,
"name": "Wen-Ding Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a655",
"hidden": false,
"name": "Alessia Borghini",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a656",
"hidden": false,
"name": "Federico Cassano",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a657",
"hidden": false,
"name": "Hongjin Su",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a658",
"hidden": false,
"name": "Jimmy Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a659",
"hidden": false,
"name": "Howard Yen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a65a",
"hidden": false,
"name": "Lasse Hansen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a65b",
"hidden": false,
"name": "Sara Hooker",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a65c",
"hidden": false,
"name": "Chenghao Xiao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a65d",
"hidden": false,
"name": "Vaibhav Adlakha",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T16:15:20.849Z",
"user": {
"_id": "61981af5d420757268e195ac",
"avatarUrl": "/avatars/8b59aaf33447224f83d497425fd7ea8f.svg",
"fullname": "Vaibhav Adlakha",
"isPro": false,
"type": "user",
"user": "vaibhavad"
}
},
{
"_id": "67b6fa9cb544aa153178a65e",
"hidden": false,
"name": "Orion Weller",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a65f",
"hidden": false,
"name": "Siva Reddy",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6fa9cb544aa153178a660",
"hidden": false,
"name": "Niklas Muennighoff",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T10:13:43 | MMTEB: Massive Multilingual Text Embedding Benchmark | Text embeddings are typically evaluated on a limited set of tasks, which are
constrained by language, domain, and task diversity. To address these
limitations and provide a more comprehensive evaluation, we introduce the
Massive Multilingual Text Embedding Benchmark (MMTEB) - a large-scale,
community-driven expansion of MTEB, covering over 500 quality-controlled
evaluation tasks across 250+ languages. MMTEB includes a diverse set of
challenging, novel tasks such as instruction following, long-document
retrieval, and code retrieval, representing the largest multilingual collection
of evaluation tasks for embedding models to date. Using this collection, we
develop several highly multilingual benchmarks, which we use to evaluate a
representative set of models. We find that while large language models (LLMs)
with billions of parameters can achieve state-of-the-art performance on certain
language subsets and task categories, the best-performing publicly available
model is multilingual-e5-large-instruct with only 560 million parameters. To
facilitate accessibility and reduce computational cost, we introduce a novel
downsampling method based on inter-task correlation, ensuring a diverse
selection while preserving relative model rankings. Furthermore, we optimize
tasks such as retrieval by sampling hard negatives, creating smaller but
effective splits. These optimizations allow us to introduce benchmarks that
drastically reduce computational demands. For instance, our newly introduced
zero-shot English benchmark maintains a ranking order similar to the full-scale
version but at a fraction of the computational cost. | 31 | 67b6fa9db544aa153178a69c | null | null |
|
2025-02-20T12:23:27.067000 | AIDE: AI-Driven Exploration in the Space of Code | 6 | {
"_id": "65f7927e7bc58032aa5bda58",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/65f7927e7bc58032aa5bda58/JxUSj-J7YBwtgj6rqQtjn.jpeg",
"followerCount": null,
"fullname": "Dex Dixing Xu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "dexhunter",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/65f7927e7bc58032aa5bda58/bkhW4LYUeFqT9_aqPd3Om.jpeg"
]
| 2502.13138 | [
{
"_id": "67b6e0829b29983879ad2312",
"hidden": false,
"name": "Zhengyao Jiang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:38:24.557Z",
"user": {
"_id": "630384837b50dd9d0a3328dc",
"avatarUrl": "/avatars/17097a93ef403592bc07c0ff6712faf3.svg",
"fullname": "Zhengyao Jiang",
"isPro": false,
"type": "user",
"user": "ZhengyaoJiang"
}
},
{
"_id": "67b6e0829b29983879ad2313",
"hidden": false,
"name": "Dominik Schmidt",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:38:52.544Z",
"user": {
"_id": "63042a72eedc089484c89aee",
"avatarUrl": "/avatars/cbbe3000e6fa783c395b23c6b67da5ab.svg",
"fullname": "Dominik Schmidt",
"isPro": false,
"type": "user",
"user": "dominikschmidt"
}
},
{
"_id": "67b6e0829b29983879ad2314",
"hidden": false,
"name": "Dhruv Srikanth",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:38:58.578Z",
"user": {
"_id": "62f3203506b6e6f54b4720f5",
"avatarUrl": "/avatars/64a395d3391e1025dbdd945c33ceee94.svg",
"fullname": "Dhruv Srikanth",
"isPro": false,
"type": "user",
"user": "dSrikanth"
}
},
{
"_id": "67b6e0829b29983879ad2315",
"hidden": false,
"name": "Dixing Xu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:52:52.450Z",
"user": {
"_id": "65f7927e7bc58032aa5bda58",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/65f7927e7bc58032aa5bda58/JxUSj-J7YBwtgj6rqQtjn.jpeg",
"fullname": "Dex Dixing Xu",
"isPro": false,
"type": "user",
"user": "dexhunter"
}
},
{
"_id": "67b6e0829b29983879ad2316",
"hidden": false,
"name": "Ian Kaplan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6e0829b29983879ad2317",
"hidden": false,
"name": "Deniss Jacenko",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6e0829b29983879ad2318",
"hidden": false,
"name": "Yuxiang Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T18:57:21 | AIDE: AI-Driven Exploration in the Space of Code | Machine learning, the foundation of modern artificial intelligence, has
driven innovations that have fundamentally transformed the world. Yet, behind
advancements lies a complex and often tedious process requiring labor and
compute intensive iteration and experimentation. Engineers and scientists
developing machine learning models spend much of their time on trial-and-error
tasks instead of conceptualizing innovative solutions or research hypotheses.
To address this challenge, we introduce AI-Driven Exploration (AIDE), a machine
learning engineering agent powered by large language models (LLMs). AIDE frames
machine learning engineering as a code optimization problem, and formulates
trial-and-error as a tree search in the space of potential solutions. By
strategically reusing and refining promising solutions, AIDE effectively trades
computational resources for enhanced performance, achieving state-of-the-art
results on multiple machine learning engineering benchmarks, including our
Kaggle evaluations, OpenAI MLE-Bench and METRs RE-Bench. | 7 | 67b6e0839b29983879ad2346 | null | null |
|
2025-02-20T12:09:53.761000 | MVL-SIB: A Massively Multilingual Vision-Language Benchmark for Cross-Modal Topical Matching | 2 | {
"_id": "64c8c2d87d0ea4e7f12995c6",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c8c2d87d0ea4e7f12995c6/h8eWJrz8kqavemy8vQ2NK.jpeg",
"followerCount": 3,
"fullname": "Fabian David Schmidt",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "fdschmidt93",
"type": "user"
} | true | null | 2502.12852 | [
{
"_id": "67b5b31f5a17526b55c3ccde",
"hidden": false,
"name": "Fabian David Schmidt",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:53:15.852Z",
"user": {
"_id": "64c8c2d87d0ea4e7f12995c6",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c8c2d87d0ea4e7f12995c6/h8eWJrz8kqavemy8vQ2NK.jpeg",
"fullname": "Fabian David Schmidt",
"isPro": false,
"type": "user",
"user": "fdschmidt93"
}
},
{
"_id": "67b5b31f5a17526b55c3ccdf",
"hidden": false,
"name": "Florian Schneider",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T10:49:49.711Z",
"user": {
"_id": "62dfd54798815401141c47fe",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dfd54798815401141c47fe/ct2OA_K0Wwpshy8DCswxy.png",
"fullname": "Flo Schneider",
"isPro": false,
"type": "user",
"user": "floschne"
}
},
{
"_id": "67b5b31f5a17526b55c3cce0",
"hidden": false,
"name": "Chris Biemann",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5b31f5a17526b55c3cce1",
"hidden": false,
"name": "Goran Glavaš",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T13:40:05 | MVL-SIB: A Massively Multilingual Vision-Language Benchmark for
Cross-Modal Topical Matching | Existing multilingual vision-language (VL) benchmarks often only cover a
handful of languages. Consequently, evaluations of large vision-language models
(LVLMs) predominantly target high-resource languages, underscoring the need for
evaluation data for low-resource languages. To address this limitation, we
introduce MVL-SIB, a massively multilingual vision-language benchmark that
evaluates both cross-modal and text-only topical matching across 205 languages
-- over 100 more than the most multilingual existing VL benchmarks encompass.
We then benchmark a range of of open-weight LVLMs together with GPT-4o(-mini)
on MVL-SIB. Our results reveal that LVLMs struggle in cross-modal topic
matching in lower-resource languages, performing no better than chance on
languages like N'Koo. Our analysis further reveals that VL support in LVLMs
declines disproportionately relative to textual support for lower-resource
languages, as evidenced by comparison of cross-modal and text-only topical
matching performance. We further observe that open-weight LVLMs do not benefit
from representing a topic with more than one image, suggesting that these
models are not yet fully effective at handling multi-image tasks. By
correlating performance on MVL-SIB with other multilingual VL benchmarks, we
highlight that MVL-SIB serves as a comprehensive probe of multilingual VL
understanding in LVLMs. | 3 | 67b5b3205a17526b55c3cd40 | null | null |
|
2025-02-20T12:07:02.880000 | Reducing Hallucinations in Language Model-based SPARQL Query Generation Using Post-Generation Memory Retrieval | 2 | {
"_id": "63e972f1ccae1fe5c6211759",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63e972f1ccae1fe5c6211759/AfKPgMdAraUtvbtJpoHFY.jpeg",
"followerCount": 2,
"fullname": "Luis Lara",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "ludolara",
"type": "user"
} | true | null | 2502.13369 | [
{
"_id": "67b7610afedfe97127f75374",
"hidden": false,
"name": "Aditya Sharma",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:37:33.974Z",
"user": {
"_id": "66d959e4fb6d15635f2b9d76",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/RpDOzFBoW_OQNW4iHuL0g.jpeg",
"fullname": "Aditya Sharma ",
"isPro": false,
"type": "user",
"user": "adityasharma001"
}
},
{
"_id": "67b7610afedfe97127f75375",
"hidden": false,
"name": "Luis Lara",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T17:11:22.814Z",
"user": {
"_id": "63e972f1ccae1fe5c6211759",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63e972f1ccae1fe5c6211759/AfKPgMdAraUtvbtJpoHFY.jpeg",
"fullname": "Luis Lara",
"isPro": false,
"type": "user",
"user": "ludolara"
}
},
{
"_id": "67b7610afedfe97127f75376",
"hidden": false,
"name": "Amal Zouaq",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:37:41.765Z",
"user": {
"_id": "64df79eff08b064990fd62db",
"avatarUrl": "/avatars/88a494615a2339bc29db4ea33a9817b2.svg",
"fullname": "Amal Zouaq",
"isPro": false,
"type": "user",
"user": "zouaq"
}
},
{
"_id": "67b7610afedfe97127f75377",
"hidden": false,
"name": "Christopher J. Pal",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T02:08:13 | Reducing Hallucinations in Language Model-based SPARQL Query Generation
Using Post-Generation Memory Retrieval | The ability to generate SPARQL queries from natural language questions is
crucial for ensuring efficient and accurate retrieval of structured data from
knowledge graphs (KG). While large language models (LLMs) have been widely
adopted for SPARQL query generation, they are often susceptible to
hallucinations and out-of-distribution errors when producing KG elements like
Uniform Resource Identifiers (URIs) based on internal parametric knowledge.
This often results in content that appears plausible but is factually
incorrect, posing significant challenges for their use in real-world
information retrieval (IR) applications. This has led to increased research
aimed at detecting and mitigating such errors. In this paper, we introduce PGMR
(Post-Generation Memory Retrieval), a modular framework that incorporates a
non-parametric memory module to retrieve KG elements and enhance LLM-based
SPARQL query generation. Our experimental results indicate that PGMR
consistently delivers strong performance across diverse datasets, data
distributions, and LLMs. Notably, PGMR significantly mitigates URI
hallucinations, nearly eliminating the problem in several scenarios. | 2 | 67b7610bfedfe97127f7539c | null | null |
|
2025-02-20T10:53:49.049000 | High-Fidelity Novel View Synthesis via Splatting-Guided Diffusion | 2 | {
"_id": "657dc1576dc01435cd9029d8",
"avatarUrl": "/avatars/3bba11ac7659fce61aeaedf40e2057a8.svg",
"followerCount": 2,
"fullname": "Xiang Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "XiangZ",
"type": "user"
} | true | null | 2502.12752 | [
{
"_id": "67b74fbdbb87b88059a9c5d3",
"hidden": false,
"name": "Xiang Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T16:06:01.193Z",
"user": {
"_id": "657dc1576dc01435cd9029d8",
"avatarUrl": "/avatars/3bba11ac7659fce61aeaedf40e2057a8.svg",
"fullname": "Xiang Zhang",
"isPro": false,
"type": "user",
"user": "XiangZ"
}
},
{
"_id": "67b74fbdbb87b88059a9c5d4",
"hidden": false,
"name": "Yang Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b74fbdbb87b88059a9c5d5",
"hidden": false,
"name": "Lukas Mehl",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b74fbdbb87b88059a9c5d6",
"hidden": false,
"name": "Markus Gross",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b74fbdbb87b88059a9c5d7",
"hidden": false,
"name": "Christopher Schroers",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T11:13:06 | High-Fidelity Novel View Synthesis via Splatting-Guided Diffusion | Despite recent advances in Novel View Synthesis (NVS), generating
high-fidelity views from single or sparse observations remains a significant
challenge. Existing splatting-based approaches often produce distorted geometry
due to splatting errors. While diffusion-based methods leverage rich 3D priors
to achieve improved geometry, they often suffer from texture hallucination. In
this paper, we introduce SplatDiff, a pixel-splatting-guided video diffusion
model designed to synthesize high-fidelity novel views from a single image.
Specifically, we propose an aligned synthesis strategy for precise control of
target viewpoints and geometry-consistent view synthesis. To mitigate texture
hallucination, we design a texture bridge module that enables high-fidelity
texture generation through adaptive feature fusion. In this manner, SplatDiff
leverages the strengths of splatting and diffusion to generate novel views with
consistent geometry and high-fidelity details. Extensive experiments verify the
state-of-the-art performance of SplatDiff in single-view NVS. Additionally,
without extra training, SplatDiff shows remarkable zero-shot performance across
diverse tasks, including sparse-view NVS and stereo video conversion. | 3 | 67b74fc7bb87b88059a9c75d | null | null |
|
2025-02-20T10:46:55.281000 | TESS 2: A Large-Scale Generalist Diffusion Language Model | 3 | {
"_id": "62608fc2ffe8827cb1d89f9f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1654027835241-62608fc2ffe8827cb1d89f9f.png",
"followerCount": 11,
"fullname": "Hamish Ivison",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "hamishivi",
"type": "user"
} | true | null | 2502.13917 | [
{
"_id": "67b698422c8b2ef925e03f4f",
"hidden": false,
"name": "Jaesung Tae",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b698422c8b2ef925e03f50",
"hidden": false,
"name": "Hamish Ivison",
"status": "extracted_confirmed",
"statusLastChangedAt": "2025-03-04T04:34:53.882Z",
"user": {
"_id": "62608fc2ffe8827cb1d89f9f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1654027835241-62608fc2ffe8827cb1d89f9f.png",
"fullname": "Hamish Ivison",
"isPro": false,
"type": "user",
"user": "hamishivi"
}
},
{
"_id": "67b698422c8b2ef925e03f51",
"hidden": false,
"name": "Sachin Kumar",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:12:28.834Z",
"user": {
"_id": "63f24d2d7ddf724fbcc0ea9c",
"avatarUrl": "/avatars/3e24c1aa9c1b4066d2dd56aeb4b0f62e.svg",
"fullname": "sachin kumar",
"isPro": false,
"type": "user",
"user": "sachinkumar"
}
},
{
"_id": "67b698422c8b2ef925e03f52",
"hidden": false,
"name": "Arman Cohan",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:12:35.587Z",
"user": {
"_id": "5f5ba21188f57f65f951f255",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1599840760465-noauth.png",
"fullname": "Arman Cohan",
"isPro": false,
"type": "user",
"user": "armanc"
}
}
]
| 2025-02-19T17:50:31 | TESS 2: A Large-Scale Generalist Diffusion Language Model | We introduce TESS 2, a general instruction-following diffusion language model
that outperforms contemporary instruction-tuned diffusion models, as well as
matches and sometimes exceeds strong autoregressive (AR) models. We train TESS
2 by first adapting a strong AR model via continued pretraining with the usual
cross-entropy as diffusion loss, and then performing further instruction
tuning. We find that adaptation training as well as the choice of the base
model is crucial for training good instruction-following diffusion models. We
further propose reward guidance, a novel and modular inference-time guidance
procedure to align model outputs without needing to train the underlying model.
Finally, we show that TESS 2 further improves with increased inference-time
compute, highlighting the utility of diffusion LMs in having fine-grained
controllability over the amount of compute used at inference time. Code and
models are available at https://github.com/hamishivi/tess-2. | 6 | 67b698432c8b2ef925e03fb4 | null | null |
|
2025-02-20T07:25:12.795000 | REFIND: Retrieval-Augmented Factuality Hallucination Detection in Large Language Models | 2 | {
"_id": "6540fbf9cb7fffd683942b43",
"avatarUrl": "/avatars/d4a64fbde511d0949e1c339179586850.svg",
"followerCount": 2,
"fullname": "DongGeon Lee",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "oneonlee",
"type": "user"
} | true | null | 2502.13622 | [
{
"_id": "67b69cf4573aa8417aec103c",
"hidden": false,
"name": "DongGeon Lee",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:35:55.480Z",
"user": {
"_id": "6540fbf9cb7fffd683942b43",
"avatarUrl": "/avatars/d4a64fbde511d0949e1c339179586850.svg",
"fullname": "DongGeon Lee",
"isPro": false,
"type": "user",
"user": "oneonlee"
}
},
{
"_id": "67b69cf4573aa8417aec103d",
"hidden": false,
"name": "Hwanjo Yu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T10:59:05 | REFIND: Retrieval-Augmented Factuality Hallucination Detection in Large
Language Models | Hallucinations in large language model (LLM) outputs severely limit their
reliability in knowledge-intensive tasks such as question answering. To address
this challenge, we introduce REFIND (Retrieval-augmented Factuality
hallucINation Detection), a novel framework that detects hallucinated spans
within LLM outputs by directly leveraging retrieved documents. As part of the
REFIND, we propose the Context Sensitivity Ratio (CSR), a novel metric that
quantifies the sensitivity of LLM outputs to retrieved evidence. This
innovative approach enables REFIND to efficiently and accurately detect
hallucinations, setting it apart from existing methods. In the evaluation,
REFIND demonstrated robustness across nine languages, including low-resource
settings, and significantly outperformed baseline models, achieving superior
IoU scores in identifying hallucinated spans. This work highlights the
effectiveness of quantifying context sensitivity for hallucination detection,
thereby paving the way for more reliable and trustworthy LLM applications
across diverse languages. | 4 | 67b69cf7573aa8417aec10bf | null | null |
|
2025-02-20T06:45:40.507000 | Train Small, Infer Large: Memory-Efficient LoRA Training for Large Language Models | 2 | {
"_id": "63fcb42c987f631186e554f2",
"avatarUrl": "/avatars/5cf87e9fa21c088c0bd8577d651d01f6.svg",
"followerCount": null,
"fullname": "Jun Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "junzhang98",
"type": "user"
} | true | null | 2502.13533 | [
{
"_id": "67b68f883cd5860d8597eace",
"hidden": false,
"name": "Jun Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:36:13.757Z",
"user": {
"_id": "63fcb42c987f631186e554f2",
"avatarUrl": "/avatars/5cf87e9fa21c088c0bd8577d651d01f6.svg",
"fullname": "Jun Zhang",
"isPro": false,
"type": "user",
"user": "junzhang98"
}
},
{
"_id": "67b68f883cd5860d8597eacf",
"hidden": false,
"name": "Jue Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68f883cd5860d8597ead0",
"hidden": false,
"name": "Huan Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68f883cd5860d8597ead1",
"hidden": false,
"name": "Lidan Shou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68f883cd5860d8597ead2",
"hidden": false,
"name": "Ke Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68f883cd5860d8597ead3",
"hidden": false,
"name": "Yang You",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68f883cd5860d8597ead4",
"hidden": false,
"name": "Guiming Xie",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68f883cd5860d8597ead5",
"hidden": false,
"name": "Xuejian Gong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68f883cd5860d8597ead6",
"hidden": false,
"name": "Kunlong Zhou",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:43:26.626Z",
"user": {
"_id": "63db54249f2687298a14dc4d",
"avatarUrl": "/avatars/7fd31c4056be5c304cd79e17e3a6e560.svg",
"fullname": "kunlong zhou",
"isPro": false,
"type": "user",
"user": "kerlomz"
}
}
]
| 2025-02-19T08:39:15 | Train Small, Infer Large: Memory-Efficient LoRA Training for Large
Language Models | Large Language Models (LLMs) have significantly advanced natural language
processing with exceptional task generalization capabilities. Low-Rank Adaption
(LoRA) offers a cost-effective fine-tuning solution, freezing the original
model parameters and training only lightweight, low-rank adapter matrices.
However, the memory footprint of LoRA is largely dominated by the original
model parameters. To mitigate this, we propose LoRAM, a memory-efficient LoRA
training scheme founded on the intuition that many neurons in
over-parameterized LLMs have low training utility but are essential for
inference. LoRAM presents a unique twist: it trains on a pruned (small) model
to obtain pruned low-rank matrices, which are then recovered and utilized with
the original (large) model for inference. Additionally, minimal-cost continual
pre-training, performed by the model publishers in advance, aligns the
knowledge discrepancy between pruned and original models. Our extensive
experiments demonstrate the efficacy of LoRAM across various pruning strategies
and downstream tasks. For a model with 70 billion parameters, LoRAM enables
training on a GPU with only 20G HBM, replacing an A100-80G GPU for LoRA
training and 15 GPUs for full fine-tuning. Specifically, QLoRAM implemented by
structured pruning combined with 4-bit quantization, for LLaMA-3.1-70B
(LLaMA-2-70B), reduces the parameter storage cost that dominates the memory
usage in low-rank matrix training by 15.81times (16.95times), while
achieving dominant performance gains over both the original LLaMA-3.1-70B
(LLaMA-2-70B) and LoRA-trained LLaMA-3.1-8B (LLaMA-2-13B). | 9 | 67b68f8b3cd5860d8597eb97 | null | null |
|
2025-02-20T05:38:39.430000 | Noise May Contain Transferable Knowledge: Understanding Semi-supervised Heterogeneous Domain Adaptation from an Empirical Perspective | 2 | {
"_id": "668bb3b14c25c09b01815a55",
"avatarUrl": "/avatars/5d46301dd5d7641e3da05b0ad560efee.svg",
"followerCount": null,
"fullname": "Yuan Yao",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "yyyaoyuan",
"type": "user"
} | true | null | 2502.13573 | [
{
"_id": "67b70459ea22340afaaf416f",
"hidden": false,
"name": "Yuan Yao",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-20T10:30:51.477Z",
"user": {
"_id": "668bb3b14c25c09b01815a55",
"avatarUrl": "/avatars/5d46301dd5d7641e3da05b0ad560efee.svg",
"fullname": "Yuan Yao",
"isPro": false,
"type": "user",
"user": "yyyaoyuan"
}
},
{
"_id": "67b70459ea22340afaaf4170",
"hidden": false,
"name": "Xiaopu Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:11:57.564Z",
"user": {
"_id": "67a2d9acd62b7924b4511564",
"avatarUrl": "/avatars/adc0c3d0e11acb64ba72c7914f5105db.svg",
"fullname": "Xiaopu Zhang",
"isPro": true,
"type": "user",
"user": "xiaopz2"
}
},
{
"_id": "67b70459ea22340afaaf4171",
"hidden": false,
"name": "Yu Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b70459ea22340afaaf4172",
"hidden": false,
"name": "Jian Jin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b70459ea22340afaaf4173",
"hidden": false,
"name": "Qiang Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T09:27:03 | Noise May Contain Transferable Knowledge: Understanding Semi-supervised
Heterogeneous Domain Adaptation from an Empirical Perspective | Semi-supervised heterogeneous domain adaptation (SHDA) addresses learning
across domains with distinct feature representations and distributions, where
source samples are labeled while most target samples are unlabeled, with only a
small fraction labeled. Moreover, there is no one-to-one correspondence between
source and target samples. Although various SHDA methods have been developed to
tackle this problem, the nature of the knowledge transferred across
heterogeneous domains remains unclear. This paper delves into this question
from an empirical perspective. We conduct extensive experiments on about 330
SHDA tasks, employing two supervised learning methods and seven representative
SHDA methods. Surprisingly, our observations indicate that both the category
and feature information of source samples do not significantly impact the
performance of the target domain. Additionally, noise drawn from simple
distributions, when used as source samples, may contain transferable knowledge.
Based on this insight, we perform a series of experiments to uncover the
underlying principles of transferable knowledge in SHDA. Specifically, we
design a unified Knowledge Transfer Framework (KTF) for SHDA. Based on the KTF,
we find that the transferable knowledge in SHDA primarily stems from the
transferability and discriminability of the source domain. Consequently,
ensuring those properties in source samples, regardless of their origin (e.g.,
image, text, noise), can enhance the effectiveness of knowledge transfer in
SHDA tasks. The codes and datasets are available at
https://github.com/yyyaoyuan/SHDA. | 2 | 67b7045bea22340afaaf41fd | null | null |
|
2025-02-20T05:19:11.890000 | GIMMICK -- Globally Inclusive Multimodal Multitask Cultural Knowledge Benchmarking | 2 | {
"_id": "62dfd54798815401141c47fe",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dfd54798815401141c47fe/ct2OA_K0Wwpshy8DCswxy.png",
"followerCount": 6,
"fullname": "Flo Schneider",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "floschne",
"type": "user"
} | true | null | 2502.13766 | [
{
"_id": "67b6faf5a96bf2b8ff8c235c",
"hidden": false,
"name": "Florian Schneider",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T10:49:37.443Z",
"user": {
"_id": "62dfd54798815401141c47fe",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dfd54798815401141c47fe/ct2OA_K0Wwpshy8DCswxy.png",
"fullname": "Flo Schneider",
"isPro": false,
"type": "user",
"user": "floschne"
}
},
{
"_id": "67b6faf5a96bf2b8ff8c235d",
"hidden": false,
"name": "Carolin Holtermann",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6faf5a96bf2b8ff8c235e",
"hidden": false,
"name": "Chris Biemann",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6faf5a96bf2b8ff8c235f",
"hidden": false,
"name": "Anne Lauscher",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:41:05.177Z",
"user": {
"_id": "626c02e7703f3b27dd590896",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1654503075060-626c02e7703f3b27dd590896.jpeg",
"fullname": "Anne Lauscher",
"isPro": false,
"type": "user",
"user": "anlausch"
}
}
]
| 2025-02-19T14:27:40 | GIMMICK -- Globally Inclusive Multimodal Multitask Cultural Knowledge
Benchmarking | Large Vision-Language Models (LVLMs) have recently gained attention due to
their distinctive performance and broad applicability. While it has been
previously shown that their efficacy in usage scenarios involving non-Western
contexts falls short, existing studies are limited in scope, covering just a
narrow range of cultures, focusing exclusively on a small number of cultural
aspects, or evaluating a limited selection of models on a single task only.
Towards globally inclusive LVLM research, we introduce GIMMICK, an extensive
multimodal benchmark designed to assess a broad spectrum of cultural knowledge
across 144 countries representing six global macro-regions. GIMMICK comprises
six tasks built upon three new datasets that span 728 unique cultural events or
facets on which we evaluated 20 LVLMs and 11 LLMs, including five proprietary
and 26 open-weight models of all sizes. We systematically examine (1) regional
cultural biases, (2) the influence of model size, (3) input modalities, and (4)
external cues. Our analyses reveal strong biases toward Western cultures across
models and tasks and highlight strong correlations between model size and
performance, as well as the effectiveness of multimodal input and external
geographic cues. We further find that models have more knowledge of tangible
than intangible aspects (e.g., food vs. rituals) and that they excel in
recognizing broad cultural origins but struggle with a more nuanced
understanding. | 3 | 67b6faf8a96bf2b8ff8c2422 | null | null |
|
2025-02-20T04:32:22.011000 | InfiR : Crafting Effective Small Language Models and Multimodal Small Language Models in Reasoning | 2 | {
"_id": "618c1ad1c74578e0a4a4d074",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/618c1ad1c74578e0a4a4d074/8u_AkeHt4d6xtQ8hzaffU.jpeg",
"followerCount": 60,
"fullname": "Drishti Sharma",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "DrishtiSharma",
"type": "user"
} | false | null | 2502.11573 | [
{
"_id": "67b6f629d9da6999328e38f5",
"hidden": false,
"name": "Congkai Xie",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:12:49.025Z",
"user": {
"_id": "6719f1ad725123d503b5ef3c",
"avatarUrl": "/avatars/08e1be1f4afa1b6e1501a15cdb786a47.svg",
"fullname": "Congkai Xie",
"isPro": false,
"type": "user",
"user": "congkai"
}
},
{
"_id": "67b6f629d9da6999328e38f6",
"hidden": false,
"name": "Shuo Cai",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e38f7",
"hidden": false,
"name": "Wenjun Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e38f8",
"hidden": true,
"name": "Pengxiang Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:13:23.938Z",
"user": {
"_id": "65d6bd8ebb862d1b367580fe",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/65d6bd8ebb862d1b367580fe/Pf5o-Oxi2IZxJPa-ZgEtV.jpeg",
"fullname": "Pengxiang Li",
"isPro": false,
"type": "user",
"user": "PengxiangLi"
}
},
{
"_id": "67b6f629d9da6999328e38f9",
"hidden": false,
"name": "Zhijie Sang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:13:31.244Z",
"user": {
"_id": "6719f2569665701fd6a6a43d",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/8gS7Tzr6rIZMl-8sS1x_r.png",
"fullname": "Zhijie Sang",
"isPro": false,
"type": "user",
"user": "SANGZHIJIE"
}
},
{
"_id": "67b6f629d9da6999328e38fa",
"hidden": false,
"name": "Kejing Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e38fb",
"hidden": true,
"name": "Yiming Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:13:47.213Z",
"user": {
"_id": "626309292597e1cb7c76ab6f",
"avatarUrl": "/avatars/594b9ccfc5f3aa61613b4bc3158fed4f.svg",
"fullname": "Yiming Zhang",
"isPro": false,
"type": "user",
"user": "yimingzhang"
}
},
{
"_id": "67b6f629d9da6999328e38fc",
"hidden": false,
"name": "Zhen Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e38fd",
"hidden": false,
"name": "Guanghao Zhu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:13:55.619Z",
"user": {
"_id": "670f8a7d4e3a710738fd13cc",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/2ITxPXZCAzdY-LhbNA6NI.png",
"fullname": "Guanghao",
"isPro": false,
"type": "user",
"user": "GuanghaoZhu"
}
},
{
"_id": "67b6f629d9da6999328e38fe",
"hidden": false,
"name": "Zeyu Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e38ff",
"hidden": false,
"name": "Yang Yu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3900",
"hidden": false,
"name": "Yuhang Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3901",
"hidden": false,
"name": "Su Lu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3902",
"hidden": false,
"name": "Baoyi He",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3903",
"hidden": false,
"name": "Qi Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3904",
"hidden": false,
"name": "Xiaotian Han",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:32:33.779Z",
"user": {
"_id": "650dde4ce14eeb01d42b37a1",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/650dde4ce14eeb01d42b37a1/n5Yv24uofZ2XJjXdYCrKd.png",
"fullname": "Xiaotian Han",
"isPro": false,
"type": "user",
"user": "xiaotianhan"
}
},
{
"_id": "67b6f629d9da6999328e3905",
"hidden": false,
"name": "Jianbo Yuan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3906",
"hidden": false,
"name": "Shengyu Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3907",
"hidden": false,
"name": "Fei Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6f629d9da6999328e3908",
"hidden": false,
"name": "Hongxia Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T09:07:32 | InfiR : Crafting Effective Small Language Models and Multimodal Small
Language Models in Reasoning | Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs)
have made significant advancements in reasoning capabilities. However, they
still face challenges such as high computational demands and privacy concerns.
This paper focuses on developing efficient Small Language Models (SLMs) and
Multimodal Small Language Models (MSLMs) that retain competitive reasoning
abilities. We introduce a novel training pipeline that enhances reasoning
capabilities and facilitates deployment on edge devices, achieving
state-of-the-art performance while minimizing development costs. \InfR~ aims to
advance AI systems by improving reasoning, reducing adoption barriers, and
addressing privacy concerns through smaller model sizes. Resources are
available at https://github. com/Reallm-Labs/InfiR. | 8 | 67b6f62ad9da6999328e3955 | null | null |
|
2025-02-20T03:56:54.121000 | ActionPiece: Contextually Tokenizing Action Sequences for Generative Recommendation | 3 | {
"_id": "64a62c2f500beb50968e5c9c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/wfL3ojJmXqyzGUmCblPf4.jpeg",
"followerCount": 5,
"fullname": "Yupeng Hou",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "hyp1231",
"type": "user"
} | true | null | 2502.13581 | [
{
"_id": "67b6ee04412c9eccae5151f5",
"hidden": false,
"name": "Yupeng Hou",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:35:14.498Z",
"user": {
"_id": "64a62c2f500beb50968e5c9c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/wfL3ojJmXqyzGUmCblPf4.jpeg",
"fullname": "Yupeng Hou",
"isPro": false,
"type": "user",
"user": "hyp1231"
}
},
{
"_id": "67b6ee04412c9eccae5151f6",
"hidden": false,
"name": "Jianmo Ni",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6ee04412c9eccae5151f7",
"hidden": false,
"name": "Zhankui He",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:40:37.697Z",
"user": {
"_id": "64daab70c38427829daf5958",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/cjL27oSvuJf1x0Zq3SrEJ.jpeg",
"fullname": "Zhankui He",
"isPro": false,
"type": "user",
"user": "ZhankuiHe"
}
},
{
"_id": "67b6ee04412c9eccae5151f8",
"hidden": false,
"name": "Noveen Sachdeva",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:40:31.037Z",
"user": {
"_id": "652e529acb01191873067b02",
"avatarUrl": "/avatars/a340d5faca6ad1d120e8a904380276e9.svg",
"fullname": "Sachdeva",
"isPro": false,
"type": "user",
"user": "Noveen"
}
},
{
"_id": "67b6ee04412c9eccae5151f9",
"hidden": false,
"name": "Wang-Cheng Kang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6ee04412c9eccae5151fa",
"hidden": false,
"name": "Ed H. Chi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6ee04412c9eccae5151fb",
"hidden": false,
"name": "Julian McAuley",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6ee04412c9eccae5151fc",
"hidden": false,
"name": "Derek Zhiyuan Cheng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:40:05.814Z",
"user": {
"_id": "624cedaa43ce7877c5af2c4e",
"avatarUrl": "/avatars/86f13f8e082e5c0b8ea671c374e4d675.svg",
"fullname": "Zhiyuan Cheng",
"isPro": false,
"type": "user",
"user": "willcee"
}
}
]
| 2025-02-19T09:45:29 | ActionPiece: Contextually Tokenizing Action Sequences for Generative
Recommendation | Generative recommendation (GR) is an emerging paradigm where user actions are
tokenized into discrete token patterns and autoregressively generated as
predictions. However, existing GR models tokenize each action independently,
assigning the same fixed tokens to identical actions across all sequences
without considering contextual relationships. This lack of context-awareness
can lead to suboptimal performance, as the same action may hold different
meanings depending on its surrounding context. To address this issue, we
propose ActionPiece to explicitly incorporate context when tokenizing action
sequences. In ActionPiece, each action is represented as a set of item
features, which serve as the initial tokens. Given the action sequence corpora,
we construct the vocabulary by merging feature patterns as new tokens, based on
their co-occurrence frequency both within individual sets and across adjacent
sets. Considering the unordered nature of feature sets, we further introduce
set permutation regularization, which produces multiple segmentations of action
sequences with the same semantics. Experiments on public datasets demonstrate
that ActionPiece consistently outperforms existing action tokenization methods,
improving NDCG@10 by 6.00% to 12.82%. | 5 | 67b6ee04412c9eccae515223 | null | null |
|
2025-02-20T02:40:09.567000 | MoM: Linear Sequence Modeling with Mixture-of-Memories | 2 | {
"_id": "6246bb33da617c00b48e4d92",
"avatarUrl": "/avatars/0304a9f6eb7f5dee4d933d03222f94e9.svg",
"followerCount": 3,
"fullname": "Weigao Sun",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "weigao266",
"type": "user"
} | true | null | 2502.13685 | [
{
"_id": "67b6dc1ba7567156c6547880",
"hidden": false,
"name": "Jusen Du",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:08:01.601Z",
"user": {
"_id": "65003e857804f04a163328d9",
"avatarUrl": "/avatars/fe32150aabfde8d283b38ccebcf6982e.svg",
"fullname": "Jusen Du",
"isPro": false,
"type": "user",
"user": "JusenK"
}
},
{
"_id": "67b6dc1ba7567156c6547881",
"hidden": false,
"name": "Weigao Sun",
"status": "extracted_confirmed",
"statusLastChangedAt": "2025-03-04T08:10:49.466Z",
"user": {
"_id": "6246bb33da617c00b48e4d92",
"avatarUrl": "/avatars/0304a9f6eb7f5dee4d933d03222f94e9.svg",
"fullname": "Weigao Sun",
"isPro": false,
"type": "user",
"user": "weigao266"
}
},
{
"_id": "67b6dc1ba7567156c6547882",
"hidden": false,
"name": "Disen Lan",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:52:54.472Z",
"user": {
"_id": "66ea643899af9ac3463639b1",
"avatarUrl": "/avatars/252d470e761a57834dee3dbc60dfefed.svg",
"fullname": "Disen Lan",
"isPro": false,
"type": "user",
"user": "landisen"
}
},
{
"_id": "67b6dc1ba7567156c6547883",
"hidden": false,
"name": "Jiaxi Hu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:08:08.108Z",
"user": {
"_id": "665dc35752ff9daa9ba5a4ed",
"avatarUrl": "/avatars/df8b01879d97e599b610fa51414d3a18.svg",
"fullname": "Hu Jiaxi",
"isPro": false,
"type": "user",
"user": "Jiaxihu2"
}
},
{
"_id": "67b6dc1ba7567156c6547884",
"hidden": false,
"name": "Yu Cheng",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T12:53:55 | MoM: Linear Sequence Modeling with Mixture-of-Memories | Linear sequence modeling methods, such as linear attention, state space
modeling, and linear RNNs, offer significant efficiency improvements by
reducing the complexity of training and inference. However, these methods
typically compress the entire input sequence into a single fixed-size memory
state, which leads to suboptimal performance on recall-intensive downstream
tasks. Drawing inspiration from neuroscience, particularly the brain's ability
to maintain robust long-term memory while mitigating "memory interference", we
introduce a novel architecture called Mixture-of-Memories (MoM). MoM utilizes
multiple independent memory states, with a router network directing input
tokens to specific memory states. This approach greatly enhances the overall
memory capacity while minimizing memory interference. As a result, MoM performs
exceptionally well on recall-intensive tasks, surpassing existing linear
sequence modeling techniques. Despite incorporating multiple memory states, the
computation of each memory state remains linear in complexity, allowing MoM to
retain the linear-complexity advantage during training, while
constant-complexity during inference. Our experimental results show that MoM
significantly outperforms current linear sequence models on downstream language
tasks, particularly recall-intensive tasks, and even achieves performance
comparable to Transformer models. The code is released at
https://github.com/OpenSparseLLMs/MoM and is also released as a part of
https://github.com/OpenSparseLLMs/Linear-MoE. | 33 | 67b6dc1ca7567156c65478b8 | null | https://github.com/OpenSparseLLMs/MoM |
|
2025-02-20T01:20:46.431000 | Presumed Cultural Identity: How Names Shape LLM Responses | 2 | {
"_id": "60c50f18754747f54fa37114",
"avatarUrl": "/avatars/648ae58b81806dbd93a68546666047e3.svg",
"followerCount": 1,
"fullname": "Siddhesh",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "sidicity",
"type": "user"
} | false | null | 2502.11995 | [
{
"_id": "67b65bbe0d878eff1a6b111d",
"hidden": false,
"name": "Siddhesh Pawar",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:11:39.727Z",
"user": {
"_id": "661e2ac200798c2e33cc49a5",
"avatarUrl": "/avatars/8e5e1672b36f86bb4ad7a7e22e8d4f4d.svg",
"fullname": "Siddhesh Pawar",
"isPro": false,
"type": "user",
"user": "Siddheshmp06"
}
},
{
"_id": "67b65bbe0d878eff1a6b111e",
"hidden": false,
"name": "Arnav Arora",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b65bbe0d878eff1a6b111f",
"hidden": false,
"name": "Lucie-Aimée Kaffee",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:53:09.702Z",
"user": {
"_id": "6531310497d7f1b4a083de7b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/ux7NRFAbgnlIVNh-Cbv9s.png",
"fullname": "Lucie-Aimée Kaffee",
"isPro": false,
"type": "user",
"user": "frimelle"
}
},
{
"_id": "67b65bbe0d878eff1a6b1120",
"hidden": false,
"name": "Isabelle Augenstein",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:36:32.278Z",
"user": {
"_id": "608918b7df398c3b285ce960",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1621507769190-608918b7df398c3b285ce960.jpeg",
"fullname": "Isabelle Augenstein",
"isPro": false,
"type": "user",
"user": "IAugenstein"
}
}
]
| 2025-02-17T16:35:15 | Presumed Cultural Identity: How Names Shape LLM Responses | Names are deeply tied to human identity. They can serve as markers of
individuality, cultural heritage, and personal history. However, using names as
a core indicator of identity can lead to over-simplification of complex
identities. When interacting with LLMs, user names are an important point of
information for personalisation. Names can enter chatbot conversations through
direct user input (requested by chatbots), as part of task contexts such as CV
reviews, or as built-in memory features that store user information for
personalisation. We study biases associated with names by measuring cultural
presumptions in the responses generated by LLMs when presented with common
suggestion-seeking queries, which might involve making assumptions about the
user. Our analyses demonstrate strong assumptions about cultural identity
associated with names present in LLM generations across multiple cultures. Our
work has implications for designing more nuanced personalisation systems that
avoid reinforcing stereotypes while maintaining meaningful customisation. | 10 | 67b65bbf0d878eff1a6b1174 | null | null |
|
2025-02-20T01:07:44.785000 | SongGen: A Single Stage Auto-regressive Transformer for Text-to-Song Generation | 2 | {
"_id": "64b4eec4faa3181a5eab9c46",
"avatarUrl": "/avatars/bcc9bf5cbf67546ad2b4c9ec8b96ac96.svg",
"followerCount": 16,
"fullname": "Jiaqi Wang",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "myownskyW7",
"type": "user"
} | true | null | 2502.13128 | [
{
"_id": "67b6c696e9b901edeaf320d5",
"hidden": false,
"name": "Zihan Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:07:49.211Z",
"user": {
"_id": "65f33b1c9f7970ccc0234cbf",
"avatarUrl": "/avatars/99fbab303912e3674663251c04279907.svg",
"fullname": "Zihan Liu",
"isPro": false,
"type": "user",
"user": "zihanliu"
}
},
{
"_id": "67b6c696e9b901edeaf320d6",
"hidden": false,
"name": "Shuangrui Ding",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:07:35.659Z",
"user": {
"_id": "65a7c0335e79abfa2ec30c52",
"avatarUrl": "/avatars/2f62f83f9c5c4cc9444571f067cd85b7.svg",
"fullname": "Shuangrui Ding",
"isPro": true,
"type": "user",
"user": "Mar2Ding"
}
},
{
"_id": "67b6c696e9b901edeaf320d7",
"hidden": false,
"name": "Zhixiong Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:07:29.113Z",
"user": {
"_id": "64b75f59324038715942628b",
"avatarUrl": "/avatars/809c533138c2a69acbc3e42a143bbf34.svg",
"fullname": "Zhixiong Zhang",
"isPro": false,
"type": "user",
"user": "zzxlmlzcg2"
}
},
{
"_id": "67b6c696e9b901edeaf320d8",
"hidden": false,
"name": "Xiaoyi Dong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6c696e9b901edeaf320d9",
"hidden": false,
"name": "Pan Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6c696e9b901edeaf320da",
"hidden": false,
"name": "Yuhang Zang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:07:11.149Z",
"user": {
"_id": "63859cf3b2906edaf83af9f0",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63859cf3b2906edaf83af9f0/iUQm5FAomzqYi6fkqIn9F.jpeg",
"fullname": "Yuhang Zang",
"isPro": false,
"type": "user",
"user": "yuhangzang"
}
},
{
"_id": "67b6c696e9b901edeaf320db",
"hidden": false,
"name": "Yuhang Cao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:07:04.833Z",
"user": {
"_id": "65000bef18830fabea469fdd",
"avatarUrl": "/avatars/b320c77dfad039d9f9c54127f610d44f.svg",
"fullname": "Cao Yuhang",
"isPro": false,
"type": "user",
"user": "yhcao"
}
},
{
"_id": "67b6c696e9b901edeaf320dc",
"hidden": false,
"name": "Dahua Lin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:06:53.745Z",
"user": {
"_id": "636317ed80c1a705a6eff396",
"avatarUrl": "/avatars/3db090e101b916d9256d0d3e043db71d.svg",
"fullname": "Dahua Lin",
"isPro": false,
"type": "user",
"user": "lindahua"
}
},
{
"_id": "67b6c696e9b901edeaf320dd",
"hidden": false,
"name": "Jiaqi Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:06:47.581Z",
"user": {
"_id": "64b4eec4faa3181a5eab9c46",
"avatarUrl": "/avatars/bcc9bf5cbf67546ad2b4c9ec8b96ac96.svg",
"fullname": "Jiaqi Wang",
"isPro": true,
"type": "user",
"user": "myownskyW7"
}
}
]
| 2025-02-18T18:52:21 | SongGen: A Single Stage Auto-regressive Transformer for Text-to-Song
Generation | Text-to-song generation, the task of creating vocals and accompaniment from
textual inputs, poses significant challenges due to domain complexity and data
scarcity. Existing approaches often employ multi-stage generation procedures,
resulting in cumbersome training and inference pipelines. In this paper, we
propose SongGen, a fully open-source, single-stage auto-regressive transformer
designed for controllable song generation. The proposed model facilitates
fine-grained control over diverse musical attributes, including lyrics and
textual descriptions of instrumentation, genre, mood, and timbre, while also
offering an optional three-second reference clip for voice cloning. Within a
unified auto-regressive framework, SongGen supports two output modes: mixed
mode, which generates a mixture of vocals and accompaniment directly, and
dual-track mode, which synthesizes them separately for greater flexibility in
downstream applications. We explore diverse token pattern strategies for each
mode, leading to notable improvements and valuable insights. Furthermore, we
design an automated data preprocessing pipeline with effective quality control.
To foster community engagement and future research, we will release our model
weights, training code, annotated data, and preprocessing pipeline. The
generated samples are showcased on our project page at
https://liuzh-19.github.io/SongGen/ , and the code will be available at
https://github.com/LiuZH-19/SongGen . | 37 | 67b6c698e9b901edeaf321a7 | null | null |
|
2025-02-19T23:54:57.669000 | Why Safeguarded Ships Run Aground? Aligned Large Language Models' Safety Mechanisms Tend to Be Anchored in The Template Region | 2 | {
"_id": "631326d6289cf15634c52369",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/631326d6289cf15634c52369/lmPWGHLsQ36H556cqcXjT.jpeg",
"followerCount": 7,
"fullname": "Cooper Leong",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "cooperleong00",
"type": "user"
} | true | null | 2502.13946 | [
{
"_id": "67b6b416b4ad845374143c31",
"hidden": false,
"name": "Chak Tou Leong",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:25:12.631Z",
"user": {
"_id": "631326d6289cf15634c52369",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/631326d6289cf15634c52369/lmPWGHLsQ36H556cqcXjT.jpeg",
"fullname": "Cooper Leong",
"isPro": false,
"type": "user",
"user": "cooperleong00"
}
},
{
"_id": "67b6b416b4ad845374143c32",
"hidden": false,
"name": "Qingyu Yin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:12:30.407Z",
"user": {
"_id": "6453cb22908e259483c0a061",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6453cb22908e259483c0a061/hMgdwZUsUbgquGalzPGzV.jpeg",
"fullname": "Qingyu_Yin",
"isPro": false,
"type": "user",
"user": "MikaStars39"
}
},
{
"_id": "67b6b416b4ad845374143c33",
"hidden": false,
"name": "Jian Wang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:53:05.099Z",
"user": {
"_id": "63a4117984a6a25c65bc2fff",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1675936454785-63a4117984a6a25c65bc2fff.jpeg",
"fullname": "Jian Wang",
"isPro": false,
"type": "user",
"user": "jwanglvy"
}
},
{
"_id": "67b6b416b4ad845374143c34",
"hidden": false,
"name": "Wenjie Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:12:36.692Z",
"user": {
"_id": "66a3710a4ee2a4c936315a5a",
"avatarUrl": "/avatars/ef8da8fb1031695d77d34a5d365aa177.svg",
"fullname": "Li",
"isPro": false,
"type": "user",
"user": "WenjieLi"
}
}
]
| 2025-02-19T18:42:45 | Why Safeguarded Ships Run Aground? Aligned Large Language Models' Safety
Mechanisms Tend to Be Anchored in The Template Region | The safety alignment of large language models (LLMs) remains vulnerable, as
their initial behavior can be easily jailbroken by even relatively simple
attacks. Since infilling a fixed template between the input instruction and
initial model output is a common practice for existing LLMs, we hypothesize
that this template is a key factor behind their vulnerabilities: LLMs'
safety-related decision-making overly relies on the aggregated information from
the template region, which largely influences these models' safety behavior. We
refer to this issue as template-anchored safety alignment. In this paper, we
conduct extensive experiments and verify that template-anchored safety
alignment is widespread across various aligned LLMs. Our mechanistic analyses
demonstrate how it leads to models' susceptibility when encountering
inference-time jailbreak attacks. Furthermore, we show that detaching safety
mechanisms from the template region is promising in mitigating vulnerabilities
to jailbreak attacks. We encourage future research to develop more robust
safety alignment techniques that reduce reliance on the template region. | 9 | 67b6b416b4ad845374143c5b | null | null |
|
2025-02-19T23:35:06.194000 | Qwen2.5-VL Technical Report | 7 | {
"_id": "63451cf0a05b51f7ded25505",
"avatarUrl": "/avatars/dec4bbee4a82b773fc58dfc2dce9dbeb.svg",
"followerCount": 14,
"fullname": "shuai bai",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "bluelike",
"type": "user"
} | true | null | 2502.13923 | [
{
"_id": "67b6b0688b56622e70b9e83e",
"hidden": false,
"name": "Shuai Bai",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T15:54:00.062Z",
"user": {
"_id": "63451cf0a05b51f7ded25505",
"avatarUrl": "/avatars/dec4bbee4a82b773fc58dfc2dce9dbeb.svg",
"fullname": "shuai bai",
"isPro": false,
"type": "user",
"user": "bluelike"
}
},
{
"_id": "67b6b0688b56622e70b9e83f",
"hidden": false,
"name": "Keqin Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T15:54:14.529Z",
"user": {
"_id": "6461d675681b2e19b6acb5a5",
"avatarUrl": "/avatars/0d95d65d30f6672ec09dc92155324d7f.svg",
"fullname": "Keqin Chen",
"isPro": false,
"type": "user",
"user": "chenkq"
}
},
{
"_id": "67b6b0688b56622e70b9e840",
"hidden": false,
"name": "Xuejing Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e841",
"hidden": false,
"name": "Jialin Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:00:48.724Z",
"user": {
"_id": "6634979161776e1d8d35b16c",
"avatarUrl": "/avatars/32a1fac0016445959c2a062c1ab76ab9.svg",
"fullname": "jialinwang",
"isPro": false,
"type": "user",
"user": "jialinwangpku"
}
},
{
"_id": "67b6b0688b56622e70b9e842",
"hidden": false,
"name": "Wenbin Ge",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:00:57.101Z",
"user": {
"_id": "634d06c6f0a69955f662e641",
"avatarUrl": "/avatars/5a0af8af0a21d2a93192f4a3c430fc60.svg",
"fullname": "Wenbin Ge",
"isPro": false,
"type": "user",
"user": "gewenbin292"
}
},
{
"_id": "67b6b0688b56622e70b9e843",
"hidden": false,
"name": "Sibo Song",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:01:03.946Z",
"user": {
"_id": "62e38a600c2a907c388d2264",
"avatarUrl": "/avatars/43250732c7c9b3737067c2f0a8ba4ec5.svg",
"fullname": "Sibo Song",
"isPro": false,
"type": "user",
"user": "StefanSong"
}
},
{
"_id": "67b6b0688b56622e70b9e844",
"hidden": false,
"name": "Kai Dang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:01:10.991Z",
"user": {
"_id": "6712930f0fac3235c56edf5b",
"avatarUrl": "/avatars/cafe7cb56ce7c3b2572f5f2d0b89357a.svg",
"fullname": "kai dang",
"isPro": false,
"type": "user",
"user": "1vk5i"
}
},
{
"_id": "67b6b0688b56622e70b9e845",
"hidden": false,
"name": "Peng Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e846",
"hidden": false,
"name": "Shijie Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e847",
"hidden": false,
"name": "Jun Tang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e848",
"hidden": false,
"name": "Humen Zhong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e849",
"hidden": false,
"name": "Yuanzhi Zhu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:01:55.087Z",
"user": {
"_id": "627d2723401f42c57b6b7c0c",
"avatarUrl": "/avatars/6ff754e56aaee63d8572881a6a966171.svg",
"fullname": "Yuanzhi Zhu",
"isPro": false,
"type": "user",
"user": "Yuanzhi"
}
},
{
"_id": "67b6b0688b56622e70b9e84a",
"hidden": false,
"name": "Mingkun Yang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:35:44.878Z",
"user": {
"_id": "6417fa211f1f3b0fa811edc0",
"avatarUrl": "/avatars/fa9e1ef1472a736c2ceebe12b77d6c89.svg",
"fullname": "Mingkun Yang",
"isPro": false,
"type": "user",
"user": "ayumiymk"
}
},
{
"_id": "67b6b0688b56622e70b9e84b",
"hidden": false,
"name": "Zhaohai Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e84c",
"hidden": false,
"name": "Jianqiang Wan",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:02:09.102Z",
"user": {
"_id": "665ea9f161b34f33e04648ef",
"avatarUrl": "/avatars/d1d7b9feada68bcb6df5484c1bb3bba4.svg",
"fullname": "Jianqiang Wang ",
"isPro": false,
"type": "user",
"user": "xbetax"
}
},
{
"_id": "67b6b0688b56622e70b9e84d",
"hidden": false,
"name": "Pengfei Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e84e",
"hidden": false,
"name": "Wei Ding",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:02:48.942Z",
"user": {
"_id": "676b7679c20c09a8c72e4ade",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/bnUqHtJngTLyC6EHD-BP9.png",
"fullname": "weiding",
"isPro": false,
"type": "user",
"user": "weiding"
}
},
{
"_id": "67b6b0688b56622e70b9e84f",
"hidden": false,
"name": "Zheren Fu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T10:49:47.484Z",
"user": {
"_id": "63ee22e75f1300034ddaaf54",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1676550873969-noauth.jpeg",
"fullname": "Zheren Fu",
"isPro": false,
"type": "user",
"user": "darkpromise"
}
},
{
"_id": "67b6b0688b56622e70b9e850",
"hidden": false,
"name": "Yiheng Xu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:03:12.665Z",
"user": {
"_id": "601d29ab913ad3afd7b7ddb8",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1620447944896-601d29ab913ad3afd7b7ddb8.jpeg",
"fullname": "Yiheng Xu",
"isPro": true,
"type": "user",
"user": "ranpox"
}
},
{
"_id": "67b6b0688b56622e70b9e851",
"hidden": false,
"name": "Jiabo Ye",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e852",
"hidden": false,
"name": "Xi Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e853",
"hidden": false,
"name": "Tianbao Xie",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:03:30.898Z",
"user": {
"_id": "618767e4238063b4615d042b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1636263880877-noauth.jpeg",
"fullname": "Tianbao Xie",
"isPro": true,
"type": "user",
"user": "tianbaoxiexxx"
}
},
{
"_id": "67b6b0688b56622e70b9e854",
"hidden": false,
"name": "Zesen Cheng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:03:38.558Z",
"user": {
"_id": "65b2529285b6c21448a10d65",
"avatarUrl": "/avatars/1b09e2742aecce1bbdc57f0c4504cf38.svg",
"fullname": "Zesen Cheng",
"isPro": false,
"type": "user",
"user": "ClownRat"
}
},
{
"_id": "67b6b0688b56622e70b9e855",
"hidden": false,
"name": "Hang Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e856",
"hidden": false,
"name": "Zhibo Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b0688b56622e70b9e857",
"hidden": false,
"name": "Haiyang Xu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:35:42.372Z",
"user": {
"_id": "645b10e80c73ea27d13f7aca",
"avatarUrl": "/avatars/95e565306472a15067440b5b43e07a6f.svg",
"fullname": "xuhaiyang",
"isPro": false,
"type": "user",
"user": "xhyandwyy"
}
},
{
"_id": "67b6b0688b56622e70b9e858",
"hidden": false,
"name": "Junyang Lin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:03:53.317Z",
"user": {
"_id": "620760a26e3b7210c2ff1943",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/620760a26e3b7210c2ff1943/VC-rKqimF6yxGESNVlPoR.jpeg",
"fullname": "Junyang Lin",
"isPro": false,
"type": "user",
"user": "JustinLin610"
}
}
]
| 2025-02-19T18:00:14 | Qwen2.5-VL Technical Report | We introduce Qwen2.5-VL, the latest flagship model of Qwen vision-language
series, which demonstrates significant advancements in both foundational
capabilities and innovative functionalities. Qwen2.5-VL achieves a major leap
forward in understanding and interacting with the world through enhanced visual
recognition, precise object localization, robust document parsing, and
long-video comprehension. A standout feature of Qwen2.5-VL is its ability to
localize objects using bounding boxes or points accurately. It provides robust
structured data extraction from invoices, forms, and tables, as well as
detailed analysis of charts, diagrams, and layouts. To handle complex inputs,
Qwen2.5-VL introduces dynamic resolution processing and absolute time encoding,
enabling it to process images of varying sizes and videos of extended durations
(up to hours) with second-level event localization. This allows the model to
natively perceive spatial scales and temporal dynamics without relying on
traditional normalization techniques. By training a native dynamic-resolution
Vision Transformer (ViT) from scratch and incorporating Window Attention, we
reduce computational overhead while maintaining native resolution. As a result,
Qwen2.5-VL excels not only in static image and document understanding but also
as an interactive visual agent capable of reasoning, tool usage, and task
execution in real-world scenarios such as operating computers and mobile
devices. Qwen2.5-VL is available in three sizes, addressing diverse use cases
from edge AI to high-performance computing. The flagship Qwen2.5-VL-72B model
matches state-of-the-art models like GPT-4o and Claude 3.5 Sonnet, particularly
excelling in document and diagram understanding. Additionally, Qwen2.5-VL
maintains robust linguistic performance, preserving the core language
competencies of the Qwen2.5 LLM. | 154 | 67b6b0688b56622e70b9e875 | null | null |
|
2025-02-19T23:34:43.424000 | Is That Your Final Answer? Test-Time Scaling Improves Selective Question Answering | 4 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | true | null | 2502.13962 | [
{
"_id": "67b691751f861500916ecd5d",
"hidden": false,
"name": "William Jurayj",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:36:09.674Z",
"user": {
"_id": "6372bc95c4267fd7cd77f4d0",
"avatarUrl": "/avatars/17a24af68f45487e601687d777b352b6.svg",
"fullname": "William Jurayj",
"isPro": false,
"type": "user",
"user": "wjurayj"
}
},
{
"_id": "67b691751f861500916ecd5e",
"hidden": false,
"name": "Jeffrey Cheng",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T15:53:07.241Z",
"user": {
"_id": "65f28eebf5cf26fe0632ce67",
"avatarUrl": "/avatars/cd970cfe4215374c82d47df57ac30795.svg",
"fullname": "Jeffrey Cheng",
"isPro": false,
"type": "user",
"user": "nexync"
}
},
{
"_id": "67b691751f861500916ecd5f",
"hidden": false,
"name": "Benjamin Van Durme",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T18:58:31 | Is That Your Final Answer? Test-Time Scaling Improves Selective Question
Answering | Scaling the test-time compute of large language models has demonstrated
impressive performance on reasoning benchmarks. However, existing evaluations
of test-time scaling make the strong assumption that a reasoning system should
always give an answer to any question provided. This overlooks concerns about
whether a model is confident in its answer, and whether it is appropriate to
always provide a response. To address these concerns, we extract confidence
scores during reasoning for thresholding model responses. We find that
increasing compute budget at inference time not only helps models answer more
questions correctly, but also increases confidence in correct responses. We
then extend the current paradigm of zero-risk responses during evaluation by
considering settings with non-zero levels of response risk, and suggest a
recipe for reporting evaluations under these settings. | 28 | 67b691761f861500916ecd8e | null | null |
|
2025-02-19T23:31:36.410000 | Thinking Preference Optimization | 4 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | false | null | 2502.13173 | [
{
"_id": "67b6b014f7e569081326494f",
"hidden": false,
"name": "Wang Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b014f7e5690813264950",
"hidden": false,
"name": "Hongye Jin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b014f7e5690813264951",
"hidden": false,
"name": "Jingfeng Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b014f7e5690813264952",
"hidden": false,
"name": "Vipin Chaudhary",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6b014f7e5690813264953",
"hidden": true,
"name": "Xiaotian Han",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:12:59.725Z",
"user": {
"_id": "650dde4ce14eeb01d42b37a1",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/650dde4ce14eeb01d42b37a1/n5Yv24uofZ2XJjXdYCrKd.png",
"fullname": "Xiaotian Han",
"isPro": false,
"type": "user",
"user": "xiaotianhan"
}
}
]
| 2025-02-17T19:56:21 | Thinking Preference Optimization | Supervised Fine-Tuning (SFT) has been a go-to and effective method for
enhancing long chain-of-thought (CoT) reasoning in relatively small LLMs by
fine-tuning them with long CoT responses from larger LLMs. To continually
improve reasoning abilities, we can either collect new high-quality long CoT
reasoning SFT data or repeatedly train on existing SFT datasets. However,
acquiring new long CoT SFT data is costly and limited, while repeated training
often results in a performance plateau or decline. To further boost the
performance with the SFT data, we propose Thinking Preference Optimization
(ThinkPO), a simple yet effective post-SFT method that enhances long CoT
reasoning without requiring new long CoT responses. Instead, ThinkPO utilizes
readily available or easily obtainable short CoT reasoning responses as
rejected answers and long CoT responses as chosen answers for the same
question. It then applies direct preference optimization to encourage the model
to favor longer reasoning outputs. Experiments show that ThinkPO further
improves the reasoning performance of SFT-ed models, e.g. it increases math
reasoning accuracy of SFT-ed models by 8.6% and output length by 25.9%.
Notably, ThinkPO is capable of continually boosting the performance of the
publicly distilled SFT model, e.g., increasing the official
DeepSeek-R1-Distill-Qwen-7B's performance on MATH500 from 87.4% to 91.2%. | 17 | 67b6b015f7e56908132649a0 | null | null |
|
2025-02-19T23:18:32.647000 | NExT-Mol: 3D Diffusion Meets 1D Language Modeling for 3D Molecule Generation | 2 | {
"_id": "6310a3cd531cc21f9e06de6a",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6310a3cd531cc21f9e06de6a/aTGMx3O41lUARK9s3dAik.jpeg",
"followerCount": 3,
"fullname": "Zhiyuan Liu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "acharkq",
"type": "user"
} | true | null | 2502.12638 | [
{
"_id": "67b6acdb3a3df2f965e7af0b",
"hidden": false,
"name": "Zhiyuan Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:43:04.070Z",
"user": {
"_id": "6310a3cd531cc21f9e06de6a",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6310a3cd531cc21f9e06de6a/aTGMx3O41lUARK9s3dAik.jpeg",
"fullname": "Zhiyuan Liu",
"isPro": false,
"type": "user",
"user": "acharkq"
}
},
{
"_id": "67b6acdb3a3df2f965e7af0c",
"hidden": false,
"name": "Yanchen Luo",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:48.054Z",
"user": {
"_id": "64f04a28f3cd962c21726459",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/MOTc7SWbzc4jdJbMcWMcK.jpeg",
"fullname": "LuoYanchen",
"isPro": false,
"type": "user",
"user": "lyc0930"
}
},
{
"_id": "67b6acdb3a3df2f965e7af0d",
"hidden": false,
"name": "Han Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6acdb3a3df2f965e7af0e",
"hidden": false,
"name": "Enzhi Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:42:41.957Z",
"user": {
"_id": "6522bdace3419abdcf8177f6",
"avatarUrl": "/avatars/872b22da6489959134b5449bb7ed9636.svg",
"fullname": "EnzhiZhang",
"isPro": false,
"type": "user",
"user": "EnzhiZhang"
}
},
{
"_id": "67b6acdb3a3df2f965e7af0f",
"hidden": false,
"name": "Sihang Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6acdb3a3df2f965e7af10",
"hidden": false,
"name": "Junfeng Fang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6acdb3a3df2f965e7af11",
"hidden": false,
"name": "Yaorui Shi",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:42:26.270Z",
"user": {
"_id": "63edd2d1f765928ceeb49057",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1676530369930-noauth.png",
"fullname": "Yaorui SHI",
"isPro": false,
"type": "user",
"user": "yrshi"
}
},
{
"_id": "67b6acdb3a3df2f965e7af12",
"hidden": false,
"name": "Xiang Wang",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-20T04:17:33.860Z",
"user": {
"_id": "65fca775fa59bdf4737b1a84",
"avatarUrl": "/avatars/a161b510bde8f57e7686cbb0b4aa6a52.svg",
"fullname": "Xiang Wang",
"isPro": false,
"type": "user",
"user": "xiangwang1223"
}
},
{
"_id": "67b6acdb3a3df2f965e7af13",
"hidden": false,
"name": "Kenji Kawaguchi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6acdb3a3df2f965e7af14",
"hidden": false,
"name": "Tat-Seng Chua",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:42:12.905Z",
"user": {
"_id": "6570ae84c4993b8fb96f41a8",
"avatarUrl": "/avatars/21f7d79d46ac4df0ecff8eca7678b33f.svg",
"fullname": "Tat-Seng Chua",
"isPro": false,
"type": "user",
"user": "chuats"
}
}
]
| 2025-02-18T08:40:13 | NExT-Mol: 3D Diffusion Meets 1D Language Modeling for 3D Molecule
Generation | 3D molecule generation is crucial for drug discovery and material design.
While prior efforts focus on 3D diffusion models for their benefits in modeling
continuous 3D conformers, they overlook the advantages of 1D SELFIES-based
Language Models (LMs), which can generate 100% valid molecules and leverage the
billion-scale 1D molecule datasets. To combine these advantages for 3D molecule
generation, we propose a foundation model -- NExT-Mol: 3D Diffusion Meets 1D
Language Modeling for 3D Molecule Generation. NExT-Mol uses an extensively
pretrained molecule LM for 1D molecule generation, and subsequently predicts
the generated molecule's 3D conformers with a 3D diffusion model. We enhance
NExT-Mol's performance by scaling up the LM's model size, refining the
diffusion neural architecture, and applying 1D to 3D transfer learning.
Notably, our 1D molecule LM significantly outperforms baselines in
distributional similarity while ensuring validity, and our 3D diffusion model
achieves leading performances in conformer prediction. Given these improvements
in 1D and 3D modeling, NExT-Mol achieves a 26% relative improvement in 3D FCD
for de novo 3D generation on GEOM-DRUGS, and a 13% average relative gain for
conditional 3D generation on QM9-2014. Our codes and pretrained checkpoints are
available at https://github.com/acharkq/NExT-Mol. | 8 | 67b6acdd3a3df2f965e7af85 | null | null |
|
2025-02-19T23:07:01.367000 | AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence | 2 | {
"_id": "6529f79e802e3d1a4f8ec662",
"avatarUrl": "/avatars/d05320c370a6497d8792ef5acb563dd5.svg",
"followerCount": 2,
"fullname": "Yuliang Liu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "yuliang03181",
"type": "user"
} | true | null | 2502.13943 | [
{
"_id": "67b6a9a7c721bee91cac2888",
"hidden": false,
"name": "Yuliang Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:11:40.282Z",
"user": {
"_id": "6529f79e802e3d1a4f8ec662",
"avatarUrl": "/avatars/d05320c370a6497d8792ef5acb563dd5.svg",
"fullname": "Yuliang Liu",
"isPro": false,
"type": "user",
"user": "yuliang03181"
}
},
{
"_id": "67b6a9a7c721bee91cac2889",
"hidden": false,
"name": "Junjie Lu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T13:05:14.854Z",
"user": {
"_id": "660e1bb6e089e44d2906a7e8",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/660e1bb6e089e44d2906a7e8/MJU8nHQ0xgiPzO8J6L8EC.jpeg",
"fullname": "Junjie Lu",
"isPro": false,
"type": "user",
"user": "Lux0926"
}
},
{
"_id": "67b6a9a7c721bee91cac288a",
"hidden": false,
"name": "Zhaoling Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a9a7c721bee91cac288b",
"hidden": false,
"name": "Chaofeng Qu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:11:10.302Z",
"user": {
"_id": "65d0ae74c4d2b2e40281256e",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/65d0ae74c4d2b2e40281256e/nNTJkAkHsu9Fru4CpRphC.jpeg",
"fullname": "Chaofeng QU",
"isPro": false,
"type": "user",
"user": "kylebrovloski"
}
},
{
"_id": "67b6a9a7c721bee91cac288c",
"hidden": false,
"name": "Jason Klein Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:11:16.565Z",
"user": {
"_id": "6731c04d5f55903a1d8c307c",
"avatarUrl": "/avatars/704b1d628c55e3141194b08736f21267.svg",
"fullname": "Jason Klein Liu",
"isPro": false,
"type": "user",
"user": "jasonkleinlove"
}
},
{
"_id": "67b6a9a7c721bee91cac288d",
"hidden": false,
"name": "Chonghan Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a9a7c721bee91cac288e",
"hidden": false,
"name": "Zefan Cai",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a9a7c721bee91cac288f",
"hidden": false,
"name": "Yunhui Xia",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a9a7c721bee91cac2890",
"hidden": false,
"name": "Li Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a9a7c721bee91cac2891",
"hidden": false,
"name": "Jiang Bian",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T17:00:08.549Z",
"user": {
"_id": "63f253f8f4e30ffd2bd308fb",
"avatarUrl": "/avatars/303f4c7ee588f638acf78a7966786e1e.svg",
"fullname": "Jiang Bian",
"isPro": false,
"type": "user",
"user": "bianjiang"
}
},
{
"_id": "67b6a9a7c721bee91cac2892",
"hidden": false,
"name": "Chuheng Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:59:49.312Z",
"user": {
"_id": "64c32727c370f29a10334d35",
"avatarUrl": "/avatars/a3284bdd2e51f437433b89a96a904448.svg",
"fullname": "ZhangChuheng",
"isPro": false,
"type": "user",
"user": "zhangchuheng123"
}
},
{
"_id": "67b6a9a7c721bee91cac2893",
"hidden": false,
"name": "Wei Shen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a9a7c721bee91cac2894",
"hidden": false,
"name": "Zhouhan Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T18:35:55 | AdaptiveStep: Automatically Dividing Reasoning Step through Model
Confidence | Current approaches for training Process Reward Models (PRMs) often involve
breaking down responses into multiple reasoning steps using rule-based
techniques, such as using predefined placeholder tokens or setting the
reasoning step's length into a fixed size. These approaches overlook the fact
that specific words do not typically mark true decision points in a text. To
address this, we propose AdaptiveStep, a method that divides reasoning steps
based on the model's confidence in predicting the next word. This division
method provides more decision-making information at each step, enhancing
downstream tasks, such as reward model learning. Moreover, our method does not
require manual annotation. We demonstrate its effectiveness through experiments
with AdaptiveStep-trained PRMs in mathematical reasoning and code generation
tasks. Experimental results indicate that the outcome PRM achieves
state-of-the-art Best-of-N performance, surpassing greedy search strategy with
token-level value-guided decoding, while also reducing construction costs by
over 30% compared to existing open-source PRMs. In addition, we provide a
thorough analysis and case study on the PRM's performance, transferability, and
generalization capabilities. | 7 | 67b6a9a8c721bee91cac28e7 | null | null |
|
2025-02-19T22:57:23.298000 | Craw4LLM: Efficient Web Crawling for LLM Pretraining | 2 | {
"_id": "6135eeeb5bc6ecdf86b60f0d",
"avatarUrl": "/avatars/43cedcf20ab6b0801a662787400e1384.svg",
"followerCount": 7,
"fullname": "Shi Yu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "yushi",
"type": "user"
} | true | null | 2502.13347 | [
{
"_id": "67b6a7e83ef3656c48f149b9",
"hidden": false,
"name": "Shi Yu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:35:47.487Z",
"user": {
"_id": "6135eeeb5bc6ecdf86b60f0d",
"avatarUrl": "/avatars/43cedcf20ab6b0801a662787400e1384.svg",
"fullname": "Shi Yu",
"isPro": false,
"type": "user",
"user": "yushi"
}
},
{
"_id": "67b6a7e83ef3656c48f149ba",
"hidden": false,
"name": "Zhiyuan Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:09:12.683Z",
"user": {
"_id": "6310a3cd531cc21f9e06de6a",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6310a3cd531cc21f9e06de6a/aTGMx3O41lUARK9s3dAik.jpeg",
"fullname": "Zhiyuan Liu",
"isPro": false,
"type": "user",
"user": "acharkq"
}
},
{
"_id": "67b6a7e83ef3656c48f149bb",
"hidden": false,
"name": "Chenyan Xiong",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:09:19.014Z",
"user": {
"_id": "617ae49bc0ff6006217aa22e",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1635543252231-617ae49bc0ff6006217aa22e.jpeg",
"fullname": "Chenyan Xiong",
"isPro": false,
"type": "user",
"user": "xiongchenyan"
}
}
]
| 2025-02-19T00:31:43 | Craw4LLM: Efficient Web Crawling for LLM Pretraining | Web crawl is a main source of large language models' (LLMs) pretraining data,
but the majority of crawled web pages are discarded in pretraining due to low
data quality. This paper presents Crawl4LLM, an efficient web crawling method
that explores the web graph based on the preference of LLM pretraining.
Specifically, it leverages the influence of a webpage in LLM pretraining as the
priority score of the web crawler's scheduler, replacing the standard graph
connectivity based priority. Our experiments on a web graph containing 900
million webpages from a commercial search engine's index demonstrate the
efficiency of Crawl4LLM in obtaining high-quality pretraining data. With just
21% URLs crawled, LLMs pretrained on Crawl4LLM data reach the same downstream
performances of previous crawls, significantly reducing the crawling waste and
alleviating the burdens on websites. Our code is publicly available at
https://github.com/cxcscmu/Crawl4LLM. | 27 | 67b6a7e93ef3656c48f149f1 | null | null |
|
2025-02-19T22:42:06.502000 | Autellix: An Efficient Serving Engine for LLM Agents as General Programs | 2 | {
"_id": "654037be97949fd2304aab7f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/654037be97949fd2304aab7f/2cSME81gcwYa2OTeVlq5Q.jpeg",
"followerCount": 3,
"fullname": "Michael Luo",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "michaelzhiluo",
"type": "user"
} | true | null | 2502.13965 | [
{
"_id": "67b6a3fa09841367596a1db5",
"hidden": false,
"name": "Michael Luo",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:25:24.729Z",
"user": {
"_id": "654037be97949fd2304aab7f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/654037be97949fd2304aab7f/2cSME81gcwYa2OTeVlq5Q.jpeg",
"fullname": "Michael Luo",
"isPro": false,
"type": "user",
"user": "michaelzhiluo"
}
},
{
"_id": "67b6a3fa09841367596a1db6",
"hidden": false,
"name": "Xiaoxiang Shi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a3fa09841367596a1db7",
"hidden": false,
"name": "Colin Cai",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a3fa09841367596a1db8",
"hidden": false,
"name": "Tianjun Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a3fa09841367596a1db9",
"hidden": false,
"name": "Justin Wong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a3fa09841367596a1dba",
"hidden": false,
"name": "Yichuan Wang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:35:50.487Z",
"user": {
"_id": "626e3449e7914f0d5ea78ad1",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/626e3449e7914f0d5ea78ad1/pVzdmdPMpNcxuj94qiIvB.jpeg",
"fullname": "Yichuan",
"isPro": false,
"type": "user",
"user": "Chrisyichuan"
}
},
{
"_id": "67b6a3fa09841367596a1dbb",
"hidden": false,
"name": "Chi Wang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-26T08:38:54.531Z",
"user": {
"_id": "67bea54186192d5a029f029b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/mLRb5e_ucqlvPdexaVk7E.png",
"fullname": "Chi Wang",
"isPro": false,
"type": "user",
"user": "chi-wang"
}
},
{
"_id": "67b6a3fa09841367596a1dbc",
"hidden": false,
"name": "Yanping Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a3fa09841367596a1dbd",
"hidden": false,
"name": "Zhifeng Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6a3fa09841367596a1dbe",
"hidden": false,
"name": "Joseph E. Gonzalez",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:10:25.466Z",
"user": {
"_id": "645d2e8401f4eaab2a0878ce",
"avatarUrl": "/avatars/1273c5fb607b4b622a746a42692fa632.svg",
"fullname": "Joseph E. Gonzalez",
"isPro": false,
"type": "user",
"user": "ProfJoeyG"
}
},
{
"_id": "67b6a3fa09841367596a1dbf",
"hidden": false,
"name": "Ion Stoica",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-19T18:59:30 | Autellix: An Efficient Serving Engine for LLM Agents as General Programs | Large language model (LLM) applications are evolving beyond simple chatbots
into dynamic, general-purpose agentic programs, which scale LLM calls and
output tokens to help AI agents reason, explore, and solve complex tasks.
However, existing LLM serving systems ignore dependencies between programs and
calls, missing significant opportunities for optimization. Our analysis reveals
that programs submitted to LLM serving engines experience long cumulative wait
times, primarily due to head-of-line blocking at both the individual LLM
request and the program. To address this, we introduce Autellix, an LLM serving
system that treats programs as first-class citizens to minimize their
end-to-end latencies. Autellix intercepts LLM calls submitted by programs,
enriching schedulers with program-level context. We propose two scheduling
algorithms-for single-threaded and distributed programs-that preempt and
prioritize LLM calls based on their programs' previously completed calls. Our
evaluation demonstrates that across diverse LLMs and agentic workloads,
Autellix improves throughput of programs by 4-15x at the same latency compared
to state-of-the-art systems, such as vLLM. | 18 | 67b6a3fb09841367596a1e06 | null | null |
|
2025-02-19T22:27:22.403000 | SearchRAG: Can Search Engines Be Helpful for LLM-based Medical Question Answering? | 2 | {
"_id": "64beb6b6140491ca9f803ebf",
"avatarUrl": "/avatars/0daa2e813a13668b8b708cd8c12763d9.svg",
"followerCount": null,
"fullname": "Yucheng SHi",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "YuchengShi",
"type": "user"
} | true | null | 2502.13233 | [
{
"_id": "67b689aeba514d2c2c969289",
"hidden": false,
"name": "Yucheng Shi",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:36:18.925Z",
"user": {
"_id": "64beb6b6140491ca9f803ebf",
"avatarUrl": "/avatars/0daa2e813a13668b8b708cd8c12763d9.svg",
"fullname": "Yucheng SHi",
"isPro": false,
"type": "user",
"user": "YuchengShi"
}
},
{
"_id": "67b689aeba514d2c2c96928a",
"hidden": false,
"name": "Tianze Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b689aeba514d2c2c96928b",
"hidden": false,
"name": "Canyu Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:58:09.392Z",
"user": {
"_id": "6483af58571c2dcfa98cae82",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6483af58571c2dcfa98cae82/d8Vnh_EG1KxeEoNAAQXHZ.jpeg",
"fullname": "Canyu Chen",
"isPro": false,
"type": "user",
"user": "canyuchen"
}
},
{
"_id": "67b689aeba514d2c2c96928c",
"hidden": false,
"name": "Quanzheng Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b689aeba514d2c2c96928d",
"hidden": false,
"name": "Tianming Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b689aeba514d2c2c96928e",
"hidden": false,
"name": "Xiang Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b689aeba514d2c2c96928f",
"hidden": false,
"name": "Ninghao Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T19:12:15 | SearchRAG: Can Search Engines Be Helpful for LLM-based Medical Question
Answering? | Large Language Models (LLMs) have shown remarkable capabilities in general
domains but often struggle with tasks requiring specialized knowledge.
Conventional Retrieval-Augmented Generation (RAG) techniques typically retrieve
external information from static knowledge bases, which can be outdated or
incomplete, missing fine-grained clinical details essential for accurate
medical question answering. In this work, we propose SearchRAG, a novel
framework that overcomes these limitations by leveraging real-time search
engines. Our method employs synthetic query generation to convert complex
medical questions into search-engine-friendly queries and utilizes
uncertainty-based knowledge selection to filter and incorporate the most
relevant and informative medical knowledge into the LLM's input. Experimental
results demonstrate that our method significantly improves response accuracy in
medical question answering tasks, particularly for complex questions requiring
detailed and up-to-date knowledge. | 13 | 67b689aeba514d2c2c9692b9 | null | null |
|
2025-02-19T22:13:49.764000 | RAD: Training an End-to-End Driving Policy via Large-Scale 3DGS-based Reinforcement Learning | 2 | {
"_id": "6536187bd34e9f02b9df1c3b",
"avatarUrl": "/avatars/0b34d62868b93053b0a05062a018b5bd.svg",
"followerCount": 1,
"fullname": "Hao Gao",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Hao605",
"type": "user"
} | true | null | 2502.13144 | [
{
"_id": "67b55c7fba22c1ddbb8d5746",
"hidden": false,
"name": "Hao Gao",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:00:48.944Z",
"user": {
"_id": "6536187bd34e9f02b9df1c3b",
"avatarUrl": "/avatars/0b34d62868b93053b0a05062a018b5bd.svg",
"fullname": "Hao Gao",
"isPro": false,
"type": "user",
"user": "Hao605"
}
},
{
"_id": "67b55c7fba22c1ddbb8d5747",
"hidden": false,
"name": "Shaoyu Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:04:07.016Z",
"user": {
"_id": "67adc154a266b54c2835cceb",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/HxrPMqiiTWE3z8qhMlW_m.png",
"fullname": "Shaoyu Chen",
"isPro": false,
"type": "user",
"user": "Atan-0221"
}
},
{
"_id": "67b55c7fba22c1ddbb8d5748",
"hidden": false,
"name": "Bo Jiang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d5749",
"hidden": false,
"name": "Bencheng Liao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d574a",
"hidden": false,
"name": "Yiang Shi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d574b",
"hidden": false,
"name": "Xiaoyang Guo",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d574c",
"hidden": false,
"name": "Yuechuan Pu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d574d",
"hidden": false,
"name": "Haoran Yin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d574e",
"hidden": false,
"name": "Xiangyu Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:05:51.144Z",
"user": {
"_id": "6481ba38814ff581861335ef",
"avatarUrl": "/avatars/183e840c33c6c9f88a009ceb0aec697a.svg",
"fullname": "xiangyu",
"isPro": false,
"type": "user",
"user": "xiangyuli"
}
},
{
"_id": "67b55c7fba22c1ddbb8d574f",
"hidden": false,
"name": "Xinbang Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d5750",
"hidden": false,
"name": "Ying Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d5751",
"hidden": false,
"name": "Wenyu Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:05:33.229Z",
"user": {
"_id": "66c2e7fc934e2f07753542ac",
"avatarUrl": "/avatars/f6fa3f94435cf1c1d06daa6c925d07d0.svg",
"fullname": "LWY",
"isPro": false,
"type": "user",
"user": "wenyuliu"
}
},
{
"_id": "67b55c7fba22c1ddbb8d5752",
"hidden": false,
"name": "Qian Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55c7fba22c1ddbb8d5753",
"hidden": false,
"name": "Xinggang Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:05:15.793Z",
"user": {
"_id": "62600de6d47e3dbae32ce1ce",
"avatarUrl": "/avatars/a536417cfec6e10ac415091bd1829426.svg",
"fullname": "Xinggang Wang",
"isPro": false,
"type": "user",
"user": "xinggangw"
}
}
]
| 2025-02-18T18:59:21 | RAD: Training an End-to-End Driving Policy via Large-Scale 3DGS-based
Reinforcement Learning | Existing end-to-end autonomous driving (AD) algorithms typically follow the
Imitation Learning (IL) paradigm, which faces challenges such as causal
confusion and the open-loop gap. In this work, we establish a 3DGS-based
closed-loop Reinforcement Learning (RL) training paradigm. By leveraging 3DGS
techniques, we construct a photorealistic digital replica of the real physical
world, enabling the AD policy to extensively explore the state space and learn
to handle out-of-distribution scenarios through large-scale trial and error. To
enhance safety, we design specialized rewards that guide the policy to
effectively respond to safety-critical events and understand real-world causal
relationships. For better alignment with human driving behavior, IL is
incorporated into RL training as a regularization term. We introduce a
closed-loop evaluation benchmark consisting of diverse, previously unseen 3DGS
environments. Compared to IL-based methods, RAD achieves stronger performance
in most closed-loop metrics, especially 3x lower collision rate. Abundant
closed-loop results are presented at https://hgao-cv.github.io/RAD. | 36 | 67b55c80ba22c1ddbb8d579c | null | null |
|
2025-02-19T21:38:13.468000 | Small Models Struggle to Learn from Strong Reasoners | 6 | {
"_id": "653df1323479e9ebbe3eb6cc",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/653df1323479e9ebbe3eb6cc/K_g-r1iMRNKj99LXPuYF3.jpeg",
"followerCount": 11,
"fullname": "Zhangchen Xu",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "flydust",
"type": "user"
} | true | null | 2502.12143 | [
{
"_id": "67b4d05a9f8a8ab661450397",
"hidden": false,
"name": "Yuetai Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4d05a9f8a8ab661450398",
"hidden": false,
"name": "Xiang Yue",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4d05a9f8a8ab661450399",
"hidden": false,
"name": "Zhangchen Xu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:32.715Z",
"user": {
"_id": "653df1323479e9ebbe3eb6cc",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/653df1323479e9ebbe3eb6cc/K_g-r1iMRNKj99LXPuYF3.jpeg",
"fullname": "Zhangchen Xu",
"isPro": true,
"type": "user",
"user": "flydust"
}
},
{
"_id": "67b4d05a9f8a8ab66145039a",
"hidden": false,
"name": "Fengqing Jiang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:09:45.083Z",
"user": {
"_id": "6531e1021dd8ebbdc1a6fd8e",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6531e1021dd8ebbdc1a6fd8e/lIcl7zCPtzRsfiUh6uY1o.jpeg",
"fullname": "Fengqing Jiang",
"isPro": false,
"type": "user",
"user": "fqjiang"
}
},
{
"_id": "67b4d05a9f8a8ab66145039b",
"hidden": false,
"name": "Luyao Niu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:09:50.890Z",
"user": {
"_id": "666dfd4770f5a2cb4aefd12f",
"avatarUrl": "/avatars/fa0e0dbc203a21e58dda8fdb4cbc67ad.svg",
"fullname": "Luyao Niu",
"isPro": false,
"type": "user",
"user": "LNIU"
}
},
{
"_id": "67b4d05a9f8a8ab66145039c",
"hidden": false,
"name": "Bill Yuchen Lin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:09:56.614Z",
"user": {
"_id": "607f666a4ad99100d63ce35c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/607f666a4ad99100d63ce35c/QxhxnvfeV6efkxwUFHwjI.png",
"fullname": "Bill Yuchen Lin",
"isPro": false,
"type": "user",
"user": "yuchenlin"
}
},
{
"_id": "67b4d05a9f8a8ab66145039d",
"hidden": false,
"name": "Bhaskar Ramasubramanian",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4d05a9f8a8ab66145039e",
"hidden": false,
"name": "Radha Poovendran",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T18:56:15 | Small Models Struggle to Learn from Strong Reasoners | Large language models (LLMs) excel in complex reasoning tasks, and distilling
their reasoning capabilities into smaller models has shown promise. However, we
uncover an interesting phenomenon, which we term the Small Model Learnability
Gap: small models (leq3B parameters) do not consistently benefit from long
chain-of-thought (CoT) reasoning or distillation from larger models. Instead,
they perform better when fine-tuned on shorter, simpler reasoning chains that
better align with their intrinsic learning capacity. To address this, we
propose Mix Distillation, a simple yet effective strategy that balances
reasoning complexity by combining long and short CoT examples or reasoning from
both larger and smaller models. Our experiments demonstrate that Mix
Distillation significantly improves small model reasoning performance compared
to training on either data alone. These findings highlight the limitations of
direct strong model distillation and underscore the importance of adapting
reasoning complexity for effective reasoning capability transfer. | 28 | 67b4d05b9f8a8ab6614503cb | null | null |
|
2025-02-19T21:35:20.931000 | LongPO: Long Context Self-Evolution of Large Language Models through Short-to-Long Preference Optimization | 2 | {
"_id": "645475e2548f22be59847604",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/645475e2548f22be59847604/EhSurrZ25u31qQ2TVXQXt.jpeg",
"followerCount": 1,
"fullname": "Chen",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Guanzheng",
"type": "user"
} | true | null | 2502.13922 | [
{
"_id": "67b6948dbef24bad725b5d4b",
"hidden": false,
"name": "Guanzheng Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:08:30.816Z",
"user": {
"_id": "645475e2548f22be59847604",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/645475e2548f22be59847604/EhSurrZ25u31qQ2TVXQXt.jpeg",
"fullname": "Chen",
"isPro": false,
"type": "user",
"user": "Guanzheng"
}
},
{
"_id": "67b6948dbef24bad725b5d4c",
"hidden": false,
"name": "Xin Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6948dbef24bad725b5d4d",
"hidden": false,
"name": "Michael Qizhe Shieh",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b6948dbef24bad725b5d4e",
"hidden": false,
"name": "Lidong Bing",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-20T16:08:48.204Z",
"user": {
"_id": "6454685a548f22be598414c4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/eMjMWKJ-AouF7eY1-RzGF.jpeg",
"fullname": "Lidong Bing",
"isPro": false,
"type": "user",
"user": "LidongBing"
}
}
]
| 2025-02-19T17:59:03 | LongPO: Long Context Self-Evolution of Large Language Models through
Short-to-Long Preference Optimization | Large Language Models (LLMs) have demonstrated remarkable capabilities
through pretraining and alignment. However, superior short-context LLMs may
underperform in long-context scenarios due to insufficient long-context
alignment. This alignment process remains challenging due to the impracticality
of human annotation for extended contexts and the difficulty in balancing
short- and long-context performance. To address these challenges, we introduce
LongPO, that enables short-context LLMs to self-evolve to excel on long-context
tasks by internally transferring short-context capabilities. LongPO harnesses
LLMs to learn from self-generated short-to-long preference data, comprising
paired responses generated for identical instructions with long-context inputs
and their compressed short-context counterparts, respectively. This preference
reveals capabilities and potentials of LLMs cultivated during short-context
alignment that may be diminished in under-aligned long-context scenarios.
Additionally, LongPO incorporates a short-to-long KL constraint to mitigate
short-context performance decline during long-context alignment. When applied
to Mistral-7B-Instruct-v0.2 from 128K to 512K context lengths, LongPO fully
retains short-context performance and largely outperforms naive SFT and DPO in
both long- and short-context tasks. Specifically, \ourMethod-trained models can
achieve results on long-context benchmarks comparable to, or even surpassing,
those of superior LLMs (e.g., GPT-4-128K) that involve extensive long-context
annotation and larger parameter scales. | 25 | 67b6948ebef24bad725b5d84 | null | null |
|
2025-02-19T20:37:51.607000 | The Hidden Risks of Large Reasoning Models: A Safety Assessment of R1 | 2 | {
"_id": "64679a226192d39142245e5e",
"avatarUrl": "/avatars/05abee0b6317f100923936ca2099e9eb.svg",
"followerCount": 4,
"fullname": "Xin Eric Wang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "xw-eric",
"type": "user"
} | false | null | 2502.12659 | [
{
"_id": "67b68700ce3055c9c0fc2987",
"hidden": false,
"name": "Kaiwen Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68700ce3055c9c0fc2988",
"hidden": false,
"name": "Chengzhi Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68700ce3055c9c0fc2989",
"hidden": false,
"name": "Xuandong Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68700ce3055c9c0fc298a",
"hidden": false,
"name": "Shreedhar Jangam",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68700ce3055c9c0fc298b",
"hidden": false,
"name": "Jayanth Srinivasa",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68700ce3055c9c0fc298c",
"hidden": false,
"name": "Gaowen Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68700ce3055c9c0fc298d",
"hidden": false,
"name": "Dawn Song",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b68700ce3055c9c0fc298e",
"hidden": false,
"name": "Xin Eric Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T09:06:07 | The Hidden Risks of Large Reasoning Models: A Safety Assessment of R1 | The rapid development of large reasoning models, such as OpenAI-o3 and
DeepSeek-R1, has led to significant improvements in complex reasoning over
non-reasoning large language models~(LLMs). However, their enhanced
capabilities, combined with the open-source access of models like DeepSeek-R1,
raise serious safety concerns, particularly regarding their potential for
misuse. In this work, we present a comprehensive safety assessment of these
reasoning models, leveraging established safety benchmarks to evaluate their
compliance with safety regulations. Furthermore, we investigate their
susceptibility to adversarial attacks, such as jailbreaking and prompt
injection, to assess their robustness in real-world applications. Through our
multi-faceted analysis, we uncover four key findings: (1) There is a
significant safety gap between the open-source R1 models and the o3-mini model,
on both safety benchmark and attack, suggesting more safety effort on R1 is
needed. (2) The distilled reasoning model shows poorer safety performance
compared to its safety-aligned base models. (3) The stronger the model's
reasoning ability, the greater the potential harm it may cause when answering
unsafe questions. (4) The thinking process in R1 models pose greater safety
concerns than their final answers. Our study provides insights into the
security implications of reasoning models and highlights the need for further
advancements in R1 models' safety to close the gap. | 6 | 67b68701ce3055c9c0fc29e4 | null | null |
|
2025-02-19T18:20:05.946000 | Scaling Autonomous Agents via Automatic Reward Modeling And Planning | 2 | {
"_id": "654e024de113b04ba5c71e2f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/654e024de113b04ba5c71e2f/WH6S_gpQU6OXqDaiPpheK.jpeg",
"followerCount": 1,
"fullname": "Rui Sun",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "ThreeSR",
"type": "user"
} | true | null | 2502.12130 | [
{
"_id": "67b657d6a267b1a747a7fed6",
"hidden": false,
"name": "Zhenfang Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b657d6a267b1a747a7fed7",
"hidden": false,
"name": "Delin Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b657d6a267b1a747a7fed8",
"hidden": false,
"name": "Rui Sun",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:36:34.773Z",
"user": {
"_id": "654e024de113b04ba5c71e2f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/654e024de113b04ba5c71e2f/WH6S_gpQU6OXqDaiPpheK.jpeg",
"fullname": "Rui Sun",
"isPro": false,
"type": "user",
"user": "ThreeSR"
}
},
{
"_id": "67b657d6a267b1a747a7fed9",
"hidden": false,
"name": "Wenjun Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b657d6a267b1a747a7feda",
"hidden": false,
"name": "Chuang Gan",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T18:49:25 | Scaling Autonomous Agents via Automatic Reward Modeling And Planning | Large language models (LLMs) have demonstrated remarkable capabilities across
a range of text-generation tasks. However, LLMs still struggle with problems
requiring multi-step decision-making and environmental feedback, such as online
shopping, scientific reasoning, and mathematical problem-solving. Unlike pure
text data, collecting large-scale decision-making data is challenging.
Moreover, many powerful LLMs are only accessible through APIs, which hinders
their fine-tuning for agent tasks due to cost and complexity. To address LLM
agents' limitations, we propose a framework that can automatically learn a
reward model from the environment without human annotations. This model can be
used to evaluate the action trajectories of LLM agents and provide heuristics
for task planning. Specifically, our approach involves employing one LLM-based
agent to navigate an environment randomly, generating diverse action
trajectories. Subsequently, a separate LLM is leveraged to assign a task intent
and synthesize a negative response alongside the correct response for each
trajectory. These triplets (task intent, positive response, and negative
response) are then utilized as training data to optimize a reward model capable
of scoring action trajectories. The effectiveness and generalizability of our
framework are demonstrated through evaluations conducted on different agent
benchmarks. In conclusion, our proposed framework represents a significant
advancement in enhancing LLM agents' decision-making capabilities. By
automating the learning of reward models, we overcome the challenges of data
scarcity and API limitations, potentially revolutionizing the application of
LLMs in complex and interactive environments. This research paves the way for
more sophisticated AI agents capable of tackling a wide range of real-world
problems requiring multi-step decision-making. | 2 | 67b657d7a267b1a747a7ff1a | null | null |
|
2025-02-19T13:39:32.672000 | YOLOv12: Attention-Centric Real-Time Object Detectors | 2 | {
"_id": "5f1158120c833276f61f1a84",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1608042047613-5f1158120c833276f61f1a84.jpeg",
"followerCount": 777,
"fullname": "Niels Rogge",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "nielsr",
"type": "user"
} | false | null | 2502.12524 | [
{
"_id": "67b608ca13df25808fbc22ae",
"hidden": false,
"name": "Yunjie Tian",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b608ca13df25808fbc22af",
"hidden": false,
"name": "Qixiang Ye",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b608ca13df25808fbc22b0",
"hidden": false,
"name": "David Doermann",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T04:20:14 | YOLOv12: Attention-Centric Real-Time Object Detectors | Enhancing the network architecture of the YOLO framework has been crucial for
a long time, but has focused on CNN-based improvements despite the proven
superiority of attention mechanisms in modeling capabilities. This is because
attention-based models cannot match the speed of CNN-based models. This paper
proposes an attention-centric YOLO framework, namely YOLOv12, that matches the
speed of previous CNN-based ones while harnessing the performance benefits of
attention mechanisms. YOLOv12 surpasses all popular real-time object detectors
in accuracy with competitive speed. For example, YOLOv12-N achieves 40.6% mAP
with an inference latency of 1.64 ms on a T4 GPU, outperforming advanced
YOLOv10-N / YOLOv11-N by 2.1%/1.2% mAP with a comparable speed. This advantage
extends to other model scales. YOLOv12 also surpasses end-to-end real-time
detectors that improve DETR, such as RT-DETR / RT-DETRv2: YOLOv12-S beats
RT-DETR-R18 / RT-DETRv2-R18 while running 42% faster, using only 36% of the
computation and 45% of the parameters. More comparisons are shown in Figure 1. | 10 | 67b608cb13df25808fbc2308 | null | null |
|
2025-02-19T10:33:08.946000 | Harnessing Vision Models for Time Series Analysis: A Survey | 2 | {
"_id": "67b5efbe38c175486e2869b9",
"avatarUrl": "/avatars/64a698259033bb8ac324e57c557a9aa9.svg",
"followerCount": null,
"fullname": "Jingchao Ni",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "nijingchao",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/67b5efbe38c175486e2869b9/iBIxlNXQX2KDabTTeqWL0.png",
"https://cdn-uploads.huggingface.co/production/uploads/67b5efbe38c175486e2869b9/cGTQawzFrVI21iLfRjpFt.png",
"https://cdn-uploads.huggingface.co/production/uploads/67b5efbe38c175486e2869b9/j-lNPZ3OqCUHj6vhnQS5v.png"
]
| 2502.08869 | [
{
"_id": "67b5f3e30e7fed1190f29f80",
"hidden": false,
"name": "Jingchao Ni",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T15:14:31.563Z",
"user": {
"_id": "67b5efbe38c175486e2869b9",
"avatarUrl": "/avatars/64a698259033bb8ac324e57c557a9aa9.svg",
"fullname": "Jingchao Ni",
"isPro": false,
"type": "user",
"user": "nijingchao"
}
},
{
"_id": "67b5f3e30e7fed1190f29f81",
"hidden": false,
"name": "Ziming Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5f3e30e7fed1190f29f82",
"hidden": false,
"name": "ChengAo Shen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:36:57.535Z",
"user": {
"_id": "66c2a909a9425c872d5213f4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/66c2a909a9425c872d5213f4/u-iodlEs2bGzbLwv-cqKx.jpeg",
"fullname": "ChengAo Shen",
"isPro": false,
"type": "user",
"user": "ChengAoShen"
}
},
{
"_id": "67b5f3e30e7fed1190f29f83",
"hidden": false,
"name": "Hanghang Tong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5f3e30e7fed1190f29f84",
"hidden": false,
"name": "Dongjin Song",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5f3e30e7fed1190f29f85",
"hidden": false,
"name": "Wei Cheng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5f3e30e7fed1190f29f86",
"hidden": false,
"name": "Dongsheng Luo",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5f3e30e7fed1190f29f87",
"hidden": false,
"name": "Haifeng Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-13T00:42:11 | Harnessing Vision Models for Time Series Analysis: A Survey | Time series analysis has witnessed the inspiring development from traditional
autoregressive models, deep learning models, to recent Transformers and Large
Language Models (LLMs). Efforts in leveraging vision models for time series
analysis have also been made along the way but are less visible to the
community due to the predominant research on sequence modeling in this domain.
However, the discrepancy between continuous time series and the discrete token
space of LLMs, and the challenges in explicitly modeling the correlations of
variates in multivariate time series have shifted some research attentions to
the equally successful Large Vision Models (LVMs) and Vision Language Models
(VLMs). To fill the blank in the existing literature, this survey discusses the
advantages of vision models over LLMs in time series analysis. It provides a
comprehensive and in-depth overview of the existing methods, with dual views of
detailed taxonomy that answer the key research questions including how to
encode time series as images and how to model the imaged time series for
various tasks. Additionally, we address the challenges in the pre- and
post-processing steps involved in this framework and outline future directions
to further advance time series analysis with vision models. | 2 | 67b5f3e30e7fed1190f29fb7 | null | null |
|
2025-02-19T08:03:59.885000 | Flow-of-Options: Diversified and Improved LLM Reasoning by Thinking Through Options | 2 | {
"_id": "643837ef581e6bf0fa9c72f8",
"avatarUrl": "/avatars/5b95d2509d1c7640d77a3405ebd53eaf.svg",
"followerCount": null,
"fullname": "Lakshmi Nair",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "lnair",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/643837ef581e6bf0fa9c72f8/HhevzVLx8wGDy7sD0zSAj.png"
]
| 2502.12929 | [
{
"_id": "67b546dc2b2ec6908f00c771",
"hidden": false,
"name": "Lakshmi Nair",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T14:37:39.033Z",
"user": {
"_id": "643837ef581e6bf0fa9c72f8",
"avatarUrl": "/avatars/5b95d2509d1c7640d77a3405ebd53eaf.svg",
"fullname": "Lakshmi Nair",
"isPro": false,
"type": "user",
"user": "lnair"
}
},
{
"_id": "67b546dc2b2ec6908f00c772",
"hidden": false,
"name": "Ian Trase",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b546dc2b2ec6908f00c773",
"hidden": false,
"name": "Mark Kim",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T15:11:46 | Flow-of-Options: Diversified and Improved LLM Reasoning by Thinking
Through Options | We present a novel reasoning approach called Flow-of-Options (FoO), designed
to address intrinsic biases in Large Language Models (LLMs). FoO enables LLMs
to systematically explore a diverse range of possibilities in their reasoning,
as demonstrated by an FoO-based agentic system for autonomously solving Machine
Learning tasks (AutoML). Our framework outperforms state-of-the-art baselines,
achieving improvements of 38.2% - 69.2% on standard data science tasks, and
37.4% - 47.9% on therapeutic chemistry tasks. With an overall operation cost
under $1 per task, our framework is well-suited for cost-sensitive
applications. Beyond classification and regression, we illustrate the broader
applicability of our FoO-based agentic system to tasks such as reinforcement
learning and image generation. Our framework presents significant advancements
compared to current state-of-the-art agentic systems for AutoML, due to the
benefits of FoO in enforcing diversity in LLM solutions through compressed,
explainable representations that also support long-term memory when combined
with case-based reasoning. | 7 | 67b546dd2b2ec6908f00c7f6 | null | null |
|
2025-02-19T07:53:04.918000 | Text2World: Benchmarking Large Language Models for Symbolic World Model Generation | 2 | {
"_id": "6237df4a5ab9df625fb70c1a",
"avatarUrl": "/avatars/c5d1a52895cb6515f28019a8e7e3e855.svg",
"followerCount": 1,
"fullname": "Mengkang Hu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "MengkangHu",
"type": "user"
} | true | null | 2502.13092 | [
{
"_id": "67b5473109afe1f3029835cb",
"hidden": false,
"name": "Mengkang Hu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:15.592Z",
"user": {
"_id": "6237df4a5ab9df625fb70c1a",
"avatarUrl": "/avatars/c5d1a52895cb6515f28019a8e7e3e855.svg",
"fullname": "Mengkang Hu",
"isPro": false,
"type": "user",
"user": "MengkangHu"
}
},
{
"_id": "67b5473109afe1f3029835cc",
"hidden": false,
"name": "Tianxing Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5473109afe1f3029835cd",
"hidden": false,
"name": "Yude Zou",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:22.709Z",
"user": {
"_id": "67b4123c9a692ad23daa8a73",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/67b4123c9a692ad23daa8a73/vhkxv0NkM2socQ2tySzqc.jpeg",
"fullname": "Yude Zou",
"isPro": false,
"type": "user",
"user": "xdzouyd"
}
},
{
"_id": "67b5473109afe1f3029835ce",
"hidden": false,
"name": "Yuheng Lei",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5473109afe1f3029835cf",
"hidden": false,
"name": "Qiguang Chen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:12.095Z",
"user": {
"_id": "636f526a6cd69d9a36ff2b53",
"avatarUrl": "/avatars/8f2271a193fcac609d9be270552b5afa.svg",
"fullname": "Qiguang Chen",
"isPro": false,
"type": "user",
"user": "LightChen2333"
}
},
{
"_id": "67b5473109afe1f3029835d0",
"hidden": false,
"name": "Ming Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5473109afe1f3029835d1",
"hidden": false,
"name": "Hongyuan Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5473109afe1f3029835d2",
"hidden": false,
"name": "Wenqi Shao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5473109afe1f3029835d3",
"hidden": false,
"name": "Ping Luo",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T17:59:48 | Text2World: Benchmarking Large Language Models for Symbolic World Model
Generation | Recently, there has been growing interest in leveraging large language models
(LLMs) to generate symbolic world models from textual descriptions. Although
LLMs have been extensively explored in the context of world modeling, prior
studies encountered several challenges, including evaluation randomness,
dependence on indirect metrics, and a limited domain scope. To address these
limitations, we introduce a novel benchmark, Text2World, based on planning
domain definition language (PDDL), featuring hundreds of diverse domains and
employing multi-criteria, execution-based metrics for a more robust evaluation.
We benchmark current LLMs using Text2World and find that reasoning models
trained with large-scale reinforcement learning outperform others. However,
even the best-performing model still demonstrates limited capabilities in world
modeling. Building on these insights, we examine several promising strategies
to enhance the world modeling capabilities of LLMs, including test-time
scaling, agent training, and more. We hope that Text2World can serve as a
crucial resource, laying the groundwork for future research in leveraging LLMs
as world models. The project page is available at
https://text-to-world.github.io/. | 12 | 67b5473209afe1f302983600 | null | null |
|
2025-02-19T06:51:04.672000 | Atom of Thoughts for Markov LLM Test-Time Scaling | 3 | {
"_id": "6402e8fb06c715b93407442d",
"avatarUrl": "/avatars/12b67f0632be5a53b56d8a68586a7f98.svg",
"followerCount": 2,
"fullname": "Fengwei Teng",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "leavendough",
"type": "user"
} | true | null | 2502.12018 | [
{
"_id": "67b5c4ed85107d20148ae710",
"hidden": false,
"name": "Fengwei Teng",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T11:49:34.612Z",
"user": {
"_id": "6402e8fb06c715b93407442d",
"avatarUrl": "/avatars/12b67f0632be5a53b56d8a68586a7f98.svg",
"fullname": "Fengwei Teng",
"isPro": false,
"type": "user",
"user": "leavendough"
}
},
{
"_id": "67b5c4ed85107d20148ae711",
"hidden": false,
"name": "Zhaoyang Yu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:43:14.422Z",
"user": {
"_id": "640dc84b474aa6f89554d518",
"avatarUrl": "/avatars/64f47f76d97c5e91b7ab8380bcada61c.svg",
"fullname": "Zhaoyang Yu",
"isPro": false,
"type": "user",
"user": "MoshiQAQ"
}
},
{
"_id": "67b5c4ed85107d20148ae712",
"hidden": false,
"name": "Quan Shi",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5c4ed85107d20148ae713",
"hidden": false,
"name": "Jiayi Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:43:54.773Z",
"user": {
"_id": "66071c8b013a0afdf40fbfd1",
"avatarUrl": "/avatars/683e4ba5991059110631759a5975eacc.svg",
"fullname": "JiaYi Zhang",
"isPro": false,
"type": "user",
"user": "Bbedd"
}
},
{
"_id": "67b5c4ed85107d20148ae714",
"hidden": false,
"name": "Chenglin Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5c4ed85107d20148ae715",
"hidden": false,
"name": "Yuyu Luo",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T16:52:42 | Atom of Thoughts for Markov LLM Test-Time Scaling | Large Language Models (LLMs) achieve superior performance through
training-time scaling, and test-time scaling further enhances their
capabilities by conducting effective reasoning during inference. However, as
the scale of reasoning increases, existing test-time scaling methods suffer
from accumulated historical information, which not only wastes computational
resources but also interferes with effective reasoning. To address this issue,
we observe that complex reasoning progress is often achieved by solving a
sequence of independent subquestions, each being self-contained and verifiable.
These subquestions are essentially atomic questions, relying primarily on their
current state rather than accumulated history, similar to the memoryless
transitions in a Markov process. Based on this observation, we propose Atom of
Thoughts (AoT), where each state transition in the reasoning process consists
of decomposing the current question into a dependency-based directed acyclic
graph and contracting its subquestions, forming a new atomic question state.
This iterative decomposition-contraction process continues until reaching
directly solvable atomic questions, naturally realizing Markov transitions
between question states. Furthermore, these atomic questions can be seamlessly
integrated into existing test-time scaling methods, enabling AoT to serve as a
plug-in enhancement for improving reasoning capabilities. Experiments across
six benchmarks demonstrate the effectiveness of AoT both as a standalone
framework and a plug-in enhancement. Notably, on HotpotQA, when applied to
gpt-4o-mini, AoT achieves an 80.6% F1 score, surpassing o3-mini by 3.4% and
DeepSeek-R1 by 10.6%. The code will be available at
https://github.com/qixucen/atom. | 15 | 67b5c4ee85107d20148ae73d | null | null |
|
2025-02-19T06:13:51.101000 | Eager Updates For Overlapped Communication and Computation in DiLoCo | 2 | {
"_id": "622792366303bf1dc304f49f",
"avatarUrl": "/avatars/975c1cc3eb2f97cf8e848162056d5bea.svg",
"followerCount": 4,
"fullname": "Arthur Douillard",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "ArthurDouillard",
"type": "user"
} | true | null | 2502.12996 | [
{
"_id": "67b5bcd091132877cf330179",
"hidden": false,
"name": "Satyen Kale",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5bcd091132877cf33017a",
"hidden": false,
"name": "Arthur Douillard",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:42:51.668Z",
"user": {
"_id": "622792366303bf1dc304f49f",
"avatarUrl": "/avatars/975c1cc3eb2f97cf8e848162056d5bea.svg",
"fullname": "Arthur Douillard",
"isPro": false,
"type": "user",
"user": "ArthurDouillard"
}
},
{
"_id": "67b5bcd091132877cf33017b",
"hidden": false,
"name": "Yanislav Donchev",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:42:57.488Z",
"user": {
"_id": "61796df92357d28cdf952511",
"avatarUrl": "/avatars/01821f364cf78c95aed1986d91d40610.svg",
"fullname": "Yanislav Donchev",
"isPro": false,
"type": "user",
"user": "yannidd"
}
}
]
| 2025-02-18T16:16:14 | Eager Updates For Overlapped Communication and Computation in DiLoCo | Distributed optimization methods such as DiLoCo have been shown to be
effective in training very large models across multiple distributed workers,
such as datacenters. These methods split updates into two parts: an inner
optimization phase, where the workers independently execute multiple
optimization steps on their own local data, and an outer optimization step,
where the inner updates are synchronized. While such approaches require orders
of magnitude less communication than standard data-parallel training, in
settings where the workers are datacenters, even the limited communication
requirements of these approaches can still cause significant slow downs due to
the blocking necessary at each outer optimization step. In this paper, we
investigate techniques to mitigate this issue by overlapping communication with
computation in a manner that allows the outer optimization step to fully
overlap with the inner optimization phase. We show that a particular variant,
dubbed eager updates, provides competitive performance with standard DiLoCo in
settings with low bandwidth between workers. | 7 | 67b5bcd191132877cf3301aa | null | null |
|
2025-02-19T04:54:27.788000 | FinMTEB: Finance Massive Text Embedding Benchmark | 2 | {
"_id": "647d834618274bce03013cc2",
"avatarUrl": "/avatars/a95c7df96dc4fb6a96193f6dd5068227.svg",
"followerCount": 2,
"fullname": "yixuan",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "yixuantt",
"type": "user"
} | true | null | 2502.10990 | [
{
"_id": "67b3ee6c1e80a69e79c3155a",
"hidden": false,
"name": "Yixuan Tang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:50.969Z",
"user": {
"_id": "647d834618274bce03013cc2",
"avatarUrl": "/avatars/a95c7df96dc4fb6a96193f6dd5068227.svg",
"fullname": "yixuan",
"isPro": true,
"type": "user",
"user": "yixuantt"
}
},
{
"_id": "67b3ee6c1e80a69e79c3155b",
"hidden": false,
"name": "Yi Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-16T04:23:52 | FinMTEB: Finance Massive Text Embedding Benchmark | Embedding models play a crucial role in representing and retrieving
information across various NLP applications. Recent advances in large language
models (LLMs) have further enhanced the performance of embedding models. While
these models are often benchmarked on general-purpose datasets, real-world
applications demand domain-specific evaluation. In this work, we introduce the
Finance Massive Text Embedding Benchmark (FinMTEB), a specialized counterpart
to MTEB designed for the financial domain. FinMTEB comprises 64 financial
domain-specific embedding datasets across 7 tasks that cover diverse textual
types in both Chinese and English, such as financial news articles, corporate
annual reports, ESG reports, regulatory filings, and earnings call transcripts.
We also develop a finance-adapted model, FinPersona-E5, using a persona-based
data synthetic method to cover diverse financial embedding tasks for training.
Through extensive evaluation of 15 embedding models, including FinPersona-E5,
we show three key findings: (1) performance on general-purpose benchmarks shows
limited correlation with financial domain tasks; (2) domain-adapted models
consistently outperform their general-purpose counterparts; and (3)
surprisingly, a simple Bag-of-Words (BoW) approach outperforms sophisticated
dense embeddings in financial Semantic Textual Similarity (STS) tasks,
underscoring current limitations in dense embedding techniques. Our work
establishes a robust evaluation framework for financial NLP applications and
provides crucial insights for developing domain-specific embedding models. | 3 | 67b3ee6d1e80a69e79c3158f | null | null |
|
2025-02-19T04:43:42.973000 | Cramming 1568 Tokens into a Single Vector and Back Again: Exploring the Limits of Embedding Space Capacity | 4 | {
"_id": "639c6e978a34ed9a404c6a7b",
"avatarUrl": "/avatars/c98ca8c9f9ed8509c2f1bb6aa994fd57.svg",
"followerCount": 7,
"fullname": "MIKHAIL BURTSEV",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "mbur",
"type": "user"
} | true | null | 2502.13063 | [
{
"_id": "67b5a7896f72266cb765e744",
"hidden": false,
"name": "Yuri Kuratov",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-19T09:42:34.422Z",
"user": {
"_id": "618b9540682ec1c38327e586",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/618b9540682ec1c38327e586/v_ZBkfh8O9Zh6C2YQpuBX.jpeg",
"fullname": "Yury Kuratov",
"isPro": false,
"type": "user",
"user": "yurakuratov"
}
},
{
"_id": "67b5a7896f72266cb765e745",
"hidden": false,
"name": "Mikhail Arkhipov",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5a7896f72266cb765e746",
"hidden": false,
"name": "Aydar Bulatov",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T14:43:06.400Z",
"user": {
"_id": "64c8b321cb2f1bf0e7c0f54b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c8b321cb2f1bf0e7c0f54b/JflXxMVnG9I0IB5YNyhXF.jpeg",
"fullname": "Aydar Bulatov",
"isPro": false,
"type": "user",
"user": "booydar"
}
},
{
"_id": "67b5a7896f72266cb765e747",
"hidden": false,
"name": "Mikhail Burtsev",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:56:59.080Z",
"user": {
"_id": "639c6e978a34ed9a404c6a7b",
"avatarUrl": "/avatars/c98ca8c9f9ed8509c2f1bb6aa994fd57.svg",
"fullname": "MIKHAIL BURTSEV",
"isPro": false,
"type": "user",
"user": "mbur"
}
}
]
| 2025-02-18T17:08:45 | Cramming 1568 Tokens into a Single Vector and Back Again: Exploring the
Limits of Embedding Space Capacity | A range of recent works addresses the problem of compression of sequence of
tokens into a shorter sequence of real-valued vectors to be used as inputs
instead of token embeddings or key-value cache. These approaches allow to
reduce the amount of compute in existing language models. Despite relying on
powerful models as encoders, the maximum attainable lossless compression ratio
is typically not higher than x10. This fact is highly intriguing because, in
theory, the maximum information capacity of large real-valued vectors is far
beyond the presented rates even for 16-bit precision and a modest vector size.
In this work, we explore the limits of compression by replacing the encoder
with a per-sample optimization procedure. We show that vectors with compression
ratios up to x1500 exist, which highlights two orders of magnitude gap between
existing and practically attainable solutions. Furthermore, we empirically show
that the compression limits are determined not by the length of the input but
by the amount of uncertainty to be reduced, namely, the cross-entropy loss on
this sequence without any conditioning. The obtained limits highlight the
substantial gap between the theoretical capacity of input embeddings and their
practical utilization, suggesting significant room for optimization in model
design. | 64 | 67b5a78a6f72266cb765e779 | null | null |
|
2025-02-19T03:03:51.930000 | You Do Not Fully Utilize Transformer's Representation Capacity | 3 | {
"_id": "63ed5676684767daecac6f8a",
"avatarUrl": "/avatars/d0e4a715f9c3fb6d74c183bab751ec35.svg",
"followerCount": 4,
"fullname": "Yaroslav Aksenov",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "yaraksen",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/63ed5676684767daecac6f8a/tZDsnW0gjHoYCpbZ-wwJi.png"
]
| 2502.09245 | [
{
"_id": "67b57a993d4f319f1fa9424b",
"hidden": false,
"name": "Gleb Gerasimov",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T10:10:30.547Z",
"user": {
"_id": "65db0871ab2f64915ce05e73",
"avatarUrl": "/avatars/77e03f493196c5413cd2a02270e93660.svg",
"fullname": "Gleb Gerasimov",
"isPro": false,
"type": "user",
"user": "gudleifrr"
}
},
{
"_id": "67b57a993d4f319f1fa9424c",
"hidden": false,
"name": "Yaroslav Aksenov",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:00:41.123Z",
"user": {
"_id": "63ed5676684767daecac6f8a",
"avatarUrl": "/avatars/d0e4a715f9c3fb6d74c183bab751ec35.svg",
"fullname": "Yaroslav Aksenov",
"isPro": false,
"type": "user",
"user": "yaraksen"
}
},
{
"_id": "67b57a993d4f319f1fa9424d",
"hidden": false,
"name": "Nikita Balagansky",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:33:26.858Z",
"user": {
"_id": "60b364e7f88532cd79eaff7b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1654185363389-60b364e7f88532cd79eaff7b.jpeg",
"fullname": "Nikita Balagansky",
"isPro": false,
"type": "user",
"user": "elephantmipt"
}
},
{
"_id": "67b57a993d4f319f1fa9424e",
"hidden": false,
"name": "Viacheslav Sinii",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T11:12:28.927Z",
"user": {
"_id": "6416272d986557e8cac64ece",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6416272d986557e8cac64ece/s3CLjNN_pGj-vJDcENFD2.jpeg",
"fullname": "Viacheslav",
"isPro": false,
"type": "user",
"user": "ummagumm-a"
}
},
{
"_id": "67b57a993d4f319f1fa9424f",
"hidden": false,
"name": "Daniil Gavrilov",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:00:43.143Z",
"user": {
"_id": "62a9c8edc19f92ae443ab37f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1669110208492-62a9c8edc19f92ae443ab37f.png",
"fullname": "Daniil Gavrilov",
"isPro": false,
"type": "user",
"user": "kefirski"
}
}
]
| 2025-02-13T12:00:50 | You Do Not Fully Utilize Transformer's Representation Capacity | In contrast to RNNs, which compress previous tokens into a single hidden
state, Transformers can attend to all previous tokens directly. However,
standard Transformers only use representations from the immediately preceding
layer. In this paper, we show that this design choice causes representation
collapse and leads to suboptimal performance. To address this issue, we
introduce Layer-Integrated Memory (LIMe), a simple yet powerful approach that
preserves the model's overall memory footprint while expanding its
representational capacity by allowing access to hidden states from earlier
layers. Through extensive experiments across various architectures and
different lookup mechanisms, we demonstrate consistent performance improvements
on a wide range of tasks. Moreover, our analysis of the learned representation
dynamics and our exploration of depthwise circuits reveal how LIMe integrates
information across layers, pointing to promising directions for future
research. | 34 | 67b57a9a3d4f319f1fa94274 | null | null |
|
2025-02-19T02:56:09.510000 | Injecting Domain-Specific Knowledge into Large Language Models: A Comprehensive Survey | 2 | {
"_id": "65407ba7a38390065750233f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/65407ba7a38390065750233f/1_IPMZbk-S9u2t18PQgMp.jpeg",
"followerCount": 1,
"fullname": "Zirui Song",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Ziruibest",
"type": "user"
} | true | null | 2502.10708 | [
{
"_id": "67b58e32e972a2806a9a0451",
"hidden": false,
"name": "Zirui Song",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:00:38.943Z",
"user": {
"_id": "65407ba7a38390065750233f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/65407ba7a38390065750233f/1_IPMZbk-S9u2t18PQgMp.jpeg",
"fullname": "Zirui Song",
"isPro": false,
"type": "user",
"user": "Ziruibest"
}
},
{
"_id": "67b58e32e972a2806a9a0452",
"hidden": false,
"name": "Bin Yan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58e32e972a2806a9a0453",
"hidden": false,
"name": "Yuhan Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:53.429Z",
"user": {
"_id": "627a124ffe55fa0f8ce0eaf7",
"avatarUrl": "/avatars/5dd30cf87e60f257ecfa7d2f871d3f33.svg",
"fullname": "Serendipity",
"isPro": false,
"type": "user",
"user": "Yuhan"
}
},
{
"_id": "67b58e32e972a2806a9a0454",
"hidden": false,
"name": "Miao Fang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58e32e972a2806a9a0455",
"hidden": false,
"name": "Mingzhe Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58e32e972a2806a9a0456",
"hidden": false,
"name": "Rui Yan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58e32e972a2806a9a0457",
"hidden": false,
"name": "Xiuying Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-15T07:43:43 | Injecting Domain-Specific Knowledge into Large Language Models: A
Comprehensive Survey | Large Language Models (LLMs) have demonstrated remarkable success in various
tasks such as natural language understanding, text summarization, and machine
translation. However, their general-purpose nature often limits their
effectiveness in domain-specific applications that require specialized
knowledge, such as healthcare, chemistry, or legal analysis. To address this,
researchers have explored diverse methods to enhance LLMs by integrating
domain-specific knowledge. In this survey, we provide a comprehensive overview
of these methods, which we categorize into four key approaches: dynamic
knowledge injection, static knowledge embedding, modular adapters, and prompt
optimization. Each approach offers unique mechanisms to equip LLMs with domain
expertise, balancing trade-offs between flexibility, scalability, and
efficiency. We discuss how these methods enable LLMs to tackle specialized
tasks, compare their advantages and disadvantages, evaluate domain-specific
LLMs against general LLMs, and highlight the challenges and opportunities in
this emerging field. For those interested in delving deeper into this area, we
also summarize the commonly used datasets and benchmarks. To keep researchers
updated on the latest studies, we maintain an open-source at:
https://github.com/abilliyb/Knowledge_Injection_Survey_Papers, dedicated to
documenting research in the field of specialized LLM. | 4 | 67b58e33e972a2806a9a04b8 | null | null |
|
2025-02-19T02:47:33.654000 | Perovskite-LLM: Knowledge-Enhanced Large Language Models for Perovskite Solar Cell Research | 2 | {
"_id": "63024676056ec3a2a8714b24",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1661093436322-noauth.jpeg",
"followerCount": 5,
"fullname": "Xiang Liu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Dominic789654",
"type": "user"
} | true | null | 2502.12669 | [
{
"_id": "67b58c806e53744c2a373351",
"hidden": false,
"name": "Xiang Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:34:03.429Z",
"user": {
"_id": "63024676056ec3a2a8714b24",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1661093436322-noauth.jpeg",
"fullname": "Xiang Liu",
"isPro": false,
"type": "user",
"user": "Dominic789654"
}
},
{
"_id": "67b58c806e53744c2a373352",
"hidden": false,
"name": "Penglei Sun",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:34:15.889Z",
"user": {
"_id": "64eded5fdfe0a679d840bc98",
"avatarUrl": "/avatars/4d4c67c13e547a4d296a301e8694e79e.svg",
"fullname": "sunpenglei",
"isPro": false,
"type": "user",
"user": "sunpenglei"
}
},
{
"_id": "67b58c806e53744c2a373353",
"hidden": false,
"name": "Shuyan Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58c806e53744c2a373354",
"hidden": false,
"name": "Longhan Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58c806e53744c2a373355",
"hidden": false,
"name": "Peijie Dong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58c806e53744c2a373356",
"hidden": false,
"name": "Huajie You",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T18:16:35.805Z",
"user": {
"_id": "660937dff7373477d86501b8",
"avatarUrl": "/avatars/1edbe8c92b41bf496b962f71b306ea7b.svg",
"fullname": "Huajie You",
"isPro": false,
"type": "user",
"user": "FrankYOU"
}
},
{
"_id": "67b58c806e53744c2a373357",
"hidden": false,
"name": "Yongqi Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:35:12.059Z",
"user": {
"_id": "64473221dcbe1333b64b2db2",
"avatarUrl": "/avatars/5e4495d3581ad3e6ea3c47650f20b993.svg",
"fullname": "yongqi zhang",
"isPro": false,
"type": "user",
"user": "yongqi2023"
}
},
{
"_id": "67b58c806e53744c2a373358",
"hidden": false,
"name": "Chang Yan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b58c806e53744c2a373359",
"hidden": false,
"name": "Xiaowen Chu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:35:20.611Z",
"user": {
"_id": "6676935fcd0b89a0115174b0",
"avatarUrl": "/avatars/4caca1b672d29e787814f9a30bf20bcc.svg",
"fullname": "Xiaowen Chu",
"isPro": false,
"type": "user",
"user": "wenxinsiju"
}
},
{
"_id": "67b58c806e53744c2a37335a",
"hidden": false,
"name": "Tong-yi Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T09:19:24 | Perovskite-LLM: Knowledge-Enhanced Large Language Models for Perovskite
Solar Cell Research | The rapid advancement of perovskite solar cells (PSCs) has led to an
exponential growth in research publications, creating an urgent need for
efficient knowledge management and reasoning systems in this domain. We present
a comprehensive knowledge-enhanced system for PSCs that integrates three key
components. First, we develop Perovskite-KG, a domain-specific knowledge graph
constructed from 1,517 research papers, containing 23,789 entities and 22,272
relationships. Second, we create two complementary datasets: Perovskite-Chat,
comprising 55,101 high-quality question-answer pairs generated through a novel
multi-agent framework, and Perovskite-Reasoning, containing 2,217 carefully
curated materials science problems. Third, we introduce two specialized large
language models: Perovskite-Chat-LLM for domain-specific knowledge assistance
and Perovskite-Reasoning-LLM for scientific reasoning tasks. Experimental
results demonstrate that our system significantly outperforms existing models
in both domain-specific knowledge retrieval and scientific reasoning tasks,
providing researchers with effective tools for literature review, experimental
design, and complex problem-solving in PSC research. | 2 | 67b58c826e53744c2a3733c2 | null | null |
|
2025-02-19T02:27:36.940000 | OctoTools: An Agentic Framework with Extensible Tools for Complex Reasoning | 3 | {
"_id": "60f5f68fa7fd83d025749234",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/60f5f68fa7fd83d025749234/gCeJAZfzaANAcEvI6v5-P.jpeg",
"followerCount": 8,
"fullname": "Pan Lu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "lupantech",
"type": "user"
} | true | null | 2502.11271 | [
{
"_id": "67b4322c217ec18a40587bec",
"hidden": false,
"name": "Pan Lu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:43.677Z",
"user": {
"_id": "60f5f68fa7fd83d025749234",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/60f5f68fa7fd83d025749234/gCeJAZfzaANAcEvI6v5-P.jpeg",
"fullname": "Pan Lu",
"isPro": false,
"type": "user",
"user": "lupantech"
}
},
{
"_id": "67b4322c217ec18a40587bed",
"hidden": false,
"name": "Bowen Chen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:37.912Z",
"user": {
"_id": "64b988b965be45c7766b105a",
"avatarUrl": "/avatars/ef1944c79cfe851d079dcba0603526ab.svg",
"fullname": "Bowen Chen",
"isPro": false,
"type": "user",
"user": "bowen118"
}
},
{
"_id": "67b4322c217ec18a40587bee",
"hidden": false,
"name": "Sheng Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:35.169Z",
"user": {
"_id": "653585666141b3927a083b4f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/653585666141b3927a083b4f/EpMnW_gKz9FV3OvogvUzi.jpeg",
"fullname": "Sheng Liu",
"isPro": false,
"type": "user",
"user": "shengliu66"
}
},
{
"_id": "67b4322c217ec18a40587bef",
"hidden": false,
"name": "Rahul Thapa",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4322c217ec18a40587bf0",
"hidden": false,
"name": "Joseph Boen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:39:08.417Z",
"user": {
"_id": "66846f30f53c83742c117277",
"avatarUrl": "/avatars/31a2434297a6c194d57f2a0234f1ca2c.svg",
"fullname": "Joseph Boen",
"isPro": false,
"type": "user",
"user": "tboen1"
}
},
{
"_id": "67b4322c217ec18a40587bf1",
"hidden": false,
"name": "James Zou",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-16T21:18:47 | OctoTools: An Agentic Framework with Extensible Tools for Complex
Reasoning | Solving complex reasoning tasks may involve visual understanding, domain
knowledge retrieval, numerical calculation, and multi-step reasoning. Existing
methods augment large language models (LLMs) with external tools but are
restricted to specialized domains, limited tool types, or require additional
training data. In this paper, we introduce OctoTools, a training-free,
user-friendly, and easily extensible open-source agentic framework designed to
tackle complex reasoning across diverse domains. OctoTools introduces
standardized tool cards to encapsulate tool functionality, a planner for both
high-level and low-level planning, and an executor to carry out tool usage. We
validate OctoTools' generality across 16 diverse tasks (including MathVista,
MMLU-Pro, MedQA, and GAIA-Text), achieving substantial average accuracy gains
of 9.3% over GPT-4o. Furthermore, OctoTools outperforms AutoGen, GPT-Functions
and LangChain by up to 10.6% when given the same set of tools. Through
comprehensive analysis and ablations, OctoTools demonstrates advantages in task
planning, effective tool usage, and multi-step problem solving. | 16 | 67b4322d217ec18a40587c27 | null | null |
|
2025-02-19T01:24:26.365000 | Pre-training Auto-regressive Robotic Models with 4D Representations | 2 | {
"_id": "667c5764186b27ef806636d3",
"avatarUrl": "/avatars/5c08f0109bc0e350624112c0aff544f6.svg",
"followerCount": null,
"fullname": "Roei Herzig",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "roeiherz",
"type": "user"
} | true | null | 2502.13142 | [
{
"_id": "67b5790132be608036ee94e5",
"hidden": false,
"name": "Dantong Niu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:12:28.457Z",
"user": {
"_id": "65c3fdf79d062be813813e45",
"avatarUrl": "/avatars/52528a61abe5bbbef4a4a431944973cd.svg",
"fullname": "Dantong Niu",
"isPro": false,
"type": "user",
"user": "NdtSoCool"
}
},
{
"_id": "67b5790132be608036ee94e6",
"hidden": false,
"name": "Yuvan Sharma",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:12:35.531Z",
"user": {
"_id": "65406e82deee4716f1c29271",
"avatarUrl": "/avatars/25331a773f8125f9ad1c3d6ac3375586.svg",
"fullname": "Yuvan Sharma",
"isPro": false,
"type": "user",
"user": "yuvansharma"
}
},
{
"_id": "67b5790132be608036ee94e7",
"hidden": false,
"name": "Haoru Xue",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5790132be608036ee94e8",
"hidden": false,
"name": "Giscard Biamby",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:12:49.219Z",
"user": {
"_id": "650bd36a7c99ca283e58e973",
"avatarUrl": "/avatars/606d24b2dac190ebcbb4b2a2e4671380.svg",
"fullname": "Giscard Biamby",
"isPro": false,
"type": "user",
"user": "gbiamby"
}
},
{
"_id": "67b5790132be608036ee94e9",
"hidden": false,
"name": "Junyi Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:02.956Z",
"user": {
"_id": "62f0ecd2700bdc19558360de",
"avatarUrl": "/avatars/5325b4b763f30c41f30e3aec0d2b59fa.svg",
"fullname": "Junyi Zhang",
"isPro": false,
"type": "user",
"user": "Junyi42"
}
},
{
"_id": "67b5790132be608036ee94ea",
"hidden": false,
"name": "Ziteng Ji",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:13:13.907Z",
"user": {
"_id": "66a09aec369dd38cf2113070",
"avatarUrl": "/avatars/cc13bdd3dc1271d33b083b61e12f1a05.svg",
"fullname": "Ziteng Ji",
"isPro": false,
"type": "user",
"user": "zitengj0618"
}
},
{
"_id": "67b5790132be608036ee94eb",
"hidden": false,
"name": "Trevor Darrell",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:13:20.379Z",
"user": {
"_id": "64cbdf02f103036e23d1c7f3",
"avatarUrl": "/avatars/496069463900dea20929b57381182d39.svg",
"fullname": "Trevor Darrell",
"isPro": false,
"type": "user",
"user": "trevordarrell"
}
},
{
"_id": "67b5790132be608036ee94ec",
"hidden": false,
"name": "Roei Herzig",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:13:26.134Z",
"user": {
"_id": "667c5764186b27ef806636d3",
"avatarUrl": "/avatars/5c08f0109bc0e350624112c0aff544f6.svg",
"fullname": "Roei Herzig",
"isPro": false,
"type": "user",
"user": "roeiherz"
}
}
]
| 2025-02-18T18:59:01 | Pre-training Auto-regressive Robotic Models with 4D Representations | Foundation models pre-trained on massive unlabeled datasets have
revolutionized natural language and computer vision, exhibiting remarkable
generalization capabilities, thus highlighting the importance of pre-training.
Yet, efforts in robotics have struggled to achieve similar success, limited by
either the need for costly robotic annotations or the lack of representations
that effectively model the physical world. In this paper, we introduce ARM4R,
an Auto-regressive Robotic Model that leverages low-level 4D Representations
learned from human video data to yield a better pre-trained robotic model.
Specifically, we focus on utilizing 3D point tracking representations from
videos derived by lifting 2D representations into 3D space via monocular depth
estimation across time. These 4D representations maintain a shared geometric
structure between the points and robot state representations up to a linear
transformation, enabling efficient transfer learning from human video data to
low-level robotic control. Our experiments show that ARM4R can transfer
efficiently from human video data to robotics and consistently improves
performance on tasks across various robot environments and configurations. | 4 | 67b5790832be608036ee9638 | null | null |
|
2025-02-19T01:21:54.836000 | PAFT: Prompt-Agnostic Fine-Tuning | 8 | {
"_id": "65ed3051492a7f35db21fea2",
"avatarUrl": "/avatars/4fc0ccc21aa88e4e8ff74a6f850570b8.svg",
"followerCount": null,
"fullname": "Chenxing Wei",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "kittttttt",
"type": "user"
} | true | null | 2502.12859 | [
{
"_id": "67b576aa489d68b981e086ad",
"hidden": false,
"name": "Chenxing Wei",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:23:00.016Z",
"user": {
"_id": "65ed3051492a7f35db21fea2",
"avatarUrl": "/avatars/4fc0ccc21aa88e4e8ff74a6f850570b8.svg",
"fullname": "Chenxing Wei",
"isPro": false,
"type": "user",
"user": "kittttttt"
}
},
{
"_id": "67b576aa489d68b981e086ae",
"hidden": false,
"name": "Yao Shu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T14:37:34.696Z",
"user": {
"_id": "66123816d7dfcea8ae55a751",
"avatarUrl": "/avatars/3f24468b63e4babd7d9a0c926ca01b23.svg",
"fullname": "Shu Yao",
"isPro": false,
"type": "user",
"user": "ZCODE0"
}
},
{
"_id": "67b576aa489d68b981e086af",
"hidden": false,
"name": "Mingwen Ou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b576aa489d68b981e086b0",
"hidden": false,
"name": "Ying Tiffany He",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b576aa489d68b981e086b1",
"hidden": false,
"name": "Fei Richard Yu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T13:46:47 | PAFT: Prompt-Agnostic Fine-Tuning | While Large Language Models (LLMs) adapt well to downstream tasks after
fine-tuning, this adaptability often compromises prompt robustness, as even
minor prompt variations can significantly degrade performance. To address this,
we propose Prompt-Agnostic Fine-Tuning(PAFT), a simple yet effective approach
that dynamically adjusts prompts during fine-tuning. This encourages the model
to learn underlying task principles rather than overfitting to specific prompt
formulations. PAFT operates in two stages: First, a diverse set of meaningful,
synthetic candidate prompts is constructed. Second, during fine-tuning, prompts
are randomly sampled from this set to create dynamic training inputs. Extensive
experiments across diverse datasets and LLMs demonstrate that models trained
with PAFT exhibit strong robustness and generalization across a wide range of
prompts, including unseen ones. This enhanced robustness improves both model
performance and inference speed while maintaining training efficiency. Ablation
studies further confirm the effectiveness of PAFT. | 15 | 67b576aa489d68b981e08708 | null | null |
|
2025-02-19T00:22:36.628000 | Soundwave: Less is More for Speech-Text Alignment in LLMs | 2 | {
"_id": "66975b9f8031bf92b428e138",
"avatarUrl": "/avatars/3254281a7bac1c8ddde1d6bc7e518b2f.svg",
"followerCount": null,
"fullname": "Yuhao Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Yoohao",
"type": "user"
} | true | null | 2502.12900 | [
{
"_id": "67b54851b986e35c41e063da",
"hidden": false,
"name": "Yuhao Zhang",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-19T02:56:18.848Z",
"user": {
"_id": "66975b9f8031bf92b428e138",
"avatarUrl": "/avatars/3254281a7bac1c8ddde1d6bc7e518b2f.svg",
"fullname": "Yuhao Zhang",
"isPro": false,
"type": "user",
"user": "Yoohao"
}
},
{
"_id": "67b54851b986e35c41e063db",
"hidden": false,
"name": "Zhiheng Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:05.678Z",
"user": {
"_id": "66597f2cf769c3c443b7cf41",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/WkZBh7hwlD9wVqCEMQtGX.png",
"fullname": "Chihang Lau",
"isPro": true,
"type": "user",
"user": "puccho"
}
},
{
"_id": "67b54851b986e35c41e063dc",
"hidden": false,
"name": "Fan Bu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:42:08.544Z",
"user": {
"_id": "668e7f46c243a12604035758",
"avatarUrl": "/avatars/35bd20032fafb7d7603266cf9a72d1e0.svg",
"fullname": "Fan Bu",
"isPro": false,
"type": "user",
"user": "FanBuCUHK"
}
},
{
"_id": "67b54851b986e35c41e063dd",
"hidden": false,
"name": "Ruiyu Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:42:14.866Z",
"user": {
"_id": "67b587c8882e49771f610b51",
"avatarUrl": "/avatars/aecfb38b44141b8284416fc261692909.svg",
"fullname": "Ruiyu Zhang",
"isPro": false,
"type": "user",
"user": "PhoenixAxis"
}
},
{
"_id": "67b54851b986e35c41e063de",
"hidden": false,
"name": "Benyou Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:42:23.845Z",
"user": {
"_id": "637c6703ca8542a0ba900ccb",
"avatarUrl": "/avatars/288ed63a1efa566c3f01e850c6ba5dd5.svg",
"fullname": "Wang",
"isPro": false,
"type": "user",
"user": "Benyou"
}
},
{
"_id": "67b54851b986e35c41e063df",
"hidden": false,
"name": "Haizhou Li",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T14:36:39 | Soundwave: Less is More for Speech-Text Alignment in LLMs | Existing end-to-end speech large language models (LLMs) usually rely on
large-scale annotated data for training, while data-efficient training has not
been discussed in depth. We focus on two fundamental problems between speech
and text: the representation space gap and sequence length inconsistency. We
propose Soundwave, which utilizes an efficient training strategy and a novel
architecture to address these issues. Results show that Soundwave outperforms
the advanced Qwen2-Audio in speech translation and AIR-Bench speech tasks,
using only one-fiftieth of the training data. Further analysis shows that
Soundwave still retains its intelligence during conversation. The project is
available at https://github.com/FreedomIntelligence/Soundwave. | 76 | 67b54852b986e35c41e06426 | null | null |
|
2025-02-18T23:51:36.910000 | Magma: A Foundation Model for Multimodal AI Agents | 6 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | true | null | 2502.13130 | [
{
"_id": "67b5625fb27eb6046b2ceec5",
"hidden": false,
"name": "Jianwei Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5625fb27eb6046b2ceec6",
"hidden": false,
"name": "Reuben Tan",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:19:01.753Z",
"user": {
"_id": "674774e8eb8fb5ea40877838",
"avatarUrl": "/avatars/7ee8599cb1f7bb4402bc8512faf6ca12.svg",
"fullname": "Reuben Tan",
"isPro": false,
"type": "user",
"user": "tanreuben"
}
},
{
"_id": "67b5625fb27eb6046b2ceec7",
"hidden": false,
"name": "Qianhui Wu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:18:38.510Z",
"user": {
"_id": "63ef330b1e695b35aa484e11",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63ef330b1e695b35aa484e11/bXwpGy0dl8JXeJwJ--ilr.jpeg",
"fullname": "Qianhui WU",
"isPro": false,
"type": "user",
"user": "qianhuiwu"
}
},
{
"_id": "67b5625fb27eb6046b2ceec8",
"hidden": false,
"name": "Ruijie Zheng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:18:32.761Z",
"user": {
"_id": "653b24ef8f8d60f204872f0a",
"avatarUrl": "/avatars/45a55219e8a78be53fd32e96ba460282.svg",
"fullname": "Ruijie Zheng",
"isPro": false,
"type": "user",
"user": "rzheng12"
}
},
{
"_id": "67b5625fb27eb6046b2ceec9",
"hidden": false,
"name": "Baolin Peng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:18:25.610Z",
"user": {
"_id": "61942296d5c2ba6daa290357",
"avatarUrl": "/avatars/594021cc183c4922d48b46f43772a062.svg",
"fullname": "Baolin Peng",
"isPro": false,
"type": "user",
"user": "Baolin"
}
},
{
"_id": "67b5625fb27eb6046b2ceeca",
"hidden": false,
"name": "Yongyuan Liang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:18:00.072Z",
"user": {
"_id": "6646d5819bb34d2b6b7455d3",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/JFH3ZTPvlaVSg4RJJBb6L.jpeg",
"fullname": "Yongyuan Liang",
"isPro": false,
"type": "user",
"user": "cheryyunl"
}
},
{
"_id": "67b5625fb27eb6046b2ceecb",
"hidden": false,
"name": "Yu Gu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:05.283Z",
"user": {
"_id": "645a02972abf6165a3ba5df8",
"avatarUrl": "/avatars/6166e4f213b9020ba99f853cb04f8db0.svg",
"fullname": "Yu Gu",
"isPro": false,
"type": "user",
"user": "aidenygu"
}
},
{
"_id": "67b5625fb27eb6046b2ceecc",
"hidden": false,
"name": "Mu Cai",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5625fb27eb6046b2ceecd",
"hidden": false,
"name": "Seonghyeon Ye",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:17:47.576Z",
"user": {
"_id": "62551f7767f0b85962624047",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1664552038624-62551f7767f0b85962624047.png",
"fullname": "Seonghyeon Ye",
"isPro": false,
"type": "user",
"user": "seonghyeonye"
}
},
{
"_id": "67b5625fb27eb6046b2ceece",
"hidden": false,
"name": "Joel Jang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:17:40.305Z",
"user": {
"_id": "613e1a9267835521a6816b04",
"avatarUrl": "/avatars/49edaa425bbce04dff92bbfb12a6b41c.svg",
"fullname": "Joel Jang",
"isPro": true,
"type": "user",
"user": "wkddydpf"
}
},
{
"_id": "67b5625fb27eb6046b2ceecf",
"hidden": false,
"name": "Yuquan Deng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5625fb27eb6046b2ceed0",
"hidden": false,
"name": "Lars Liden",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:17:09.378Z",
"user": {
"_id": "65bc13c7719492167d777718",
"avatarUrl": "/avatars/d1a3dc52c130b84a47c8a4ddd2e74be8.svg",
"fullname": "Lars Liden",
"isPro": false,
"type": "user",
"user": "larsliden"
}
},
{
"_id": "67b5625fb27eb6046b2ceed1",
"hidden": false,
"name": "Jianfeng Gao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:17:03.043Z",
"user": {
"_id": "641904caf9d6f1d772ec7af7",
"avatarUrl": "/avatars/4a63eac71eb30f70b1a0e9d4708f26c1.svg",
"fullname": "Jianfeng Gao",
"isPro": false,
"type": "user",
"user": "wyngjf"
}
}
]
| 2025-02-18T18:55:21 | Magma: A Foundation Model for Multimodal AI Agents | We present Magma, a foundation model that serves multimodal AI agentic tasks
in both the digital and physical worlds. Magma is a significant extension of
vision-language (VL) models in that it not only retains the VL understanding
ability (verbal intelligence) of the latter, but is also equipped with the
ability to plan and act in the visual-spatial world (spatial-temporal
intelligence) and complete agentic tasks ranging from UI navigation to robot
manipulation. To endow the agentic capabilities, Magma is pretrained on large
amounts of heterogeneous datasets spanning from images, videos to robotics
data, where the actionable visual objects (e.g., clickable buttons in GUI) in
images are labeled by Set-of-Mark (SoM) for action grounding, and the object
movements (e.g., the trace of human hands or robotic arms) in videos are
labeled by Trace-of-Mark (ToM) for action planning. Extensive experiments show
that SoM and ToM reach great synergy and facilitate the acquisition of
spatial-temporal intelligence for our Magma model, which is fundamental to a
wide range of tasks as shown in Fig.1. In particular, Magma creates new
state-of-the-art results on UI navigation and robotic manipulation tasks,
outperforming previous models that are specifically tailored to these tasks. On
image and video-related multimodal tasks, Magma also compares favorably to
popular large multimodal models that are trained on much larger datasets. We
make our model and code public for reproducibility at
https://microsoft.github.io/Magma. | 54 | 67b56265b27eb6046b2cf08f | null | null |
|
2025-02-18T23:37:46.756000 | Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities? | 2 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | false | null | 2502.12215 | [
{
"_id": "67b56007fa141a55e51d9d78",
"hidden": false,
"name": "Zhiyuan Zeng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b56007fa141a55e51d9d79",
"hidden": false,
"name": "Qinyuan Cheng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b56007fa141a55e51d9d7a",
"hidden": false,
"name": "Zhangyue Yin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T11:50:17.993Z",
"user": {
"_id": "628c5da32f09ccf530204dbe",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1653366416287-628c5da32f09ccf530204dbe.jpeg",
"fullname": "Zhangyue Yin",
"isPro": false,
"type": "user",
"user": "yinzhangyue"
}
},
{
"_id": "67b56007fa141a55e51d9d7b",
"hidden": false,
"name": "Yunhua Zhou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b56007fa141a55e51d9d7c",
"hidden": false,
"name": "Xipeng Qiu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T11:50:29.919Z",
"user": {
"_id": "61457b8deff2c9fdb4de4988",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1632381702899-61457b8deff2c9fdb4de4988.jpeg",
"fullname": "Xipeng Qiu",
"isPro": false,
"type": "user",
"user": "xpqiu"
}
}
]
| 2025-02-17T07:21:11 | Revisiting the Test-Time Scaling of o1-like Models: Do they Truly
Possess Test-Time Scaling Capabilities? | The advent of test-time scaling in large language models (LLMs), exemplified
by OpenAI's o1 series, has advanced reasoning capabilities by scaling
computational resource allocation during inference. While successors like QwQ,
Deepseek-R1 (R1) and LIMO replicate these advancements, whether these models
truly possess test-time scaling capabilities remains underexplored. This study
found that longer CoTs of these o1-like models do not consistently enhance
accuracy; in fact, correct solutions are often shorter than incorrect ones for
the same questions. Further investigation shows this phenomenon is closely
related to models' self-revision capabilities - longer CoTs contain more
self-revisions, which often lead to performance degradation. We then compare
sequential and parallel scaling strategies on QwQ, R1 and LIMO, finding that
parallel scaling achieves better coverage and scalability. Based on these
insights, we propose Shortest Majority Vote, a method that combines parallel
scaling strategies with CoT length characteristics, significantly improving
models' test-time scalability compared to conventional majority voting
approaches. | 16 | 67b56007fa141a55e51d9da7 | null | null |
|
2025-02-18T23:23:34.214000 | SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models | 2 | {
"_id": "64ad5f59b7e4b2c1ce47eb43",
"avatarUrl": "/avatars/1f13ebe21a90d8c99920aa2c8cd9ac45.svg",
"followerCount": 4,
"fullname": "Seanie Lee",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Seanie-lee",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/64ad5f59b7e4b2c1ce47eb43/ZEq_vSLjsXuPX3O-TWIpE.png"
]
| 2502.12464 | [
{
"_id": "67b55b2cc92c4aa82c13562d",
"hidden": false,
"name": "Seanie Lee",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:00:53.341Z",
"user": {
"_id": "64ad5f59b7e4b2c1ce47eb43",
"avatarUrl": "/avatars/1f13ebe21a90d8c99920aa2c8cd9ac45.svg",
"fullname": "Seanie Lee",
"isPro": false,
"type": "user",
"user": "Seanie-lee"
}
},
{
"_id": "67b55b2cc92c4aa82c13562e",
"hidden": false,
"name": "Dong Bok Lee",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55b2cc92c4aa82c13562f",
"hidden": false,
"name": "Dominik Wagner",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T11:12:27.148Z",
"user": {
"_id": "6311ba6f05cc08a1408d910a",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1662997515866-6311ba6f05cc08a1408d910a.png",
"fullname": "Dominik Wagner",
"isPro": false,
"type": "user",
"user": "dwgnr"
}
},
{
"_id": "67b55b2cc92c4aa82c135630",
"hidden": false,
"name": "Minki Kang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55b2cc92c4aa82c135631",
"hidden": false,
"name": "Haebin Seong",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:51:37.783Z",
"user": {
"_id": "63a9379e2e05ca32e352d93b",
"avatarUrl": "/avatars/6cda37befc873a92ed6d5dcba507954a.svg",
"fullname": "Haebin Seong",
"isPro": false,
"type": "user",
"user": "hbseong"
}
},
{
"_id": "67b55b2cc92c4aa82c135632",
"hidden": false,
"name": "Tobias Bocklet",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55b2cc92c4aa82c135633",
"hidden": false,
"name": "Juho Lee",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55b2cc92c4aa82c135634",
"hidden": false,
"name": "Sung Ju Hwang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T02:51:17 | SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety
Guardrails in Large Language Models | Deploying large language models (LLMs) in real-world applications requires
robust safety guard models to detect and block harmful user prompts. While
large safety guard models achieve strong performance, their computational cost
is substantial. To mitigate this, smaller distilled models are used, but they
often underperform on "hard" examples where the larger model provides accurate
predictions. We observe that many inputs can be reliably handled by the smaller
model, while only a small fraction require the larger model's capacity.
Motivated by this, we propose SafeRoute, a binary router that distinguishes
hard examples from easy ones. Our method selectively applies the larger safety
guard model to the data that the router considers hard, improving efficiency
while maintaining accuracy compared to solely using the larger safety guard
model. Experimental results on multiple benchmark datasets demonstrate that our
adaptive model selection significantly enhances the trade-off between
computational cost and safety performance, outperforming relevant baselines. | 27 | 67b55b2dc92c4aa82c13568b | null | null |
|
2025-02-18T22:59:16.530000 | MUDDFormer: Breaking Residual Bottlenecks in Transformers via Multiway Dynamic Dense Connections | 2 | {
"_id": "62d77440bad37ef354028365",
"avatarUrl": "/avatars/df0dea879e06fa814867e9aad03d1e68.svg",
"followerCount": null,
"fullname": "Da Xiao",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "xiaoda99",
"type": "user"
} | false | null | 2502.12170 | [
{
"_id": "67b5434f2b2ec6908fffe75e",
"hidden": false,
"name": "Da Xiao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5434f2b2ec6908fffe75f",
"hidden": false,
"name": "Qingye Meng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:23:45.040Z",
"user": {
"_id": "634f9c94cdc89e42cc7b194a",
"avatarUrl": "/avatars/42376474afe9b68dc44184c71e210529.svg",
"fullname": "Qingye Meng",
"isPro": false,
"type": "user",
"user": "Hilbertmeng"
}
},
{
"_id": "67b5434f2b2ec6908fffe760",
"hidden": false,
"name": "Shengping Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:23:38.331Z",
"user": {
"_id": "6706a37cca9b1a88fc3951ea",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/Y4GvpKiVgqm_2Kod52_B8.png",
"fullname": "lishengping",
"isPro": false,
"type": "user",
"user": "lishengping"
}
},
{
"_id": "67b5434f2b2ec6908fffe761",
"hidden": false,
"name": "Xingyuan Yuan",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-13T10:26:27 | MUDDFormer: Breaking Residual Bottlenecks in Transformers via Multiway
Dynamic Dense Connections | We propose MUltiway Dynamic Dense (MUDD) connections, a simple yet effective
method to address the limitations of residual connections and enhance
cross-layer information flow in Transformers. Unlike existing dense connection
approaches with static and shared connection weights, MUDD generates connection
weights dynamically depending on hidden states at each sequence position and
for each decoupled input stream (the query, key, value or residual) of a
Transformer block. MUDD connections can be seamlessly integrated into any
Transformer architecture to create MUDDFormer. Extensive experiments show that
MUDDFormer significantly outperforms Transformers across various model
architectures and scales in language modeling, achieving the performance of
Transformers trained with 1.8X-2.4X compute. Notably, MUDDPythia-2.8B matches
Pythia-6.9B in pretraining ppl and downstream tasks and even rivals Pythia-12B
in five-shot settings, while adding only 0.23% parameters and 0.4% computation.
Code in JAX and PyTorch and pre-trained models are available at
https://github.com/Caiyun-AI/MUDDFormer . | 12 | 67b543502b2ec6908fffe788 | null | null |
|
2025-02-18T22:46:16.586000 | Multilingual Encoder Knows more than You Realize: Shared Weights Pretraining for Extremely Low-Resource Languages | 2 | {
"_id": "6430bdd8cd31d174a9f900fb",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/Y9SPnRfpKSbYc7MhNdP-H.jpeg",
"followerCount": 2,
"fullname": "Ziyin Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Geralt-Targaryen",
"type": "user"
} | true | null | 2502.10852 | [
{
"_id": "67b55321f703732d151de666",
"hidden": false,
"name": "Zeli Su",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55321f703732d151de667",
"hidden": false,
"name": "Ziyin Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:10:05.400Z",
"user": {
"_id": "6430bdd8cd31d174a9f900fb",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/Y9SPnRfpKSbYc7MhNdP-H.jpeg",
"fullname": "Ziyin Zhang",
"isPro": false,
"type": "user",
"user": "Geralt-Targaryen"
}
},
{
"_id": "67b55321f703732d151de668",
"hidden": false,
"name": "Guixian Xu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:09:59.600Z",
"user": {
"_id": "6747329d228a652d5707e094",
"avatarUrl": "/avatars/f33e118950e329ce5612877413806e49.svg",
"fullname": "GUIXIAN XU",
"isPro": false,
"type": "user",
"user": "Stuart-Xu"
}
},
{
"_id": "67b55321f703732d151de669",
"hidden": false,
"name": "Jianing Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55321f703732d151de66a",
"hidden": false,
"name": "XU Han",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55321f703732d151de66b",
"hidden": false,
"name": "Ting Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55321f703732d151de66c",
"hidden": false,
"name": "Yushuang Dong",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-15T16:53:10 | Multilingual Encoder Knows more than You Realize: Shared Weights
Pretraining for Extremely Low-Resource Languages | While multilingual language models like XLM-R have advanced multilingualism
in NLP, they still perform poorly in extremely low-resource languages. This
situation is exacerbated by the fact that modern LLMs such as LLaMA and Qwen
support far fewer languages than XLM-R, making text generation models
non-existent for many languages in the world. To tackle this challenge, we
propose a novel framework for adapting multilingual encoders to text generation
in extremely low-resource languages. By reusing the weights between the encoder
and the decoder, our framework allows the model to leverage the learned
semantic space of the encoder, enabling efficient learning and effective
generalization in low-resource languages. Applying this framework to four
Chinese minority languages, we present XLM-SWCM, and demonstrate its superior
performance on various downstream tasks even when compared with much larger
models. | 2 | 67b55322f703732d151de69d | null | null |
|
2025-02-18T22:43:02.567000 | Continuous Diffusion Model for Language Modeling | 4 | {
"_id": "65e5bd4568234ef5d6decadc",
"avatarUrl": "/avatars/c41095a946c0176b949c0b3566136c05.svg",
"followerCount": 4,
"fullname": "Jaehyeong Jo",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "harryjo97",
"type": "user"
} | true | null | 2502.11564 | [
{
"_id": "67b40f93aba9e111862052ab",
"hidden": false,
"name": "Jaehyeong Jo",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:27.544Z",
"user": {
"_id": "65e5bd4568234ef5d6decadc",
"avatarUrl": "/avatars/c41095a946c0176b949c0b3566136c05.svg",
"fullname": "Jaehyeong Jo",
"isPro": false,
"type": "user",
"user": "harryjo97"
}
},
{
"_id": "67b40f93aba9e111862052ac",
"hidden": false,
"name": "Sung Ju Hwang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T08:54:29 | Continuous Diffusion Model for Language Modeling | Diffusion models have emerged as a promising alternative to autoregressive
models in modeling discrete categorical data. Yet diffusion models that
directly work on discrete data space do not fully exploit the power of
iterative refinement, as the signals are lost during the transition between
discrete states. Existing continuous diffusion models for discrete data have
limited performance compared to discrete approaches, and the unclear link
between them restricts the development of diffusion models for discrete data.
In this work, we propose a continuous diffusion model for language modeling
that incorporates the geometry of the underlying categorical distribution. We
establish a connection between the discrete diffusion and continuous flow on
the statistical manifold, and building on the analogy, we introduce a simple
design for the diffusion process that generalizes previous discrete diffusion
models. We further propose a simulation-free training framework based on radial
symmetry and a simple technique to address the high dimensionality of the
manifold. Comprehensive experiments on language modeling benchmarks and other
modalities show that our method outperforms existing discrete diffusion models
and approaches the performance of autoregressive models. Codes available at
https://github.com/harryjo97/RDLM{https://github.com/harryjo97/RDLM}. | 50 | 67b40f94aba9e111862052d5 | null | null |
|
2025-02-18T22:35:23.066000 | HealthGPT: A Medical Large Vision-Language Model for Unifying Comprehension and Generation via Heterogeneous Knowledge Adaptation | 2 | {
"_id": "65fc18edfb66882aba4d548e",
"avatarUrl": "/avatars/f70d47fe4aba98b5a5cd64f7e002dfd2.svg",
"followerCount": null,
"fullname": "wenqiao",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "wannature",
"type": "user"
} | true | null | 2502.09838 | [
{
"_id": "67b55078a64445f58c771d84",
"hidden": true,
"name": "Tianwei Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d85",
"hidden": false,
"name": "Wenqiao Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:40:41.324Z",
"user": {
"_id": "65fc18edfb66882aba4d548e",
"avatarUrl": "/avatars/f70d47fe4aba98b5a5cd64f7e002dfd2.svg",
"fullname": "wenqiao",
"isPro": false,
"type": "user",
"user": "wannature"
}
},
{
"_id": "67b55078a64445f58c771d86",
"hidden": false,
"name": "Sijing Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d87",
"hidden": false,
"name": "Yuqian Yuan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d88",
"hidden": false,
"name": "Binhe Yu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:41:15.907Z",
"user": {
"_id": "648ad056dc6a5b4b88306ae2",
"avatarUrl": "/avatars/66a47a11626c05b706de33a3184182e9.svg",
"fullname": "yu",
"isPro": false,
"type": "user",
"user": "binheyu1991"
}
},
{
"_id": "67b55078a64445f58c771d89",
"hidden": false,
"name": "Haoyuan Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d8a",
"hidden": false,
"name": "Wanggui He",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d8b",
"hidden": false,
"name": "Hao Jiang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d8c",
"hidden": false,
"name": "Mengze Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d8d",
"hidden": false,
"name": "Xiaohui Song",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T14:42:09.080Z",
"user": {
"_id": "635156f1f74cdcca6f7acf70",
"avatarUrl": "/avatars/133e88b30255425f6da7777737f91e81.svg",
"fullname": "Xiaohui Song",
"isPro": false,
"type": "user",
"user": "fpcsong"
}
},
{
"_id": "67b55078a64445f58c771d8e",
"hidden": false,
"name": "Siliang Tang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d8f",
"hidden": false,
"name": "Jun Xiao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d90",
"hidden": false,
"name": "Hui Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d91",
"hidden": false,
"name": "Yueting Zhuang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b55078a64445f58c771d92",
"hidden": false,
"name": "Beng Chin Ooi",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-14T00:42:36 | HealthGPT: A Medical Large Vision-Language Model for Unifying
Comprehension and Generation via Heterogeneous Knowledge Adaptation | We present HealthGPT, a powerful Medical Large Vision-Language Model
(Med-LVLM) that integrates medical visual comprehension and generation
capabilities within a unified autoregressive paradigm. Our bootstrapping
philosophy is to progressively adapt heterogeneous comprehension and generation
knowledge to pre-trained large language models (LLMs). This is achieved through
a novel heterogeneous low-rank adaptation (H-LoRA) technique, which is
complemented by a tailored hierarchical visual perception approach and a
three-stage learning strategy. To effectively learn the HealthGPT, we devise a
comprehensive medical domain-specific comprehension and generation dataset
called VL-Health. Experimental results demonstrate exceptional performance and
scalability of HealthGPT in medical visual unified tasks. Our project can be
accessed at https://github.com/DCDmllm/HealthGPT. | 10 | 67b5507aa64445f58c771df9 | null | null |
|
2025-02-18T22:08:27.750000 | Multimodal Mamba: Decoder-only Multimodal State Space Model via Quadratic to Linear Distillation | 2 | {
"_id": "6577073fc2bf55b1f6bafb49",
"avatarUrl": "/avatars/58803398b1a918b7570db17893e65122.svg",
"followerCount": 4,
"fullname": "liao",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "LegendBC",
"type": "user"
} | true | null | 2502.13145 | [
{
"_id": "67b54b04bd51b4e46e39d287",
"hidden": false,
"name": "Bencheng Liao",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:00.934Z",
"user": {
"_id": "6577073fc2bf55b1f6bafb49",
"avatarUrl": "/avatars/58803398b1a918b7570db17893e65122.svg",
"fullname": "liao",
"isPro": false,
"type": "user",
"user": "LegendBC"
}
},
{
"_id": "67b54b04bd51b4e46e39d288",
"hidden": false,
"name": "Hongyuan Tao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:13:43.349Z",
"user": {
"_id": "66a105bb456284adf458d656",
"avatarUrl": "/avatars/b543a324f7e159d6e84bc68915e93d24.svg",
"fullname": "Tao Hongyuan",
"isPro": false,
"type": "user",
"user": "HongyuanTao"
}
},
{
"_id": "67b54b04bd51b4e46e39d289",
"hidden": false,
"name": "Qian Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b54b04bd51b4e46e39d28a",
"hidden": false,
"name": "Tianheng Cheng",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:00:58.351Z",
"user": {
"_id": "646b3db131968a60a01e4cf5",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/646b3db131968a60a01e4cf5/DhfdqUYQaD1Qa8Svw996J.jpeg",
"fullname": "Tianheng Cheng",
"isPro": false,
"type": "user",
"user": "wondervictor"
}
},
{
"_id": "67b54b04bd51b4e46e39d28b",
"hidden": false,
"name": "Yingyue Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b54b04bd51b4e46e39d28c",
"hidden": false,
"name": "Haoran Yin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b54b04bd51b4e46e39d28d",
"hidden": false,
"name": "Wenyu Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:12:36.866Z",
"user": {
"_id": "66c2e7fc934e2f07753542ac",
"avatarUrl": "/avatars/f6fa3f94435cf1c1d06daa6c925d07d0.svg",
"fullname": "LWY",
"isPro": false,
"type": "user",
"user": "wenyuliu"
}
},
{
"_id": "67b54b04bd51b4e46e39d28e",
"hidden": false,
"name": "Xinggang Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T18:59:57 | Multimodal Mamba: Decoder-only Multimodal State Space Model via
Quadratic to Linear Distillation | Recent Multimodal Large Language Models (MLLMs) have achieved remarkable
performance but face deployment challenges due to their quadratic computational
complexity, growing Key-Value cache requirements, and reliance on separate
vision encoders. We propose mmMamba, a framework for developing
linear-complexity native multimodal state space models through progressive
distillation from existing MLLMs using moderate academic computational
resources. Our approach enables the direct conversion of trained decoder-only
MLLMs to linear-complexity architectures without requiring pre-trained
RNN-based LLM or vision encoders. We propose an seeding strategy to carve Mamba
from trained Transformer and a three-stage distillation recipe, which can
effectively transfer the knowledge from Transformer to Mamba while preserving
multimodal capabilities. Our method also supports flexible hybrid architectures
that combine Transformer and Mamba layers for customizable
efficiency-performance trade-offs. Distilled from the Transformer-based
decoder-only HoVLE, mmMamba-linear achieves competitive performance against
existing linear and quadratic-complexity VLMs, while mmMamba-hybrid further
improves performance significantly, approaching HoVLE's capabilities. At 103K
tokens, mmMamba-linear demonstrates 20.6times speedup and 75.8% GPU memory
reduction compared to HoVLE, while mmMamba-hybrid achieves 13.5times speedup
and 60.2% memory savings. Code and models are released at
https://github.com/hustvl/mmMamba | 36 | 67b54b05bd51b4e46e39d2bb | null | null |
|
2025-02-18T22:06:19.200000 | FLAG-Trader: Fusion LLM-Agent with Gradient-based Reinforcement Learning for Financial Trading | 2 | {
"_id": "63b58ed5889aa6707f0bb0f4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63b58ed5889aa6707f0bb0f4/znl74_aMswlV8VtHrfj3G.jpeg",
"followerCount": 15,
"fullname": "Jimin Huang",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "jiminHuang",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/63b58ed5889aa6707f0bb0f4/2C9mhT-1Qz14hik7sxjf2.png"
]
| 2502.11433 | [
{
"_id": "67b54a644508bd0617598c21",
"hidden": false,
"name": "Guojun Xiong",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:14:19.641Z",
"user": {
"_id": "67b54cbcd9f66be7f6f3f7de",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/RZwRfp6AcseTCDVXW_eUb.png",
"fullname": "Guojun Xiong",
"isPro": false,
"type": "user",
"user": "xionggj001"
}
},
{
"_id": "67b54a644508bd0617598c22",
"hidden": false,
"name": "Zhiyang Deng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:14:39.220Z",
"user": {
"_id": "668dedbdf278aa900ce400c9",
"avatarUrl": "/avatars/b38e29c6b5092f5892bc2e9a7e625c88.svg",
"fullname": "Zhiyang Deng",
"isPro": false,
"type": "user",
"user": "zdeng10"
}
},
{
"_id": "67b54a644508bd0617598c23",
"hidden": false,
"name": "Keyi Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b54a644508bd0617598c24",
"hidden": false,
"name": "Yupeng Cao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:15:07.509Z",
"user": {
"_id": "62dd8f328456396d4f8aa894",
"avatarUrl": "/avatars/af8f5dc7ff937e3e849ecdfd9ca4750b.svg",
"fullname": "Yupeng Cao",
"isPro": false,
"type": "user",
"user": "YupengCao"
}
},
{
"_id": "67b54a644508bd0617598c25",
"hidden": false,
"name": "Haohang Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:15:15.689Z",
"user": {
"_id": "634cabd104491d9f7111eea3",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1665969099097-noauth.jpeg",
"fullname": "Haohang Li",
"isPro": true,
"type": "user",
"user": "Acatsama"
}
},
{
"_id": "67b54a644508bd0617598c26",
"hidden": false,
"name": "Yangyang Yu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:15:22.050Z",
"user": {
"_id": "64f757c6016d60f3199ef5e6",
"avatarUrl": "/avatars/2659ba698081265d0480b08161718013.svg",
"fullname": "Yangyang Yu",
"isPro": false,
"type": "user",
"user": "ShirleyY"
}
},
{
"_id": "67b54a644508bd0617598c27",
"hidden": false,
"name": "Xueqing Peng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:15:29.692Z",
"user": {
"_id": "63a0c0803c8841cfe2cd1f15",
"avatarUrl": "/avatars/bbe216db7a33612f23d23ce4ed4ba3f9.svg",
"fullname": "Xueqing Peng",
"isPro": false,
"type": "user",
"user": "Xueqing"
}
},
{
"_id": "67b54a644508bd0617598c28",
"hidden": false,
"name": "Mingquan Lin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:15:52.483Z",
"user": {
"_id": "6650e0a99ccb17d9679653c5",
"avatarUrl": "/avatars/ed08188e64bdebf58329304742f9ac16.svg",
"fullname": "Mingquan Lin",
"isPro": false,
"type": "user",
"user": "mq0051"
}
},
{
"_id": "67b54a644508bd0617598c29",
"hidden": false,
"name": "Kaleb E Smith",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b54a644508bd0617598c2a",
"hidden": false,
"name": "Xiao-Yang Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b54a644508bd0617598c2b",
"hidden": false,
"name": "Jimin Huang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:03.181Z",
"user": {
"_id": "63b58ed5889aa6707f0bb0f4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63b58ed5889aa6707f0bb0f4/znl74_aMswlV8VtHrfj3G.jpeg",
"fullname": "Jimin Huang",
"isPro": true,
"type": "user",
"user": "jiminHuang"
}
},
{
"_id": "67b54a644508bd0617598c2c",
"hidden": false,
"name": "Sophia Ananiadou",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:16:09.788Z",
"user": {
"_id": "66f6cb352c5d4ef3578a9c3f",
"avatarUrl": "/avatars/0a70c94072bc5e1d018cf12da0904ff0.svg",
"fullname": "Sophia Ananiadou",
"isPro": false,
"type": "user",
"user": "Effoula"
}
},
{
"_id": "67b54a644508bd0617598c2d",
"hidden": false,
"name": "Qianqian Xie",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:16:24.340Z",
"user": {
"_id": "6479f4317c18dca75e9a9324",
"avatarUrl": "/avatars/9aa709230b057f57ee4415c04a622c63.svg",
"fullname": "Xie",
"isPro": false,
"type": "user",
"user": "QianqianXie1994"
}
}
]
| 2025-02-17T04:45:53 | FLAG-Trader: Fusion LLM-Agent with Gradient-based Reinforcement Learning
for Financial Trading | Large language models (LLMs) fine-tuned on multimodal financial data have
demonstrated impressive reasoning capabilities in various financial tasks.
However, they often struggle with multi-step, goal-oriented scenarios in
interactive financial markets, such as trading, where complex agentic
approaches are required to improve decision-making. To address this, we propose
FLAG-Trader, a unified architecture integrating linguistic processing
(via LLMs) with gradient-driven reinforcement learning (RL) policy
optimization, in which a partially fine-tuned LLM acts as the policy network,
leveraging pre-trained knowledge while adapting to the financial domain through
parameter-efficient fine-tuning. Through policy gradient optimization driven by
trading rewards, our framework not only enhances LLM performance in trading but
also improves results on other financial-domain tasks. We present extensive
empirical evidence to validate these enhancements. | 31 | 67b54a654508bd0617598c7e | null | null |
|
2025-02-18T21:59:45.466000 | Rethinking Diverse Human Preference Learning through Principal Component Analysis | 3 | {
"_id": "64d45451c34a346181b130dd",
"avatarUrl": "/avatars/9bb8205b889337df5d321539c9b5d69d.svg",
"followerCount": 6,
"fullname": "Rui Yang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Ray2333",
"type": "user"
} | true | null | 2502.13131 | [
{
"_id": "67b5461d29cc269e5a4eb823",
"hidden": false,
"name": "Feng Luo",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5461d29cc269e5a4eb824",
"hidden": true,
"name": "Rui Yang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:23.095Z",
"user": {
"_id": "64d45451c34a346181b130dd",
"avatarUrl": "/avatars/9bb8205b889337df5d321539c9b5d69d.svg",
"fullname": "Rui Yang",
"isPro": false,
"type": "user",
"user": "Ray2333"
}
},
{
"_id": "67b5461d29cc269e5a4eb825",
"hidden": false,
"name": "Hao Sun",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5461d29cc269e5a4eb826",
"hidden": false,
"name": "Chunyuan Deng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:56:33.053Z",
"user": {
"_id": "634b9914dcf125e4da02498b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/634b9914dcf125e4da02498b/crRgFroWq5U6XWtvlTXSZ.jpeg",
"fullname": "Chunyuan Deng",
"isPro": false,
"type": "user",
"user": "CharlesDDDD"
}
},
{
"_id": "67b5461d29cc269e5a4eb827",
"hidden": false,
"name": "Jiarui Yao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5461d29cc269e5a4eb828",
"hidden": false,
"name": "Jingyan Shen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b5461d29cc269e5a4eb829",
"hidden": false,
"name": "Huan Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:52:47.329Z",
"user": {
"_id": "6719d581a6cad13741b8bc7f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6719d581a6cad13741b8bc7f/w4EttqfXRgWZJc6HpYOS9.jpeg",
"fullname": "Huan Zhang",
"isPro": false,
"type": "user",
"user": "huanzhang12"
}
},
{
"_id": "67b5461d29cc269e5a4eb82a",
"hidden": false,
"name": "Hanjie Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T18:55:26 | Rethinking Diverse Human Preference Learning through Principal Component
Analysis | Understanding human preferences is crucial for improving foundation models
and building personalized AI systems. However, preferences are inherently
diverse and complex, making it difficult for traditional reward models to
capture their full range. While fine-grained preference data can help,
collecting it is expensive and hard to scale. In this paper, we introduce
Decomposed Reward Models (DRMs), a novel approach that extracts diverse human
preferences from binary comparisons without requiring fine-grained annotations.
Our key insight is to represent human preferences as vectors and analyze them
using Principal Component Analysis (PCA). By constructing a dataset of
embedding differences between preferred and rejected responses, DRMs identify
orthogonal basis vectors that capture distinct aspects of preference. These
decomposed rewards can be flexibly combined to align with different user needs,
offering an interpretable and scalable alternative to traditional reward
models. We demonstrate that DRMs effectively extract meaningful preference
dimensions (e.g., helpfulness, safety, humor) and adapt to new users without
additional training. Our results highlight DRMs as a powerful framework for
personalized and interpretable LLM alignment. | 35 | 67b5461f29cc269e5a4eb8bc | null | null |
|
2025-02-18T21:57:00.289000 | HeadInfer: Memory-Efficient LLM Inference by Head-wise Offloading | 2 | {
"_id": "64cb48f7667f4f808535107e",
"avatarUrl": "/avatars/8f77f378ad665b246e1ea3aaba2153ae.svg",
"followerCount": 1,
"fullname": "chengluo",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "wdlctc",
"type": "user"
} | true | null | 2502.12574 | [
{
"_id": "67b547f555d0424a31b9c384",
"hidden": false,
"name": "Cheng Luo",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:40:25.130Z",
"user": {
"_id": "64cb48f7667f4f808535107e",
"avatarUrl": "/avatars/8f77f378ad665b246e1ea3aaba2153ae.svg",
"fullname": "chengluo",
"isPro": false,
"type": "user",
"user": "wdlctc"
}
},
{
"_id": "67b547f555d0424a31b9c385",
"hidden": false,
"name": "Zefan Cai",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:40:47.077Z",
"user": {
"_id": "64b15284372d4340772a3dca",
"avatarUrl": "/avatars/417d5f1bc1bcb5e4d5de6169673c2cf7.svg",
"fullname": "Zefan Cai",
"isPro": false,
"type": "user",
"user": "ZefanCai"
}
},
{
"_id": "67b547f555d0424a31b9c386",
"hidden": false,
"name": "Hanshi Sun",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547f555d0424a31b9c387",
"hidden": false,
"name": "Jinqi Xiao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:41:01.931Z",
"user": {
"_id": "64c15c5bea792b1950e302e4",
"avatarUrl": "/avatars/51f84365cc08a1dcd5da70968389aed2.svg",
"fullname": "Jinqi Xiao",
"isPro": false,
"type": "user",
"user": "jinqixiao"
}
},
{
"_id": "67b547f555d0424a31b9c388",
"hidden": false,
"name": "Bo Yuan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547f555d0424a31b9c389",
"hidden": false,
"name": "Wen Xiao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547f555d0424a31b9c38a",
"hidden": false,
"name": "Junjie Hu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:41:18.304Z",
"user": {
"_id": "675f8271a63fff7b5bcbc478",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/9tJn7NyzLMreCJVH4wRho.png",
"fullname": "Junjie Hu",
"isPro": false,
"type": "user",
"user": "junjiehu"
}
},
{
"_id": "67b547f555d0424a31b9c38b",
"hidden": false,
"name": "Jiawei Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547f555d0424a31b9c38c",
"hidden": false,
"name": "Beidi Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:39:20.563Z",
"user": {
"_id": "64b732f832403871593e082c",
"avatarUrl": "/avatars/dd21932b0c167131ee7545a622c46c3c.svg",
"fullname": "Beidi Chen",
"isPro": false,
"type": "user",
"user": "beidic"
}
},
{
"_id": "67b547f555d0424a31b9c38d",
"hidden": false,
"name": "Anima Anandkumar",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:39:15.091Z",
"user": {
"_id": "6532920b3e385cfc6002938d",
"avatarUrl": "/avatars/cb9cc6d2733031582c83f56dc6cd1dd5.svg",
"fullname": "Anima Anandkumar",
"isPro": false,
"type": "user",
"user": "animakumar"
}
}
]
| 2025-02-18T06:26:05 | HeadInfer: Memory-Efficient LLM Inference by Head-wise Offloading | Transformer-based large language models (LLMs) demonstrate impressive
performance in long context generation. Extending the context length has
disproportionately shifted the memory footprint of LLMs during inference to the
key-value cache (KV cache). In this paper, we propose HEADINFER, which offloads
the KV cache to CPU RAM while avoiding the need to fully store the KV cache for
any transformer layer on the GPU. HEADINFER employs a fine-grained, head-wise
offloading strategy, maintaining only selective attention heads KV cache on the
GPU while computing attention output dynamically. Through roofline analysis, we
demonstrate that HEADINFER maintains computational efficiency while
significantly reducing memory footprint. We evaluate HEADINFER on the
Llama-3-8B model with a 1-million-token sequence, reducing the GPU memory
footprint of the KV cache from 128 GB to 1 GB and the total GPU memory usage
from 207 GB to 17 GB, achieving a 92% reduction compared to BF16 baseline
inference. Notably, HEADINFER enables 4-million-token inference with an 8B
model on a single consumer GPU with 24GB memory (e.g., NVIDIA RTX 4090) without
approximation methods. | 11 | 67b547f755d0424a31b9c3e5 | null | null |
|
2025-02-18T21:56:39.407000 | Phantom: Subject-consistent video generation via cross-modal alignment | 2 | {
"_id": "63a950ac3453852ef5394178",
"avatarUrl": "/avatars/48a5e537b10e2247a17e63502e3201a6.svg",
"followerCount": 1,
"fullname": "Lijie Liu",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "liulj13",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/63a950ac3453852ef5394178/HuVZ5d9xTlI4R1onRv_F5.mp4"
]
| 2502.11079 | [
{
"_id": "67b40141ad717fe02e188c1a",
"hidden": false,
"name": "Lijie Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:42.570Z",
"user": {
"_id": "63a950ac3453852ef5394178",
"avatarUrl": "/avatars/48a5e537b10e2247a17e63502e3201a6.svg",
"fullname": "Lijie Liu",
"isPro": false,
"type": "user",
"user": "liulj13"
}
},
{
"_id": "67b40141ad717fe02e188c1b",
"hidden": false,
"name": "Tianxiang Ma",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:45:00.117Z",
"user": {
"_id": "657ab4705e1c941f4c2f7877",
"avatarUrl": "/avatars/c450f81f83dd0436ae120ab15616c4f7.svg",
"fullname": "Tianxiang Ma",
"isPro": false,
"type": "user",
"user": "Grayson111"
}
},
{
"_id": "67b40141ad717fe02e188c1c",
"hidden": false,
"name": "Bingchuan Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:47:57.441Z",
"user": {
"_id": "63b415037af2e415f2599c18",
"avatarUrl": "/avatars/4afbe7d6d05a702f1beeed9c53e78153.svg",
"fullname": "Bingchuan Li",
"isPro": false,
"type": "user",
"user": "lbc402"
}
},
{
"_id": "67b40141ad717fe02e188c1d",
"hidden": false,
"name": "Zhuowei Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T09:47:50.995Z",
"user": {
"_id": "6304e2dabad6ce7fc0287d57",
"avatarUrl": "/avatars/3fd4a9a62b0ef98db2573411463a9247.svg",
"fullname": "Zhuowei_Chen",
"isPro": false,
"type": "user",
"user": "ZhuoweiChen"
}
},
{
"_id": "67b40141ad717fe02e188c1e",
"hidden": false,
"name": "Jiawei Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b40141ad717fe02e188c1f",
"hidden": false,
"name": "Qian He",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b40141ad717fe02e188c20",
"hidden": false,
"name": "Xinglong Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-16T11:02:50 | Phantom: Subject-consistent video generation via cross-modal alignment | The continuous development of foundational models for video generation is
evolving into various applications, with subject-consistent video generation
still in the exploratory stage. We refer to this as Subject-to-Video, which
extracts subject elements from reference images and generates
subject-consistent video through textual instructions. We believe that the
essence of subject-to-video lies in balancing the dual-modal prompts of text
and image, thereby deeply and simultaneously aligning both text and visual
content. To this end, we propose Phantom, a unified video generation framework
for both single and multi-subject references. Building on existing
text-to-video and image-to-video architectures, we redesign the joint
text-image injection model and drive it to learn cross-modal alignment via
text-image-video triplet data. In particular, we emphasize subject consistency
in human generation, covering existing ID-preserving video generation while
offering enhanced advantages. The project homepage is here
https://phantom-video.github.io/Phantom/. | 52 | 67b40144ad717fe02e188cb2 | null | null |
|
2025-02-18T21:55:26.822000 | Crowd Comparative Reasoning: Unlocking Comprehensive Evaluations for LLM-as-a-Judge | 2 | {
"_id": "62a42f22c683d02f5b63320c",
"avatarUrl": "/avatars/bc611abe9c4ef8d378123cb8ac9fdbf2.svg",
"followerCount": null,
"fullname": "Qiyuan Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "DonJoey",
"type": "user"
} | true | null | 2502.12501 | [
{
"_id": "67b547ffc9071a3e97139532",
"hidden": false,
"name": "Qiyuan Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:10.215Z",
"user": {
"_id": "62a42f22c683d02f5b63320c",
"avatarUrl": "/avatars/bc611abe9c4ef8d378123cb8ac9fdbf2.svg",
"fullname": "Qiyuan Zhang",
"isPro": false,
"type": "user",
"user": "DonJoey"
}
},
{
"_id": "67b547ffc9071a3e97139533",
"hidden": false,
"name": "Yufei Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547ffc9071a3e97139534",
"hidden": false,
"name": "Yuxin Jiang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:08.101Z",
"user": {
"_id": "63c20105726f62e411fbe882",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63c20105726f62e411fbe882/2UsU9O2psbDjJzz-sAmGH.jpeg",
"fullname": "Yuxin Jiang",
"isPro": false,
"type": "user",
"user": "YuxinJiang"
}
},
{
"_id": "67b547ffc9071a3e97139535",
"hidden": false,
"name": "Liangyou Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547ffc9071a3e97139536",
"hidden": false,
"name": "Chuhan Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547ffc9071a3e97139537",
"hidden": false,
"name": "Yasheng Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b547ffc9071a3e97139538",
"hidden": false,
"name": "Xin Jiang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T11:54:48.156Z",
"user": {
"_id": "647415007afa69c3c7a98f1f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/647415007afa69c3c7a98f1f/pEl0PozmzNK8_PwUMiikd.jpeg",
"fullname": "Xin Jiang",
"isPro": false,
"type": "user",
"user": "horiz94"
}
},
{
"_id": "67b547ffc9071a3e97139539",
"hidden": false,
"name": "Lifeng Shang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T11:54:33.330Z",
"user": {
"_id": "655b1360c11dee7f7e7cf794",
"avatarUrl": "/avatars/efb4b91e9bb8ab531331c8e4296f754c.svg",
"fullname": "lifengshang",
"isPro": false,
"type": "user",
"user": "lifengshang"
}
},
{
"_id": "67b547ffc9071a3e9713953a",
"hidden": false,
"name": "Ruiming Tang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T11:54:25.814Z",
"user": {
"_id": "6728c3b8d5ceae39aa1d2fdd",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/VV4-DjhxzEVceNLmsokn2.png",
"fullname": "tang ruiming",
"isPro": false,
"type": "user",
"user": "zhangsan5421"
}
},
{
"_id": "67b547ffc9071a3e9713953b",
"hidden": false,
"name": "Fuyuan Lyu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T11:54:14.440Z",
"user": {
"_id": "65d2bb5c6130ef7be012d235",
"avatarUrl": "/avatars/1c1e3bbb2c683a5c9d1f792a2c13fc4a.svg",
"fullname": "Fuyuan Lyu",
"isPro": false,
"type": "user",
"user": "silentspring2"
}
},
{
"_id": "67b547ffc9071a3e9713953c",
"hidden": false,
"name": "Chen Ma",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T03:31:06 | Crowd Comparative Reasoning: Unlocking Comprehensive Evaluations for
LLM-as-a-Judge | LLM-as-a-Judge, which generates chain-of-thought (CoT) judgments, has become
a widely adopted auto-evaluation method. However, its reliability is
compromised by the CoT reasoning's inability to capture comprehensive and
deeper details, often leading to incomplete outcomes. Existing methods mainly
rely on majority voting or criteria expansion, which is insufficient to address
the limitation in CoT. We propose Crowd-based Comparative Evaluation, which
introduces additional crowd responses to compare with the candidate responses,
thereby exposing deeper and more comprehensive details within the candidate
responses. This process effectively guides LLM-as-a-Judge to provide a more
detailed CoT judgment. Extensive experiments demonstrate that our approach
enhances evaluation reliability, achieving an average accuracy gain of 6.7%
across five benchmarks. Moreover, our method produces higher-quality CoTs that
facilitate judge distillation and exhibit superior performance in rejection
sampling for supervised fine-tuning (SFT), referred to as crowd rejection
sampling, thereby enabling more efficient SFT. Our analysis confirms that CoTs
generated by ours are more comprehensive and of higher quality, and evaluation
accuracy improves as inference scales. | 6 | 67b54800c9071a3e9713956c | null | null |
|
2025-02-18T21:52:22.326000 | RealSyn: An Effective and Scalable Multimodal Interleaved Document Transformation Paradigm | 2 | {
"_id": "63e202f352b7578dba448ab5",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63e202f352b7578dba448ab5/8itVBLcv14m7OVsoF8h1o.jpeg",
"followerCount": 4,
"fullname": "Yang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Kaichengalex",
"type": "user"
} | true | null | 2502.12513 | [
{
"_id": "67b545fd88527668fa8bcc14",
"hidden": false,
"name": "Tiancheng Gu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:19:15.243Z",
"user": {
"_id": "6508712e7ee07e274b0f4c94",
"avatarUrl": "/avatars/23fe5593b0bce36c2167c3142e57e0e9.svg",
"fullname": "Tiancheng Gu",
"isPro": false,
"type": "user",
"user": "GaryGuuu"
}
},
{
"_id": "67b545fd88527668fa8bcc15",
"hidden": false,
"name": "Kaicheng Yang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T14:37:43.180Z",
"user": {
"_id": "63e202f352b7578dba448ab5",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63e202f352b7578dba448ab5/8itVBLcv14m7OVsoF8h1o.jpeg",
"fullname": "Yang",
"isPro": false,
"type": "user",
"user": "Kaichengalex"
}
},
{
"_id": "67b545fd88527668fa8bcc16",
"hidden": false,
"name": "Chaoyi Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b545fd88527668fa8bcc17",
"hidden": false,
"name": "Yin Xie",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b545fd88527668fa8bcc18",
"hidden": true,
"name": "Xiang An",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-26T08:39:00.405Z",
"user": {
"_id": "6478679d7b370854241b2ad8",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6478679d7b370854241b2ad8/dBczWYYdfEt9tQcnVGhQk.jpeg",
"fullname": "xiangan",
"isPro": false,
"type": "user",
"user": "xiangan"
}
},
{
"_id": "67b545fd88527668fa8bcc19",
"hidden": false,
"name": "Ziyong Feng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b545fd88527668fa8bcc1a",
"hidden": false,
"name": "Dongnan Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:20:00.745Z",
"user": {
"_id": "65e7f19e14856e8859fd8adc",
"avatarUrl": "/avatars/203e919c361db94028a1b3c6ea52f0c2.svg",
"fullname": "Dongnan Liu",
"isPro": false,
"type": "user",
"user": "Nina0607"
}
},
{
"_id": "67b545fd88527668fa8bcc1b",
"hidden": false,
"name": "Weidong Cai",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:20:18.377Z",
"user": {
"_id": "6760a8f5e4b55ba1b2b0a7b4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/NddUMmwmZFbS25v1q8KyS.png",
"fullname": "Weidong Cai",
"isPro": false,
"type": "user",
"user": "SeriousBro"
}
},
{
"_id": "67b545fd88527668fa8bcc1c",
"hidden": false,
"name": "Jiankang Deng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:20:25.455Z",
"user": {
"_id": "62cc7a38376917c0223dd24b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1657566065867-noauth.png",
"fullname": "JiankangDeng",
"isPro": false,
"type": "user",
"user": "JiankangDeng"
}
}
]
| 2025-02-18T03:58:38 | RealSyn: An Effective and Scalable Multimodal Interleaved Document
Transformation Paradigm | After pre-training on extensive image-text pairs, Contrastive Language-Image
Pre-training (CLIP) demonstrates promising performance on a wide variety of
benchmarks. However, a substantial volume of non-paired data, such as
multimodal interleaved documents, remains underutilized for vision-language
representation learning. To fully leverage these unpaired documents, we
initially establish a Real-World Data Extraction pipeline to extract
high-quality images and texts. Then we design a hierarchical retrieval method
to efficiently associate each image with multiple semantically relevant
realistic texts. To further enhance fine-grained visual information, we propose
an image semantic augmented generation module for synthetic text production.
Furthermore, we employ a semantic balance sampling strategy to improve dataset
diversity, enabling better learning of long-tail concepts. Based on these
innovations, we construct RealSyn, a dataset combining realistic and synthetic
texts, available in three scales: 15M, 30M, and 100M. Extensive experiments
demonstrate that RealSyn effectively advances vision-language representation
learning and exhibits strong scalability. Models pre-trained on RealSyn achieve
state-of-the-art performance on multiple downstream tasks. To facilitate future
research, the RealSyn dataset and pre-trained model weights are released at
https://github.com/deepglint/RealSyn. | 15 | 67b545fe88527668fa8bcc65 | null | null |
|
2025-02-18T21:51:33.957000 | SoFar: Language-Grounded Orientation Bridges Spatial Reasoning and Object Manipulation | 2 | {
"_id": "63c3e8abc7d7f4c63a515a02",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63c3e8abc7d7f4c63a515a02/npMHnVP2hHLbvoUGe7C4O.jpeg",
"followerCount": 2,
"fullname": "Zekun Qi",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "qizekun",
"type": "user"
} | true | null | 2502.13143 | [
{
"_id": "67b546c0d8a1eac02c605f6a",
"hidden": false,
"name": "Zekun Qi",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:21.001Z",
"user": {
"_id": "63c3e8abc7d7f4c63a515a02",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63c3e8abc7d7f4c63a515a02/npMHnVP2hHLbvoUGe7C4O.jpeg",
"fullname": "Zekun Qi",
"isPro": false,
"type": "user",
"user": "qizekun"
}
},
{
"_id": "67b546c0d8a1eac02c605f6b",
"hidden": false,
"name": "Wenyao Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:08:31.789Z",
"user": {
"_id": "65f9533b136fb8ddbd14e1fa",
"avatarUrl": "/avatars/d88f75da0448093ccd1babba2a37d73f.svg",
"fullname": "Zhang",
"isPro": false,
"type": "user",
"user": "WenyaoZhang"
}
},
{
"_id": "67b546c0d8a1eac02c605f6c",
"hidden": false,
"name": "Yufei Ding",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:08:57.294Z",
"user": {
"_id": "66bde456198f9d79f2be2d17",
"avatarUrl": "/avatars/8c349aecb8a3a7cd7ef9d69e94eca8bd.svg",
"fullname": "Yufei Ding",
"isPro": false,
"type": "user",
"user": "YufeiD"
}
},
{
"_id": "67b546c0d8a1eac02c605f6d",
"hidden": false,
"name": "Runpei Dong",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:01:18.622Z",
"user": {
"_id": "6201fc5d91d53938a6432fbf",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6201fc5d91d53938a6432fbf/VLs8ZYaZrop4KBpZn53fH.jpeg",
"fullname": "Runpei Dong",
"isPro": false,
"type": "user",
"user": "RunpeiDong"
}
},
{
"_id": "67b546c0d8a1eac02c605f6e",
"hidden": false,
"name": "Xinqiang Yu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:27.869Z",
"user": {
"_id": "675296fdea02ef7b84609893",
"avatarUrl": "/avatars/27b3886ca048f108ac26a942f151410b.svg",
"fullname": "Yu",
"isPro": false,
"type": "user",
"user": "XinXinQiang"
}
},
{
"_id": "67b546c0d8a1eac02c605f6f",
"hidden": false,
"name": "Jingwen Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b546c0d8a1eac02c605f70",
"hidden": false,
"name": "Lingyun Xu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:09:41.570Z",
"user": {
"_id": "6745c4a80739b09408862ac9",
"avatarUrl": "/avatars/8bc082fcbbf933b150a252c78d1bb3be.svg",
"fullname": "lingyun xu",
"isPro": false,
"type": "user",
"user": "codered010"
}
},
{
"_id": "67b546c0d8a1eac02c605f71",
"hidden": false,
"name": "Baoyu Li",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:09:48.223Z",
"user": {
"_id": "67302fa362930cbc461511a8",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/L9x68oxL8Ot88pyeU6w7Z.png",
"fullname": "Baoyu Li",
"isPro": false,
"type": "user",
"user": "boeyyyy"
}
},
{
"_id": "67b546c0d8a1eac02c605f72",
"hidden": false,
"name": "Xialin He",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:10:04.207Z",
"user": {
"_id": "67b42039cca7aff798979e80",
"avatarUrl": "/avatars/d410b3617395cd7e2a9c0c89ff12f23d.svg",
"fullname": "Xialin He",
"isPro": false,
"type": "user",
"user": "XialinHe"
}
},
{
"_id": "67b546c0d8a1eac02c605f73",
"hidden": false,
"name": "Guofan Fan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b546c0d8a1eac02c605f74",
"hidden": false,
"name": "Jiazhao Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:11:41.498Z",
"user": {
"_id": "65658233d35fc55406e8b00d",
"avatarUrl": "/avatars/660eaa1923cd3e3478cec8197936a75c.svg",
"fullname": "Jiazhao Zhang",
"isPro": false,
"type": "user",
"user": "Jzzhang"
}
},
{
"_id": "67b546c0d8a1eac02c605f75",
"hidden": false,
"name": "Jiawei He",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T14:37:41.045Z",
"user": {
"_id": "649cd8deccfa6c1a3c4d05ec",
"avatarUrl": "/avatars/2b8f72a0643dfd74bc08fba5ed98ce95.svg",
"fullname": "Jiawei",
"isPro": false,
"type": "user",
"user": "jiaweihe"
}
},
{
"_id": "67b546c0d8a1eac02c605f76",
"hidden": false,
"name": "Jiayuan Gu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T10:11:49.058Z",
"user": {
"_id": "638aa283cebef0d13aa2ec2e",
"avatarUrl": "/avatars/3b67e6a6073033864a817230e97c27ca.svg",
"fullname": "Jiayuan Gu",
"isPro": false,
"type": "user",
"user": "jigu"
}
},
{
"_id": "67b546c0d8a1eac02c605f77",
"hidden": false,
"name": "Xin Jin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b546c0d8a1eac02c605f78",
"hidden": false,
"name": "Kaisheng Ma",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b546c0d8a1eac02c605f79",
"hidden": false,
"name": "Zhizheng Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b546c0d8a1eac02c605f7a",
"hidden": false,
"name": "He Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b546c0d8a1eac02c605f7b",
"hidden": false,
"name": "Li Yi",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T18:59:02 | SoFar: Language-Grounded Orientation Bridges Spatial Reasoning and
Object Manipulation | Spatial intelligence is a critical component of embodied AI, promoting robots
to understand and interact with their environments. While recent advances have
enhanced the ability of VLMs to perceive object locations and positional
relationships, they still lack the capability to precisely understand object
orientations-a key requirement for tasks involving fine-grained manipulations.
Addressing this limitation not only requires geometric reasoning but also an
expressive and intuitive way to represent orientation. In this context, we
propose that natural language offers a more flexible representation space than
canonical frames, making it particularly suitable for instruction-following
robotic systems. In this paper, we introduce the concept of semantic
orientation, which defines object orientations using natural language in a
reference-frame-free manner (e.g., the ''plug-in'' direction of a USB or the
''handle'' direction of a knife). To support this, we construct OrienText300K,
a large-scale dataset of 3D models annotated with semantic orientations that
link geometric understanding to functional semantics. By integrating semantic
orientation into a VLM system, we enable robots to generate manipulation
actions with both positional and orientational constraints. Extensive
experiments in simulation and real world demonstrate that our approach
significantly enhances robotic manipulation capabilities, e.g., 48.7% accuracy
on Open6DOR and 74.9% accuracy on SIMPLER. | 29 | 67b546c5d8a1eac02c606090 | null | null |
|
2025-02-18T21:18:22.741000 | Sailor2: Sailing in South-East Asia with Inclusive Multilingual LLMs | 4 | {
"_id": "6214e4ee1e35c843d42d1f88",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6214e4ee1e35c843d42d1f88/fj-9wuIdPhvogh3BrcXTB.jpeg",
"followerCount": 15,
"fullname": "Longxu Dou",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "dreamerdeo",
"type": "user"
} | true | null | 2502.12982 | [
{
"_id": "67b53f572b2ec6908ffef365",
"hidden": false,
"name": "Longxu Dou",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-19T02:17:59.980Z",
"user": {
"_id": "6214e4ee1e35c843d42d1f88",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6214e4ee1e35c843d42d1f88/fj-9wuIdPhvogh3BrcXTB.jpeg",
"fullname": "Longxu Dou",
"isPro": true,
"type": "user",
"user": "dreamerdeo"
}
},
{
"_id": "67b53f572b2ec6908ffef366",
"hidden": false,
"name": "Qian Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef367",
"hidden": false,
"name": "Fan Zhou",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:51:54.538Z",
"user": {
"_id": "628f6e5ab90dde28ef57d293",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/628f6e5ab90dde28ef57d293/AxNzR2nvrND6Rf3RPkYMk.jpeg",
"fullname": "Fan Zhou",
"isPro": false,
"type": "user",
"user": "koalazf99"
}
},
{
"_id": "67b53f572b2ec6908ffef368",
"hidden": false,
"name": "Changyu Chen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-26T08:39:02.389Z",
"user": {
"_id": "64e416dc54e18f390ef79ba4",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/5n01J00ZaVRrebsON8iYA.jpeg",
"fullname": "Changyu Chen",
"isPro": true,
"type": "user",
"user": "Cameron-Chen"
}
},
{
"_id": "67b53f572b2ec6908ffef369",
"hidden": false,
"name": "Zili Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef36a",
"hidden": false,
"name": "Ziqi Jin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef36b",
"hidden": false,
"name": "Zichen Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-27T09:17:58.404Z",
"user": {
"_id": "65f5392c68b8e0cb3c9977a2",
"avatarUrl": "/avatars/aa64772475098e8a135c13072fde6744.svg",
"fullname": "Zichen",
"isPro": false,
"type": "user",
"user": "lkevinzc"
}
},
{
"_id": "67b53f572b2ec6908ffef36c",
"hidden": false,
"name": "Tongyao Zhu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef36d",
"hidden": false,
"name": "Cunxiao Du",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef36e",
"hidden": false,
"name": "Penghui Yang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:03:36.624Z",
"user": {
"_id": "6508463c423b46492eec64e2",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6508463c423b46492eec64e2/WSU7NSqjk92Pr2xUIWjCk.png",
"fullname": "Penghui Yang",
"isPro": false,
"type": "user",
"user": "phyang"
}
},
{
"_id": "67b53f572b2ec6908ffef36f",
"hidden": false,
"name": "Haonan Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef370",
"hidden": false,
"name": "Jiaheng Liu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:55.378Z",
"user": {
"_id": "65377c30e48353201e6fdda0",
"avatarUrl": "/avatars/a8f803b6f2e598eaee9c52c0d2ddfc16.svg",
"fullname": "Jiaheng Liu",
"isPro": false,
"type": "user",
"user": "CheeryLJH"
}
},
{
"_id": "67b53f572b2ec6908ffef371",
"hidden": false,
"name": "Yongchi Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef372",
"hidden": false,
"name": "Xiachong Feng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef373",
"hidden": false,
"name": "Xin Mao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef374",
"hidden": false,
"name": "Man Tsung Yeung",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef375",
"hidden": false,
"name": "Kunat Pipatanakul",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef376",
"hidden": false,
"name": "Fajri Koto",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef377",
"hidden": false,
"name": "Min Si Thu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T14:37:45.250Z",
"user": {
"_id": "63ff6038e7767a895335bd48",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1678198534680-63ff6038e7767a895335bd48.jpeg",
"fullname": "Min Si Thu",
"isPro": false,
"type": "user",
"user": "jojo-ai-mst"
}
},
{
"_id": "67b53f572b2ec6908ffef378",
"hidden": false,
"name": "Hynek Kydlíček",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef379",
"hidden": false,
"name": "Zeyi Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef37a",
"hidden": false,
"name": "Qunshu Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef37b",
"hidden": false,
"name": "Sittipong Sripaisarnmongkol",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef37c",
"hidden": false,
"name": "Kridtaphad Sae-Khow",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef37d",
"hidden": false,
"name": "Nirattisai Thongchim",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef37e",
"hidden": false,
"name": "Taechawat Konkaew",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef37f",
"hidden": false,
"name": "Narong Borijindargoon",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef380",
"hidden": false,
"name": "Anh Dao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef381",
"hidden": false,
"name": "Matichon Maneegard",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef382",
"hidden": false,
"name": "Phakphum Artkaew",
"status": "claimed_verified",
"statusLastChangedAt": "2025-03-03T08:07:36.639Z",
"user": {
"_id": "631a4855300a072a8da70abd",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/631a4855300a072a8da70abd/jRnzdW5JBjICYKCmkUFI-.jpeg",
"fullname": "phakphum artkaew",
"isPro": false,
"type": "user",
"user": "pakphum"
}
},
{
"_id": "67b53f572b2ec6908ffef383",
"hidden": false,
"name": "Zheng-Xin Yong",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:30.404Z",
"user": {
"_id": "61424bf4f0d914a5f606a823",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/61424bf4f0d914a5f606a823/0td8lR4elBaVvJUD9Pojh.png",
"fullname": "Yong Zheng-Xin",
"isPro": false,
"type": "user",
"user": "yongzx"
}
},
{
"_id": "67b53f572b2ec6908ffef384",
"hidden": false,
"name": "Quan Nguyen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef385",
"hidden": false,
"name": "Wannaphong Phatthiyaphaibun",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T14:37:47.061Z",
"user": {
"_id": "60dc25da6155a8319f008a6f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1630322686754-60dc25da6155a8319f008a6f.jpeg",
"fullname": "Wannaphong Phatthiyaphaibun",
"isPro": false,
"type": "user",
"user": "wannaphong"
}
},
{
"_id": "67b53f572b2ec6908ffef386",
"hidden": false,
"name": "Hoang H. Tran",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-24T09:25:29.514Z",
"user": {
"_id": "65c19c3172fd754bba256112",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/65c19c3172fd754bba256112/cfgYvKfO36PSTFYKQottM.jpeg",
"fullname": "Ryan Tran",
"isPro": false,
"type": "user",
"user": "ryanhoangt"
}
},
{
"_id": "67b53f572b2ec6908ffef387",
"hidden": false,
"name": "Mike Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:03:21.404Z",
"user": {
"_id": "60d33fbbd7b174177faabd4f",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/60d33fbbd7b174177faabd4f/pfyv_xj2B2m2N4F4sT9zJ.jpeg",
"fullname": "Mike Zhang",
"isPro": true,
"type": "user",
"user": "jjzha"
}
},
{
"_id": "67b53f572b2ec6908ffef388",
"hidden": false,
"name": "Shiqi Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef389",
"hidden": false,
"name": "Tianyu Pang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef38a",
"hidden": false,
"name": "Chao Du",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef38b",
"hidden": false,
"name": "Xinyi Wan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef38c",
"hidden": false,
"name": "Wei Lu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b53f572b2ec6908ffef38d",
"hidden": false,
"name": "Min Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-18T16:04:57 | Sailor2: Sailing in South-East Asia with Inclusive Multilingual LLMs | Sailor2 is a family of cutting-edge multilingual language models for
South-East Asian (SEA) languages, available in 1B, 8B, and 20B sizes to suit
diverse applications. Building on Qwen2.5, Sailor2 undergoes continuous
pre-training on 500B tokens (400B SEA-specific and 100B replay tokens) to
support 13 SEA languages while retaining proficiency in Chinese and English.
Sailor2-20B model achieves a 50-50 win rate against GPT-4o across SEA
languages. We also deliver a comprehensive cookbook on how to develop the
multilingual model in an efficient manner, including five key aspects: data
curation, pre-training, post-training, model customization and evaluation. We
hope that Sailor2 model (Apache 2.0 license) will drive language development in
the SEA region, and Sailor2 cookbook will inspire researchers to build more
inclusive LLMs for other under-served languages. | 14 | 67b53f572b2ec6908ffef3c9 | null | null |
|
2025-02-18T20:05:09.186000 | ExaGPT: Example-Based Machine-Generated Text Detection for Human Interpretability | 2 | {
"_id": "6538e649f940c8a0358aa8b8",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538e649f940c8a0358aa8b8/veNw6QJuZu8anWCXtOXxu.jpeg",
"followerCount": null,
"fullname": "Ryuto Koike",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "ryuryukke",
"type": "user"
} | false | [
"https://cdn-uploads.huggingface.co/production/uploads/6538e649f940c8a0358aa8b8/LTS6uI3uy5AxEeoD9-oMX.png"
]
| 2502.11336 | [
{
"_id": "67b52de36007d463b988b202",
"hidden": false,
"name": "Ryuto Koike",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:03:41.013Z",
"user": {
"_id": "6538e649f940c8a0358aa8b8",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538e649f940c8a0358aa8b8/veNw6QJuZu8anWCXtOXxu.jpeg",
"fullname": "Ryuto Koike",
"isPro": false,
"type": "user",
"user": "ryuryukke"
}
},
{
"_id": "67b52de36007d463b988b203",
"hidden": false,
"name": "Masahiro Kaneko",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:46:06.323Z",
"user": {
"_id": "652bdb25756a15d750071787",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/652bdb25756a15d750071787/qWjkU0_1egP7XYX3sXoSZ.jpeg",
"fullname": "Masahiro Kaneko",
"isPro": false,
"type": "user",
"user": "MasahiroKaneko"
}
},
{
"_id": "67b52de36007d463b988b204",
"hidden": false,
"name": "Ayana Niwa",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b52de36007d463b988b205",
"hidden": false,
"name": "Preslav Nakov",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:46:29.303Z",
"user": {
"_id": "647f7eb25e1bc4753746bf9f",
"avatarUrl": "/avatars/cc9c6210fdc822d8a106937e747dff41.svg",
"fullname": "Preslav Nakov",
"isPro": false,
"type": "user",
"user": "preslavnakov"
}
},
{
"_id": "67b52de36007d463b988b206",
"hidden": false,
"name": "Naoaki Okazaki",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:46:36.241Z",
"user": {
"_id": "630c0a9b09eceb8fafe89cc6",
"avatarUrl": "/avatars/6325a9b34da54d5cbddb814c3987a2fe.svg",
"fullname": "Naoaki Okazaki",
"isPro": false,
"type": "user",
"user": "nokazaki"
}
}
]
| 2025-02-17T01:15:07 | ExaGPT: Example-Based Machine-Generated Text Detection for Human
Interpretability | Detecting texts generated by Large Language Models (LLMs) could cause grave
mistakes due to incorrect decisions, such as undermining student's academic
dignity. LLM text detection thus needs to ensure the interpretability of the
decision, which can help users judge how reliably correct its prediction is.
When humans verify whether a text is human-written or LLM-generated, they
intuitively investigate with which of them it shares more similar spans.
However, existing interpretable detectors are not aligned with the human
decision-making process and fail to offer evidence that users easily
understand. To bridge this gap, we introduce ExaGPT, an interpretable detection
approach grounded in the human decision-making process for verifying the origin
of a text. ExaGPT identifies a text by checking whether it shares more similar
spans with human-written vs. with LLM-generated texts from a datastore. This
approach can provide similar span examples that contribute to the decision for
each span in the text as evidence. Our human evaluation demonstrates that
providing similar span examples contributes more effectively to judging the
correctness of the decision than existing interpretable methods. Moreover,
extensive experiments in four domains and three generators show that ExaGPT
massively outperforms prior powerful detectors by up to +40.9 points of
accuracy at a false positive rate of 1%. | 0 | 67b52de46007d463b988b279 | null | null |
|
2025-02-18T18:58:34.838000 | Diffusion Models without Classifier-free Guidance | 2 | {
"_id": "6372f265112fb535baf254c4",
"avatarUrl": "/avatars/9b821bc533175c7dded48cdb3a3e1a12.svg",
"followerCount": 2,
"fullname": "tzco",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "tzco",
"type": "user"
} | true | null | 2502.12154 | [
{
"_id": "67b400719ff3ff79dae14701",
"hidden": false,
"name": "Zhicong Tang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:45.361Z",
"user": {
"_id": "6372f265112fb535baf254c4",
"avatarUrl": "/avatars/9b821bc533175c7dded48cdb3a3e1a12.svg",
"fullname": "tzco",
"isPro": false,
"type": "user",
"user": "tzco"
}
},
{
"_id": "67b400719ff3ff79dae14702",
"hidden": false,
"name": "Jianmin Bao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b400719ff3ff79dae14703",
"hidden": false,
"name": "Dong Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b400719ff3ff79dae14704",
"hidden": false,
"name": "Baining Guo",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T18:59:50 | Diffusion Models without Classifier-free Guidance | This paper presents Model-guidance (MG), a novel objective for training
diffusion model that addresses and removes of the commonly used Classifier-free
guidance (CFG). Our innovative approach transcends the standard modeling of
solely data distribution to incorporating the posterior probability of
conditions. The proposed technique originates from the idea of CFG and is easy
yet effective, making it a plug-and-play module for existing models. Our method
significantly accelerates the training process, doubles the inference speed,
and achieve exceptional quality that parallel and even surpass concurrent
diffusion models with CFG. Extensive experiments demonstrate the effectiveness,
efficiency, scalability on different models and datasets. Finally, we establish
state-of-the-art performance on ImageNet 256 benchmarks with an FID of 1.34.
Our code is available at https://github.com/tzco/Diffusion-wo-CFG. | 4 | 67b400789ff3ff79dae147ee | null | null |
|
2025-02-18T14:56:45.613000 | EQ-VAE: Equivariance Regularized Latent Space for Improved Generative Image Modeling | 2 | {
"_id": "661ba524bd9243bf7e598355",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/661ba524bd9243bf7e598355/i77yD4XgJn2vUbn_mIsT8.jpeg",
"followerCount": 2,
"fullname": "Ioannis Kakogeorgiou",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "gkakogeorgiou",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/661ba524bd9243bf7e598355/9XkVow22TY84dDgXm-Duc.gif"
]
| 2502.09509 | [
{
"_id": "67b4e4259beded220ad14729",
"hidden": false,
"name": "Theodoros Kouzelis",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T16:15:22.578Z",
"user": {
"_id": "6383aa17834d3558a3955186",
"avatarUrl": "/avatars/1f6aed0a762379df334bc6a734d42f86.svg",
"fullname": "Kouzelis",
"isPro": false,
"type": "user",
"user": "zelaki"
}
},
{
"_id": "67b4e4259beded220ad1472a",
"hidden": false,
"name": "Ioannis Kakogeorgiou",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:03:49.086Z",
"user": {
"_id": "661ba524bd9243bf7e598355",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/661ba524bd9243bf7e598355/i77yD4XgJn2vUbn_mIsT8.jpeg",
"fullname": "Ioannis Kakogeorgiou",
"isPro": false,
"type": "user",
"user": "gkakogeorgiou"
}
},
{
"_id": "67b4e4259beded220ad1472b",
"hidden": false,
"name": "Spyros Gidaris",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4e4259beded220ad1472c",
"hidden": false,
"name": "Nikos Komodakis",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-13T17:21:51 | EQ-VAE: Equivariance Regularized Latent Space for Improved Generative
Image Modeling | Latent generative models have emerged as a leading approach for high-quality
image synthesis. These models rely on an autoencoder to compress images into a
latent space, followed by a generative model to learn the latent distribution.
We identify that existing autoencoders lack equivariance to semantic-preserving
transformations like scaling and rotation, resulting in complex latent spaces
that hinder generative performance. To address this, we propose EQ-VAE, a
simple regularization approach that enforces equivariance in the latent space,
reducing its complexity without degrading reconstruction quality. By finetuning
pre-trained autoencoders with EQ-VAE, we enhance the performance of several
state-of-the-art generative models, including DiT, SiT, REPA and MaskGIT,
achieving a 7 speedup on DiT-XL/2 with only five epochs of SD-VAE fine-tuning.
EQ-VAE is compatible with both continuous and discrete autoencoders, thus
offering a versatile enhancement for a wide range of latent generative models.
Project page and code: https://eq-vae.github.io/. | 7 | 67b4e4289beded220ad147c7 | null | null |
|
2025-02-18T13:59:31.380000 | Ask in Any Modality: A Comprehensive Survey on Multimodal Retrieval-Augmented Generation | 2 | {
"_id": "64ba58d377dd483716aba098",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64ba58d377dd483716aba098/6VASAUkFpDC-PR01yUJWj.png",
"followerCount": 3,
"fullname": "Mahdi Abootorabi",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "aboots",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/64ba58d377dd483716aba098/N0fZ0I60EfZjITEnf6gPc.png",
"https://cdn-uploads.huggingface.co/production/uploads/64ba58d377dd483716aba098/CtLxMqUEhWr6d9ztU1YZq.jpeg",
"https://cdn-uploads.huggingface.co/production/uploads/64ba58d377dd483716aba098/HczPPOjzArOwgdwb5yv5Z.jpeg"
]
| 2502.08826 | [
{
"_id": "67b303f18bd6e9a5cad8bc4d",
"hidden": false,
"name": "Mohammad Mahdi Abootorabi",
"status": "extracted_confirmed",
"statusLastChangedAt": "2025-02-17T09:40:27.588Z",
"user": {
"_id": "64ba58d377dd483716aba098",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64ba58d377dd483716aba098/6VASAUkFpDC-PR01yUJWj.png",
"fullname": "Mahdi Abootorabi",
"isPro": false,
"type": "user",
"user": "aboots"
}
},
{
"_id": "67b303f18bd6e9a5cad8bc4e",
"hidden": false,
"name": "Amirhosein Zobeiri",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T10:49:51.840Z",
"user": {
"_id": "659d22a87ca51193252ec403",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/659d22a87ca51193252ec403/4r6RO3enW8q6gv_eWFZhl.jpeg",
"fullname": "Amirhosein Zobeiri",
"isPro": false,
"type": "user",
"user": "ZobeiriA"
}
},
{
"_id": "67b303f18bd6e9a5cad8bc4f",
"hidden": false,
"name": "Mahdi Dehghani",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T09:37:45.858Z",
"user": {
"_id": "67b61b5692649eb787279693",
"avatarUrl": "/avatars/1d3b1f4d48ec8064a25ae5f27ea58576.svg",
"fullname": "Mahdi Dehghani",
"isPro": false,
"type": "user",
"user": "Mahdi-dh"
}
},
{
"_id": "67b303f18bd6e9a5cad8bc50",
"hidden": false,
"name": "Mohammadali Mohammadkhani",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:55.291Z",
"user": {
"_id": "64a43776f489f7ced54a4c4b",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64a43776f489f7ced54a4c4b/zO36YPuyaSoTbug99PxYs.jpeg",
"fullname": "Mohammadali Mohammadkhani",
"isPro": false,
"type": "user",
"user": "moali-mkh-2000"
}
},
{
"_id": "67b303f18bd6e9a5cad8bc51",
"hidden": false,
"name": "Bardia Mohammadi",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T14:37:52.589Z",
"user": {
"_id": "6531334db5672703b2efcf6f",
"avatarUrl": "/avatars/598d6b4922d5f7d434ed1cf9ffc94f6d.svg",
"fullname": "Bardia Mohammadi",
"isPro": false,
"type": "user",
"user": "bardia79mhd"
}
},
{
"_id": "67b303f18bd6e9a5cad8bc52",
"hidden": false,
"name": "Omid Ghahroodi",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:33:28.667Z",
"user": {
"_id": "6250819a5bf543dbd2607081",
"avatarUrl": "/avatars/dfade22a578b8349ab544acda5e8bcad.svg",
"fullname": "Omid Ghahroodi",
"isPro": false,
"type": "user",
"user": "omidgh"
}
},
{
"_id": "67b303f18bd6e9a5cad8bc53",
"hidden": false,
"name": "Mahdieh Soleymani Baghshah",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-17T09:40:02.049Z",
"user": {
"_id": "661a88f0bcd78151e521bc60",
"avatarUrl": "/avatars/bedab01ce7909ecde7a60f891770c18c.svg",
"fullname": "Mahdieh Soleymani Baghshah",
"isPro": false,
"type": "user",
"user": "Soleymani"
}
},
{
"_id": "67b303f18bd6e9a5cad8bc54",
"hidden": false,
"name": "Ehsaneddin Asgari",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-12T22:33:41 | Ask in Any Modality: A Comprehensive Survey on Multimodal
Retrieval-Augmented Generation | Large Language Models (LLMs) struggle with hallucinations and outdated
knowledge due to their reliance on static training data. Retrieval-Augmented
Generation (RAG) mitigates these issues by integrating external dynamic
information enhancing factual and updated grounding. Recent advances in
multimodal learning have led to the development of Multimodal RAG,
incorporating multiple modalities such as text, images, audio, and video to
enhance the generated outputs. However, cross-modal alignment and reasoning
introduce unique challenges to Multimodal RAG, distinguishing it from
traditional unimodal RAG. This survey offers a structured and comprehensive
analysis of Multimodal RAG systems, covering datasets, metrics, benchmarks,
evaluation, methodologies, and innovations in retrieval, fusion, augmentation,
and generation. We precisely review training strategies, robustness
enhancements, and loss functions, while also exploring the diverse Multimodal
RAG scenarios. Furthermore, we discuss open challenges and future research
directions to support advancements in this evolving field. This survey lays the
foundation for developing more capable and reliable AI systems that effectively
leverage multimodal dynamic external knowledge bases. Resources are available
at https://github.com/llm-lab-org/Multimodal-RAG-Survey. | 17 | 67b303f28bd6e9a5cad8bc85 | null | null |
|
2025-02-18T13:21:05.722000 | IHEval: Evaluating Language Models on Following the Instruction Hierarchy | 2 | {
"_id": "63bf9695da08ed054400205e",
"avatarUrl": "/avatars/b6fca49559a61cf66628088c60d26c10.svg",
"followerCount": 1,
"fullname": "Zhihan Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "zhihz0535",
"type": "user"
} | true | null | 2502.08745 | [
{
"_id": "67b4cf1994ec5e365fb7995d",
"hidden": false,
"name": "Zhihan Zhang",
"status": "extracted_confirmed",
"statusLastChangedAt": "2025-02-18T18:19:31.455Z",
"user": {
"_id": "63bf9695da08ed054400205e",
"avatarUrl": "/avatars/b6fca49559a61cf66628088c60d26c10.svg",
"fullname": "Zhihan Zhang",
"isPro": false,
"type": "user",
"user": "zhihz0535"
}
},
{
"_id": "67b4cf1994ec5e365fb7995e",
"hidden": false,
"name": "Shiyang Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb7995f",
"hidden": false,
"name": "Zixuan Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb79960",
"hidden": false,
"name": "Xin Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb79961",
"hidden": false,
"name": "Haoming Jiang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb79962",
"hidden": false,
"name": "Xianfeng Tang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:25:56.312Z",
"user": {
"_id": "6465f6467ff8fcbef7d22513",
"avatarUrl": "/avatars/07992835c235fbb07016a0ea4f1d61cb.svg",
"fullname": "Xianfeng Tang",
"isPro": false,
"type": "user",
"user": "xianft"
}
},
{
"_id": "67b4cf1994ec5e365fb79963",
"hidden": false,
"name": "Yifan Gao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb79964",
"hidden": false,
"name": "Zheng Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb79965",
"hidden": false,
"name": "Haodong Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb79966",
"hidden": false,
"name": "Zhaoxuan Tan",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:26:12.966Z",
"user": {
"_id": "638eaca4a26f3d5af37be8b3",
"avatarUrl": "/avatars/476890d83191d0cbdb9a3d5351a129da.svg",
"fullname": "Zhaoxuan_Tan",
"isPro": false,
"type": "user",
"user": "Zhaoxuan"
}
},
{
"_id": "67b4cf1994ec5e365fb79967",
"hidden": false,
"name": "Yichuan Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb79968",
"hidden": false,
"name": "Qingyu Yin",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:26:54.938Z",
"user": {
"_id": "6453cb22908e259483c0a061",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6453cb22908e259483c0a061/hMgdwZUsUbgquGalzPGzV.jpeg",
"fullname": "Qingyu_Yin",
"isPro": false,
"type": "user",
"user": "MikaStars39"
}
},
{
"_id": "67b4cf1994ec5e365fb79969",
"hidden": false,
"name": "Bing Yin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4cf1994ec5e365fb7996a",
"hidden": false,
"name": "Meng Jiang",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-12T19:35:28 | IHEval: Evaluating Language Models on Following the Instruction
Hierarchy | The instruction hierarchy, which establishes a priority order from system
messages to user messages, conversation history, and tool outputs, is essential
for ensuring consistent and safe behavior in language models (LMs). Despite its
importance, this topic receives limited attention, and there is a lack of
comprehensive benchmarks for evaluating models' ability to follow the
instruction hierarchy. We bridge this gap by introducing IHEval, a novel
benchmark comprising 3,538 examples across nine tasks, covering cases where
instructions in different priorities either align or conflict. Our evaluation
of popular LMs highlights their struggle to recognize instruction priorities.
All evaluated models experience a sharp performance decline when facing
conflicting instructions, compared to their original instruction-following
performance. Moreover, the most competitive open-source model only achieves 48%
accuracy in resolving such conflicts. Our results underscore the need for
targeted optimization in the future development of LMs. | 18 | 67b4cf1a94ec5e365fb799c1 | null | null |
|
2025-02-18T13:04:04.423000 | Data Valuation using Neural Networks for Efficient Instruction Fine-Tuning | 2 | {
"_id": "6391e4e984afa726d66180b9",
"avatarUrl": "/avatars/e437e2820745b522a868b8da27d9a11f.svg",
"followerCount": 0,
"fullname": "Ishika Agarwal",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "ishikaa",
"type": "user"
} | true | null | 2502.09969 | [
{
"_id": "67b4cb6c777b7676c8b3c43d",
"hidden": false,
"name": "Ishika Agarwal",
"status": "extracted_confirmed",
"statusLastChangedAt": "2025-02-18T18:06:42.786Z",
"user": {
"_id": "6391e4e984afa726d66180b9",
"avatarUrl": "/avatars/e437e2820745b522a868b8da27d9a11f.svg",
"fullname": "Ishika Agarwal",
"isPro": false,
"type": "user",
"user": "ishikaa"
}
},
{
"_id": "67b4cb6c777b7676c8b3c43e",
"hidden": false,
"name": "Dilek Hakkani-Tür",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-14T07:55:47 | Data Valuation using Neural Networks for Efficient Instruction
Fine-Tuning | Influence functions provide crucial insights into model training, but
existing methods suffer from large computational costs and limited
generalization. Particularly, recent works have proposed various metrics and
algorithms to calculate the influence of data using language models, which do
not scale well with large models and datasets. This is because of the expensive
forward and backward passes required for computation, substantial memory
requirements to store large models, and poor generalization of influence
estimates to new data. In this paper, we explore the use of small neural
networks -- which we refer to as the InfluenceNetwork -- to estimate influence
values, achieving up to 99% cost reduction. Our evaluation demonstrates that
influence values can be estimated with models just 0.0027% the size of full
language models (we use 7B and 8B versions). We apply our algorithm of
estimating influence values (called NN-CIFT: Neural Networks for effiCient
Instruction Fine-Tuning) to the downstream task of subset selection for general
instruction fine-tuning. In our study, we include four state-of-the-art
influence functions and show no compromise in performance, despite large
speedups, between NN-CIFT and the original influence functions. We provide an
in-depth hyperparameter analyses of NN-CIFT. The code for our method can be
found here: https://github.com/agarwalishika/NN-CIFT. | 1 | 67b4cb6d777b7676c8b3c45c | null | null |
|
2025-02-18T11:57:43.538000 | Explorer: Scaling Exploration-driven Web Trajectory Synthesis for Multimodal Web Agents | 2 | {
"_id": "6556717676fe5cfa6a115405",
"avatarUrl": "/avatars/570dd8f4eb6baaff12d7ebe11dde6348.svg",
"followerCount": 1,
"fullname": "Vardaan Pahuja",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "vardaan123",
"type": "user"
} | true | null | 2502.11357 | [
{
"_id": "67b3f1f1f5bd60d66133e1f3",
"hidden": false,
"name": "Vardaan Pahuja",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:47.969Z",
"user": {
"_id": "6556717676fe5cfa6a115405",
"avatarUrl": "/avatars/570dd8f4eb6baaff12d7ebe11dde6348.svg",
"fullname": "Vardaan Pahuja",
"isPro": false,
"type": "user",
"user": "vardaan123"
}
},
{
"_id": "67b3f1f1f5bd60d66133e1f4",
"hidden": false,
"name": "Yadong Lu",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-18T02:35:29.988Z",
"user": {
"_id": "664bbd75a6bd1b3d2ac7fc34",
"avatarUrl": "/avatars/127bf5d611b46ef95a1859a8cf21a160.svg",
"fullname": "Yadong Lu",
"isPro": false,
"type": "user",
"user": "adamlu1"
}
},
{
"_id": "67b3f1f1f5bd60d66133e1f5",
"hidden": false,
"name": "Corby Rosset",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b3f1f1f5bd60d66133e1f6",
"hidden": false,
"name": "Boyu Gou",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-20T17:33:00.709Z",
"user": {
"_id": "6500870f1e14749e84f8f887",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6500870f1e14749e84f8f887/wfvx4BZvh2OyW-vpq5jEy.jpeg",
"fullname": "Boyu Gou",
"isPro": false,
"type": "user",
"user": "BoyuNLP"
}
},
{
"_id": "67b3f1f1f5bd60d66133e1f7",
"hidden": false,
"name": "Arindam Mitra",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b3f1f1f5bd60d66133e1f8",
"hidden": false,
"name": "Spencer Whitehead",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b3f1f1f5bd60d66133e1f9",
"hidden": false,
"name": "Yu Su",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b3f1f1f5bd60d66133e1fa",
"hidden": false,
"name": "Ahmed Awadallah",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T02:13:48 | Explorer: Scaling Exploration-driven Web Trajectory Synthesis for
Multimodal Web Agents | Recent success in large multimodal models (LMMs) has sparked promising
applications of agents capable of autonomously completing complex web tasks.
While open-source LMM agents have made significant advances in offline
evaluation benchmarks, their performance still falls substantially short of
human-level capabilities in more realistic online settings. A key bottleneck is
the lack of diverse and large-scale trajectory-level datasets across various
domains, which are expensive to collect. In this paper, we address this
challenge by developing a scalable recipe to synthesize the largest and most
diverse trajectory-level dataset to date, containing over 94K successful
multimodal web trajectories, spanning 49K unique URLs, 720K screenshots, and
33M web elements. In particular, we leverage extensive web exploration and
refinement to obtain diverse task intents. The average cost is 28 cents per
successful trajectory, making it affordable to a wide range of users in the
community. Leveraging this dataset, we train Explorer, a multimodal web agent,
and demonstrate strong performance on both offline and online web agent
benchmarks such as Mind2Web-Live, Multimodal-Mind2Web, and MiniWob++.
Additionally, our experiments highlight data scaling as a key driver for
improving web agent capabilities. We hope this study makes state-of-the-art
LMM-based agent research at a larger scale more accessible. | 9 | 67b3f1f1f5bd60d66133e24b | null | null |
|
2025-02-18T11:42:58.976000 | ILIAS: Instance-Level Image retrieval At Scale | 2 | {
"_id": "66a3ae59f33ff23e1c027ccd",
"avatarUrl": "/avatars/216717d547bf785a2b1696171e5f4b11.svg",
"followerCount": 1,
"fullname": "Vladan Stojnic",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "stojnvla",
"type": "user"
} | true | null | 2502.11748 | [
{
"_id": "67b465600e5142133055d7c1",
"hidden": false,
"name": "Giorgos Kordopatis-Zilos",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:38.791Z",
"user": {
"_id": "673c934bdf13003bd11746fd",
"avatarUrl": "/avatars/1aec1157549be85963b39eb54845b695.svg",
"fullname": "Giorgos Kordopatis-Zilos",
"isPro": false,
"type": "user",
"user": "gkordo"
}
},
{
"_id": "67b465600e5142133055d7c2",
"hidden": false,
"name": "Vladan Stojnić",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T16:39:41.653Z",
"user": {
"_id": "66a3ae59f33ff23e1c027ccd",
"avatarUrl": "/avatars/216717d547bf785a2b1696171e5f4b11.svg",
"fullname": "Vladan Stojnic",
"isPro": false,
"type": "user",
"user": "stojnvla"
}
},
{
"_id": "67b465600e5142133055d7c3",
"hidden": false,
"name": "Anna Manko",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b465600e5142133055d7c4",
"hidden": false,
"name": "Pavel Šuma",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:41.052Z",
"user": {
"_id": "67ab9acd412358244419d946",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/tq-eUQMYGdlDZVn15DckY.png",
"fullname": "Pavel Suma",
"isPro": false,
"type": "user",
"user": "pavelsuma"
}
},
{
"_id": "67b465600e5142133055d7c5",
"hidden": false,
"name": "Nikolaos-Antonios Ypsilantis",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b465600e5142133055d7c6",
"hidden": false,
"name": "Nikos Efthymiadis",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T11:49:36.494Z",
"user": {
"_id": "663e63c5902b965ba35a0308",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/663e63c5902b965ba35a0308/OFX5yvY2LL8KORKsiDhAy.jpeg",
"fullname": "Nikos Efthymiadis",
"isPro": false,
"type": "user",
"user": "nikos-efth"
}
},
{
"_id": "67b465600e5142133055d7c7",
"hidden": false,
"name": "Zakaria Laskar",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b465600e5142133055d7c8",
"hidden": false,
"name": "Jiří Matas",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b465600e5142133055d7c9",
"hidden": false,
"name": "Ondřej Chum",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b465600e5142133055d7ca",
"hidden": false,
"name": "Giorgos Tolias",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T12:42:38 | ILIAS: Instance-Level Image retrieval At Scale | This work introduces ILIAS, a new test dataset for Instance-Level Image
retrieval At Scale. It is designed to evaluate the ability of current and
future foundation models and retrieval techniques to recognize particular
objects. The key benefits over existing datasets include large scale, domain
diversity, accurate ground truth, and a performance that is far from saturated.
ILIAS includes query and positive images for 1,000 object instances, manually
collected to capture challenging conditions and diverse domains. Large-scale
retrieval is conducted against 100 million distractor images from YFCC100M. To
avoid false negatives without extra annotation effort, we include only query
objects confirmed to have emerged after 2014, i.e. the compilation date of
YFCC100M. An extensive benchmarking is performed with the following
observations: i) models fine-tuned on specific domains, such as landmarks or
products, excel in that domain but fail on ILIAS ii) learning a linear
adaptation layer using multi-domain class supervision results in performance
improvements, especially for vision-language models iii) local descriptors in
retrieval re-ranking are still a key ingredient, especially in the presence of
severe background clutter iv) the text-to-image performance of the
vision-language foundation models is surprisingly close to the corresponding
image-to-image case. website: https://vrg.fel.cvut.cz/ilias/ | 4 | 67b465680e5142133055d97d | null | null |
|
2025-02-18T08:59:34.204000 | Can a Single Model Master Both Multi-turn Conversations and Tool Use? CALM: A Unified Conversational Agentic Language Model | 2 | {
"_id": "63888d3fd68e37abd599f428",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63888d3fd68e37abd599f428/YaNyxG_oM6IgrHTkFZ6Eq.jpeg",
"followerCount": 12,
"fullname": "emre can",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "emrecanacikgoz",
"type": "user"
} | true | null | 2502.08820 | [
{
"_id": "67aece59f2e8a2ee35b5affd",
"hidden": false,
"name": "Emre Can Acikgoz",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:34:01.421Z",
"user": {
"_id": "63888d3fd68e37abd599f428",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/63888d3fd68e37abd599f428/YaNyxG_oM6IgrHTkFZ6Eq.jpeg",
"fullname": "emre can",
"isPro": true,
"type": "user",
"user": "emrecanacikgoz"
}
},
{
"_id": "67aece59f2e8a2ee35b5affe",
"hidden": false,
"name": "Jeremiah Greer",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67aece59f2e8a2ee35b5afff",
"hidden": false,
"name": "Akul Datta",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67aece59f2e8a2ee35b5b000",
"hidden": false,
"name": "Ze Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67aece59f2e8a2ee35b5b001",
"hidden": false,
"name": "William Zeng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67aece59f2e8a2ee35b5b002",
"hidden": false,
"name": "Oussama Elachqar",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67aece59f2e8a2ee35b5b003",
"hidden": false,
"name": "Emmanouil Koukoumidis",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67aece59f2e8a2ee35b5b004",
"hidden": false,
"name": "Dilek Hakkani-Tür",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67aece59f2e8a2ee35b5b005",
"hidden": false,
"name": "Gokhan Tur",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-12T22:18:34 | Can a Single Model Master Both Multi-turn Conversations and Tool Use?
CALM: A Unified Conversational Agentic Language Model | Large Language Models (LLMs) with API-calling capabilities enabled building
effective Language Agents (LA), while also revolutionizing the conventional
task-oriented dialogue (TOD) paradigm. However, current approaches face a
critical dilemma: TOD systems are often trained on a limited set of target
APIs, requiring new data to maintain their quality when interfacing with new
services, while LAs are not trained to maintain user intent over multi-turn
conversations. Because both robust multi-turn management and advanced function
calling are crucial for effective conversational agents, we evaluate these
skills on three popular benchmarks: MultiWOZ 2.4 (TOD), BFCL V3 (LA), and
API-Bank (LA), and our analyses reveal that specialized approaches excel in one
domain but underperform in the other. To bridge this chasm, we introduce CALM
(Conversational Agentic Language Model), a unified approach that integrates
both conversational and agentic capabilities. We created CALM-IT, a carefully
constructed multi-task dataset that interleave multi-turn ReAct reasoning with
complex API usage. Using CALM-IT, we train three models CALM 8B, CALM 70B, and
CALM 405B, which outperform top domain-specific models, including GPT-4o,
across all three benchmarks. | 4 | 67aece5af2e8a2ee35b5b03e | null | null |
|
2025-02-18T07:33:17.294000 | The Mirage of Model Editing: Revisiting Evaluation in the Wild | 2 | {
"_id": "64e4090f222b232f03fe5f63",
"avatarUrl": "/avatars/1e97328de374d726f64bf16528d36ca4.svg",
"followerCount": null,
"fullname": "Wanli Yang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "WenDingY",
"type": "user"
} | false | null | 2502.11177 | [
{
"_id": "67b47dd2e638b35196b8e014",
"hidden": false,
"name": "Wanli Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b47dd2e638b35196b8e015",
"hidden": false,
"name": "Fei Sun",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b47dd2e638b35196b8e016",
"hidden": false,
"name": "Jiajun Tan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b47dd2e638b35196b8e017",
"hidden": false,
"name": "Xinyu Ma",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:36.300Z",
"user": {
"_id": "62cd2f13979d883655cd5377",
"avatarUrl": "/avatars/400c252d20d68aca56e0d0280498ce17.svg",
"fullname": "Xinyu Ma",
"isPro": false,
"type": "user",
"user": "xyma"
}
},
{
"_id": "67b47dd2e638b35196b8e018",
"hidden": false,
"name": "Qi Cao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b47dd2e638b35196b8e019",
"hidden": false,
"name": "Dawei Yin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b47dd2e638b35196b8e01a",
"hidden": false,
"name": "Huawei Shen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b47dd2e638b35196b8e01b",
"hidden": false,
"name": "Xueqi Cheng",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-16T15:57:55 | The Mirage of Model Editing: Revisiting Evaluation in the Wild | Despite near-perfect results in artificial evaluations, the effectiveness of
model editing in real-world applications remains unexplored. To bridge this
gap, we propose to study model editing in question answering (QA) by
establishing a rigorous evaluation practice to assess the effectiveness of
editing methods in correcting LLMs' errors. It consists of QAEdit, a new
benchmark derived from popular QA datasets, and a standardized evaluation
framework. Our single editing experiments indicate that current editing methods
perform substantially worse than previously reported (38.5% vs. ~96%). Through
module analysis and controlled experiments, we demonstrate that this
performance decline stems from issues in evaluation practices of prior editing
research. One key issue is the inappropriate use of teacher forcing in testing
prevents error propagation by feeding ground truth tokens (inaccessible in
real-world scenarios) as input. Furthermore, we simulate real-world deployment
by sequential editing, revealing that current approaches fail drastically with
only 1000 edits. Our analysis provides a fundamental reexamination of both the
real-world applicability of existing model editing methods and their evaluation
practices, and establishes a rigorous evaluation framework with key insights to
advance reliable and practical model editing research. | 10 | 67b47dd2e638b35196b8e03a | null | null |
|
2025-02-18T07:16:07.632000 | Memory, Benchmark & Robots: A Benchmark for Solving Complex Tasks with Reinforcement Learning | 2 | {
"_id": "6668687caee0993c95b0eb81",
"avatarUrl": "/avatars/301fe1f395e0a129b1c9785868fa9858.svg",
"followerCount": 2,
"fullname": "Egor Cherepanov",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "avanturist",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/6668687caee0993c95b0eb81/zl6FgeOWq-7PC7PRLEyzW.qt"
]
| 2502.10550 | [
{
"_id": "67b478517fa6ecaa21d1498d",
"hidden": false,
"name": "Egor Cherepanov",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T16:39:34.993Z",
"user": {
"_id": "6668687caee0993c95b0eb81",
"avatarUrl": "/avatars/301fe1f395e0a129b1c9785868fa9858.svg",
"fullname": "Egor Cherepanov",
"isPro": false,
"type": "user",
"user": "avanturist"
}
},
{
"_id": "67b478517fa6ecaa21d1498e",
"hidden": false,
"name": "Nikita Kachaev",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b478517fa6ecaa21d1498f",
"hidden": false,
"name": "Alexey K. Kovalev",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b478517fa6ecaa21d14990",
"hidden": false,
"name": "Aleksandr I. Panov",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-14T20:46:19 | Memory, Benchmark & Robots: A Benchmark for Solving Complex Tasks with
Reinforcement Learning | Memory is crucial for enabling agents to tackle complex tasks with temporal
and spatial dependencies. While many reinforcement learning (RL) algorithms
incorporate memory, the field lacks a universal benchmark to assess an agent's
memory capabilities across diverse scenarios. This gap is particularly evident
in tabletop robotic manipulation, where memory is essential for solving tasks
with partial observability and ensuring robust performance, yet no standardized
benchmarks exist. To address this, we introduce MIKASA (Memory-Intensive Skills
Assessment Suite for Agents), a comprehensive benchmark for memory RL, with
three key contributions: (1) we propose a comprehensive classification
framework for memory-intensive RL tasks, (2) we collect MIKASA-Base - a unified
benchmark that enables systematic evaluation of memory-enhanced agents across
diverse scenarios, and (3) we develop MIKASA-Robo - a novel benchmark of 32
carefully designed memory-intensive tasks that assess memory capabilities in
tabletop robotic manipulation. Our contributions establish a unified framework
for advancing memory RL research, driving the development of more reliable
systems for real-world applications. The code is available at
https://sites.google.com/view/memorybenchrobots/. | 5 | 67b478557fa6ecaa21d14a24 | null | null |
|
2025-02-18T06:33:31.888000 | Dyve: Thinking Fast and Slow for Dynamic Process Verification | 2 | {
"_id": "6608fa4f5baec84322ec85ea",
"avatarUrl": "/avatars/13bdaff931676b065fa1efef06fef922.svg",
"followerCount": 1,
"fullname": "Zhong",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Jianyuan1",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/6608fa4f5baec84322ec85ea/iiYwe_FlXRwT1RjPvzF-b.png"
]
| 2502.11157 | [
{
"_id": "67b44baa5fd91177ed7760a2",
"hidden": false,
"name": "Jianyuan Zhong",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:30:45.385Z",
"user": {
"_id": "6608fa4f5baec84322ec85ea",
"avatarUrl": "/avatars/13bdaff931676b065fa1efef06fef922.svg",
"fullname": "Zhong",
"isPro": false,
"type": "user",
"user": "Jianyuan1"
}
},
{
"_id": "67b44baa5fd91177ed7760a3",
"hidden": false,
"name": "Zeju Li",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T16:39:43.674Z",
"user": {
"_id": "664ac4f7fe822b08e6f06814",
"avatarUrl": "/avatars/23193494fcc8e58faf1eee5f1223aca6.svg",
"fullname": "Zeju Li",
"isPro": false,
"type": "user",
"user": "zeju-0727"
}
},
{
"_id": "67b44baa5fd91177ed7760a4",
"hidden": false,
"name": "Zhijian Xu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44baa5fd91177ed7760a5",
"hidden": false,
"name": "Xiangyu Wen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T16:39:45.646Z",
"user": {
"_id": "641b1b36a5f876fe30c49542",
"avatarUrl": "/avatars/ac9267925f45d325c2adb2eb0e38077b.svg",
"fullname": "Xiangyu Wen",
"isPro": false,
"type": "user",
"user": "XiangyuWen"
}
},
{
"_id": "67b44baa5fd91177ed7760a6",
"hidden": false,
"name": "Qiang Xu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-16T15:11:19 | Dyve: Thinking Fast and Slow for Dynamic Process Verification | We present Dyve, a dynamic process verifier that enhances reasoning error
detection in large language models by integrating fast and slow thinking,
inspired by Kahneman's Systems Theory. Dyve adaptively applies immediate
token-level confirmation System 1 for straightforward steps and comprehensive
analysis System 2 for complex ones. Leveraging a novel step-wise
consensus-filtered process supervision technique, combining Monte Carlo
estimation with LLM based evaluation, Dyve curates high-quality supervision
signals from noisy data. Experimental results on ProcessBench and the MATH
dataset confirm that Dyve significantly outperforms existing process-based
verifiers and boosts performance in Best-of-N settings. | 6 | 67b44bab5fd91177ed7760ca | null | null |
|
2025-02-18T06:07:36.212000 | Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention | 9 | {
"_id": "645e054ff7a55f0d780a8ff7",
"avatarUrl": "/avatars/9614510443bee3bd5d6266efd1c39fc1.svg",
"followerCount": 5,
"fullname": "Chunjiang Ge",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "HelloJiang",
"type": "user"
} | true | null | 2502.11089 | [
{
"_id": "67b43211d3c5f50aa9c03a2d",
"hidden": false,
"name": "Jingyang Yuan",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b43211d3c5f50aa9c03a2e",
"hidden": false,
"name": "Huazuo Gao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:43:19.672Z",
"user": {
"_id": "64e370be59aa5366642ac329",
"avatarUrl": "/avatars/0fa1eb6ac6c1aeff3e65bc86a6617f64.svg",
"fullname": "Huazuo Gao",
"isPro": false,
"type": "user",
"user": "gaohuazuo"
}
},
{
"_id": "67b43211d3c5f50aa9c03a2f",
"hidden": false,
"name": "Damai Dai",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:43:30.267Z",
"user": {
"_id": "659389f8de82e1ef7b9a8b13",
"avatarUrl": "/avatars/896ed9f4cdbd317493b303d070b7e12a.svg",
"fullname": "Damai Dai",
"isPro": false,
"type": "user",
"user": "DeepSeekDDM"
}
},
{
"_id": "67b43211d3c5f50aa9c03a30",
"hidden": false,
"name": "Junyu Luo",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:43:36.295Z",
"user": {
"_id": "66e6c6372c78909baf44cdf8",
"avatarUrl": "/avatars/458ea1d545d7c022b0463e7fbbd91db1.svg",
"fullname": "Junyu Luo",
"isPro": false,
"type": "user",
"user": "junyuluo"
}
},
{
"_id": "67b43211d3c5f50aa9c03a31",
"hidden": false,
"name": "Liang Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b43211d3c5f50aa9c03a32",
"hidden": false,
"name": "Zhengyan Zhang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:44:02.477Z",
"user": {
"_id": "65654ed2219af7f841640f27",
"avatarUrl": "/avatars/e6904b3479fc5e65ea1f752919ca8290.svg",
"fullname": "Zhengyan Zhang",
"isPro": false,
"type": "user",
"user": "ZhengyanZhang"
}
},
{
"_id": "67b43211d3c5f50aa9c03a33",
"hidden": false,
"name": "Zhenda Xie",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:44:16.691Z",
"user": {
"_id": "6797ca96e9e2793006a15110",
"avatarUrl": "/avatars/2d393d6e5fc2e1a867f7fdd44e055a2f.svg",
"fullname": "zhenda xie",
"isPro": false,
"type": "user",
"user": "Zhendaxie"
}
},
{
"_id": "67b43211d3c5f50aa9c03a34",
"hidden": false,
"name": "Y. X. Wei",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b43211d3c5f50aa9c03a35",
"hidden": false,
"name": "Lean Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:44:26.979Z",
"user": {
"_id": "650c509472afb1e60e6151ae",
"avatarUrl": "/avatars/c16ab5053a586819dc2b965303215ff7.svg",
"fullname": "Lean Wang",
"isPro": false,
"type": "user",
"user": "AdaHousman"
}
},
{
"_id": "67b43211d3c5f50aa9c03a36",
"hidden": false,
"name": "Zhiping Xiao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:44:34.873Z",
"user": {
"_id": "66ab566e30c55e83b02aa050",
"avatarUrl": "/avatars/62692be88b9ad34ad3f474fb0359ae20.svg",
"fullname": "Zhiping Xiao",
"isPro": false,
"type": "user",
"user": "Shockzipper"
}
},
{
"_id": "67b43211d3c5f50aa9c03a37",
"hidden": false,
"name": "Yuqing Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b43211d3c5f50aa9c03a38",
"hidden": false,
"name": "Chong Ruan",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-18T16:45:33.988Z",
"user": {
"_id": "6398203609f12714ed1935c2",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6398203609f12714ed1935c2/uXgl0LgKnFYjq1Wz39-a6.jpeg",
"fullname": "Chong Ruan",
"isPro": false,
"type": "user",
"user": "Chester111"
}
},
{
"_id": "67b43211d3c5f50aa9c03a39",
"hidden": false,
"name": "Ming Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b43211d3c5f50aa9c03a3a",
"hidden": false,
"name": "Wenfeng Liang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b43211d3c5f50aa9c03a3b",
"hidden": false,
"name": "Wangding Zeng",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-16T11:53:44 | Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse
Attention | Long-context modeling is crucial for next-generation language models, yet the
high computational cost of standard attention mechanisms poses significant
computational challenges. Sparse attention offers a promising direction for
improving efficiency while maintaining model capabilities. We present NSA, a
Natively trainable Sparse Attention mechanism that integrates algorithmic
innovations with hardware-aligned optimizations to achieve efficient
long-context modeling. NSA employs a dynamic hierarchical sparse strategy,
combining coarse-grained token compression with fine-grained token selection to
preserve both global context awareness and local precision. Our approach
advances sparse attention design with two key innovations: (1) We achieve
substantial speedups through arithmetic intensity-balanced algorithm design,
with implementation optimizations for modern hardware. (2) We enable end-to-end
training, reducing pretraining computation without sacrificing model
performance. As shown in Figure 1, experiments show the model pretrained with
NSA maintains or exceeds Full Attention models across general benchmarks,
long-context tasks, and instruction-based reasoning. Meanwhile, NSA achieves
substantial speedups over Full Attention on 64k-length sequences across
decoding, forward propagation, and backward propagation, validating its
efficiency throughout the model lifecycle. | 139 | 67b43212d3c5f50aa9c03a5c | null | null |
|
2025-02-18T05:28:54.029000 | Better Embeddings with Coupled Adam | 3 | {
"_id": "66867e1675f10ce7ef96180e",
"avatarUrl": "/avatars/ac85c00ba9d4dc48887b8864a0626743.svg",
"followerCount": null,
"fullname": "Felix Stollenwerk",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "flxst",
"type": "user"
} | true | null | 2502.08441 | [
{
"_id": "67b30311a2b3622dd42a51ff",
"hidden": false,
"name": "Felix Stollenwerk",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:32:36.770Z",
"user": {
"_id": "66867e1675f10ce7ef96180e",
"avatarUrl": "/avatars/ac85c00ba9d4dc48887b8864a0626743.svg",
"fullname": "Felix Stollenwerk",
"isPro": false,
"type": "user",
"user": "flxst"
}
},
{
"_id": "67b30311a2b3622dd42a5200",
"hidden": false,
"name": "Tobias Stollenwerk",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-12T14:32:17 | Better Embeddings with Coupled Adam | Despite their remarkable capabilities, LLMs learn word representations that
exhibit the undesirable yet poorly understood feature of anisotropy. In this
paper, we argue that the second moment in Adam is a cause of anisotropic
embeddings, and suggest a modified optimizer called Coupled Adam to mitigate
the problem. Our experiments demonstrate that Coupled Adam significantly
improves the quality of embeddings, while also leading to better upstream and
downstream performance on large enough datasets. | 1 | 67b30312a2b3622dd42a522d | null | null |
|
2025-02-18T04:37:21.573000 | Show Me the Work: Fact-Checkers' Requirements for Explainable Automated Fact-Checking | 2 | {
"_id": "6698cffdb2ebada9f4a7e7d7",
"avatarUrl": "/avatars/e66d946c14595d3b008185f2be8d2f57.svg",
"followerCount": 2,
"fullname": "Greta Warren",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "gretawarren",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/6698cffdb2ebada9f4a7e7d7/55xAEeg9Xsk87DXHTH9gM.png"
]
| 2502.09083 | [
{
"_id": "67b30726d4665a0448e6436d",
"hidden": false,
"name": "Greta Warren",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:32:34.585Z",
"user": {
"_id": "6698cffdb2ebada9f4a7e7d7",
"avatarUrl": "/avatars/e66d946c14595d3b008185f2be8d2f57.svg",
"fullname": "Greta Warren",
"isPro": false,
"type": "user",
"user": "gretawarren"
}
},
{
"_id": "67b30726d4665a0448e6436e",
"hidden": false,
"name": "Irina Shklovski",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b30726d4665a0448e6436f",
"hidden": false,
"name": "Isabelle Augenstein",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:49:35.332Z",
"user": {
"_id": "608918b7df398c3b285ce960",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1621507769190-608918b7df398c3b285ce960.jpeg",
"fullname": "Isabelle Augenstein",
"isPro": false,
"type": "user",
"user": "IAugenstein"
}
}
]
| 2025-02-13T08:56:25 | Show Me the Work: Fact-Checkers' Requirements for Explainable Automated
Fact-Checking | The pervasiveness of large language models and generative AI in online media
has amplified the need for effective automated fact-checking to assist
fact-checkers in tackling the increasing volume and sophistication of
misinformation. The complex nature of fact-checking demands that automated
fact-checking systems provide explanations that enable fact-checkers to
scrutinise their outputs. However, it is unclear how these explanations should
align with the decision-making and reasoning processes of fact-checkers to be
effectively integrated into their workflows. Through semi-structured interviews
with fact-checking professionals, we bridge this gap by: (i) providing an
account of how fact-checkers assess evidence, make decisions, and explain their
processes; (ii) examining how fact-checkers use automated tools in practice;
and (iii) identifying fact-checker explanation requirements for automated
fact-checking tools. The findings show unmet explanation needs and identify
important criteria for replicable fact-checking explanations that trace the
model's reasoning path, reference specific evidence, and highlight uncertainty
and information gaps. | 4 | 67b30727d4665a0448e6438d | null | null |
|
2025-02-18T04:34:15.786000 | MagicArticulate: Make Your 3D Models Articulation-Ready | 2 | {
"_id": "64fb31a34c8924c4fe7498bc",
"avatarUrl": "/avatars/6c8e4a66e1b8b3c786a4000210089392.svg",
"followerCount": 4,
"fullname": "Chaoyue Song",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "chaoyue7",
"type": "user"
} | true | null | 2502.12135 | [
{
"_id": "67b4028237db78705fb256e1",
"hidden": false,
"name": "Chaoyue Song",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:40.771Z",
"user": {
"_id": "64fb31a34c8924c4fe7498bc",
"avatarUrl": "/avatars/6c8e4a66e1b8b3c786a4000210089392.svg",
"fullname": "Chaoyue Song",
"isPro": false,
"type": "user",
"user": "chaoyue7"
}
},
{
"_id": "67b4028237db78705fb256e2",
"hidden": false,
"name": "Jianfeng Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256e3",
"hidden": false,
"name": "Xiu Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256e4",
"hidden": false,
"name": "Fan Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256e5",
"hidden": false,
"name": "Yiwen Chen",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256e6",
"hidden": false,
"name": "Zhongcong Xu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256e7",
"hidden": false,
"name": "Jun Hao Liew",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256e8",
"hidden": false,
"name": "Xiaoyang Guo",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256e9",
"hidden": false,
"name": "Fayao Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256ea",
"hidden": false,
"name": "Jiashi Feng",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4028237db78705fb256eb",
"hidden": false,
"name": "Guosheng Lin",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T18:53:27 | MagicArticulate: Make Your 3D Models Articulation-Ready | With the explosive growth of 3D content creation, there is an increasing
demand for automatically converting static 3D models into articulation-ready
versions that support realistic animation. Traditional approaches rely heavily
on manual annotation, which is both time-consuming and labor-intensive.
Moreover, the lack of large-scale benchmarks has hindered the development of
learning-based solutions. In this work, we present MagicArticulate, an
effective framework that automatically transforms static 3D models into
articulation-ready assets. Our key contributions are threefold. First, we
introduce Articulation-XL, a large-scale benchmark containing over 33k 3D
models with high-quality articulation annotations, carefully curated from
Objaverse-XL. Second, we propose a novel skeleton generation method that
formulates the task as a sequence modeling problem, leveraging an
auto-regressive transformer to naturally handle varying numbers of bones or
joints within skeletons and their inherent dependencies across different 3D
models. Third, we predict skinning weights using a functional diffusion process
that incorporates volumetric geodesic distance priors between vertices and
joints. Extensive experiments demonstrate that MagicArticulate significantly
outperforms existing methods across diverse object categories, achieving
high-quality articulation that enables realistic animation. Project page:
https://chaoyuesong.github.io/MagicArticulate. | 8 | 67b4028437db78705fb25726 | null | null |
|
2025-02-18T04:33:41.120000 | I Think, Therefore I Diffuse: Enabling Multimodal In-Context Reasoning in Diffusion Models | 3 | {
"_id": "6354bda206d707b33249c4c2",
"avatarUrl": "/avatars/bbd9f76274ac52214df92084d50bc7b5.svg",
"followerCount": 1,
"fullname": "Zhenxing Mi",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Mifucius",
"type": "user"
} | true | null | 2502.10458 | [
{
"_id": "67b3ea0f4dd7ea0538ce589d",
"hidden": false,
"name": "Zhenxing Mi",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:52.837Z",
"user": {
"_id": "6354bda206d707b33249c4c2",
"avatarUrl": "/avatars/bbd9f76274ac52214df92084d50bc7b5.svg",
"fullname": "Zhenxing Mi",
"isPro": false,
"type": "user",
"user": "Mifucius"
}
},
{
"_id": "67b3ea0f4dd7ea0538ce589e",
"hidden": false,
"name": "Kuan-Chieh Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:21:46.349Z",
"user": {
"_id": "648ca58a39d2584ee47efef6",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/648ca58a39d2584ee47efef6/R7B72bnwc59mdK45rmzYS.png",
"fullname": "Kuan-Chieh Wang",
"isPro": false,
"type": "user",
"user": "wangkua1"
}
},
{
"_id": "67b3ea0f4dd7ea0538ce589f",
"hidden": false,
"name": "Guocheng Qian",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:21:52.861Z",
"user": {
"_id": "645fed74335c21d19f3bf76c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/645fed74335c21d19f3bf76c/gwVsllRWtSHbg4a1erkdF.jpeg",
"fullname": "Guocheng Qian",
"isPro": false,
"type": "user",
"user": "guochengqian"
}
},
{
"_id": "67b3ea0f4dd7ea0538ce58a0",
"hidden": false,
"name": "Hanrong Ye",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:21:58.865Z",
"user": {
"_id": "62d3ae4d894e7fe42def988f",
"avatarUrl": "/avatars/3aafc55d9783459f9a79546fc31dd68a.svg",
"fullname": "Hanrong Ye",
"isPro": false,
"type": "user",
"user": "leoye"
}
},
{
"_id": "67b3ea0f4dd7ea0538ce58a1",
"hidden": false,
"name": "Runtao Liu",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:22:08.197Z",
"user": {
"_id": "64a653e330dd11336539c439",
"avatarUrl": "/avatars/348910ea160829707ac5e74f9f824c60.svg",
"fullname": "liuruntao",
"isPro": false,
"type": "user",
"user": "runtao"
}
},
{
"_id": "67b3ea0f4dd7ea0538ce58a2",
"hidden": false,
"name": "Sergey Tulyakov",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b3ea0f4dd7ea0538ce58a3",
"hidden": false,
"name": "Kfir Aberman",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:22:17.444Z",
"user": {
"_id": "64db29097266618e853dd6ec",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64db29097266618e853dd6ec/r0MaPQCfAxeKv3ycdKYLK.jpeg",
"fullname": "Kfir Aberman",
"isPro": false,
"type": "user",
"user": "kaberman"
}
},
{
"_id": "67b3ea0f4dd7ea0538ce58a4",
"hidden": false,
"name": "Dan Xu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:53.095Z",
"user": {
"_id": "66feab48651e00e22f33222e",
"avatarUrl": "/avatars/7344377e2c796c7ec85194bb2fc78521.svg",
"fullname": "Dan Xu",
"isPro": false,
"type": "user",
"user": "danxuhk"
}
}
]
| 2025-02-12T05:30:08 | I Think, Therefore I Diffuse: Enabling Multimodal In-Context Reasoning
in Diffusion Models | This paper presents ThinkDiff, a novel alignment paradigm that empowers
text-to-image diffusion models with multimodal in-context understanding and
reasoning capabilities by integrating the strengths of vision-language models
(VLMs). Existing multimodal diffusion finetuning methods largely focus on
pixel-level reconstruction rather than in-context reasoning, and are
constrained by the complexity and limited availability of reasoning-based
datasets. ThinkDiff addresses these challenges by leveraging vision-language
training as a proxy task, aligning VLMs with the decoder of an encoder-decoder
large language model (LLM) instead of a diffusion decoder. This proxy task
builds on the observation that the LLM decoder shares the same input
feature space with diffusion decoders that use the corresponding
LLM encoder for prompt embedding. As a result, aligning VLMs with
diffusion decoders can be simplified through alignment with the LLM decoder.
Without complex training and datasets, ThinkDiff effectively unleashes
understanding, reasoning, and composing capabilities in diffusion models.
Experiments demonstrate that ThinkDiff significantly improves accuracy from
19.2% to 46.3% on the challenging CoBSAT benchmark for multimodal in-context
reasoning generation, with only 5 hours of training on 4 A100 GPUs.
Additionally, ThinkDiff demonstrates exceptional performance in composing
multiple images and texts into logically coherent images. Project page:
https://mizhenxing.github.io/ThinkDiff. | 30 | 67b3ea124dd7ea0538ce592d | https://mizhenxing.github.io/ThinkDiff | https://github.com/MiZhenxing/ThinkDiff |
|
2025-02-18T04:20:25.916000 | Intuitive physics understanding emerges from self-supervised pretraining on natural videos | 2 | {
"_id": "5f1158120c833276f61f1a84",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1608042047613-5f1158120c833276f61f1a84.jpeg",
"followerCount": 777,
"fullname": "Niels Rogge",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "nielsr",
"type": "user"
} | false | null | 2502.11831 | [
{
"_id": "67b450cf315f7b69956df3d6",
"hidden": false,
"name": "Quentin Garrido",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:28:09.217Z",
"user": {
"_id": "63049022412a1b9d381b9dcb",
"avatarUrl": "/avatars/7382c0a0e3f5609b754ec09a309d33f6.svg",
"fullname": "Quentin Garrido",
"isPro": false,
"type": "user",
"user": "garridoq"
}
},
{
"_id": "67b450cf315f7b69956df3d7",
"hidden": false,
"name": "Nicolas Ballas",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b450cf315f7b69956df3d8",
"hidden": false,
"name": "Mahmoud Assran",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b450cf315f7b69956df3d9",
"hidden": false,
"name": "Adrien Bardes",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b450cf315f7b69956df3da",
"hidden": false,
"name": "Laurent Najman",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b450cf315f7b69956df3db",
"hidden": false,
"name": "Michael Rabbat",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b450cf315f7b69956df3dc",
"hidden": false,
"name": "Emmanuel Dupoux",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:44:00.610Z",
"user": {
"_id": "63317d2118711776b4663c3a",
"avatarUrl": "/avatars/7dedd1934c1000b6f81a2a37ec348347.svg",
"fullname": "Emmanuel Dupoux",
"isPro": false,
"type": "user",
"user": "edupoux"
}
},
{
"_id": "67b450cf315f7b69956df3dd",
"hidden": false,
"name": "Yann LeCun",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:43:53.902Z",
"user": {
"_id": "64ed0b8c2203a126eb1a5b9a",
"avatarUrl": "/avatars/9156dc406ed3f9ee62b73657ac20f5ed.svg",
"fullname": "Yann LeCun",
"isPro": false,
"type": "user",
"user": "ylecun"
}
}
]
| 2025-02-17T14:27:14 | Intuitive physics understanding emerges from self-supervised pretraining
on natural videos | We investigate the emergence of intuitive physics understanding in
general-purpose deep neural network models trained to predict masked regions in
natural videos. Leveraging the violation-of-expectation framework, we find that
video prediction models trained to predict outcomes in a learned representation
space demonstrate an understanding of various intuitive physics properties,
such as object permanence and shape consistency. In contrast, video prediction
in pixel space and multimodal large language models, which reason through text,
achieve performance closer to chance. Our comparisons of these architectures
reveal that jointly learning an abstract representation space while predicting
missing parts of sensory input, akin to predictive coding, is sufficient to
acquire an understanding of intuitive physics, and that even models trained on
one week of unique video achieve above chance performance. This challenges the
idea that core knowledge -- a set of innate systems to help understand the
world -- needs to be hardwired to develop an understanding of intuitive
physics. | 18 | 67b450d0315f7b69956df3f9 | null | https://github.com/facebookresearch/jepa-intuitive-physics |
|
2025-02-18T04:16:28.219000 | Towards Data-Efficient Pretraining for Atomic Property Prediction | 3 | {
"_id": "642b51385bf2355d02a23d15",
"avatarUrl": "/avatars/87985347643b2647555f2453fa4d94fb.svg",
"followerCount": 4,
"fullname": "Hasan Abed Al Kader Hammoud",
"isHf": false,
"isMod": false,
"isPro": true,
"name": "hammh0a",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/642b51385bf2355d02a23d15/bLvTbh56AkUmcmRst8mT3.png"
]
| 2502.11085 | [
{
"_id": "67b44f44620ae0bad17d6699",
"hidden": false,
"name": "Yasir Ghunaim",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44f44620ae0bad17d669a",
"hidden": false,
"name": "Hasan Abed Al Kader Hammoud",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:30:43.057Z",
"user": {
"_id": "642b51385bf2355d02a23d15",
"avatarUrl": "/avatars/87985347643b2647555f2453fa4d94fb.svg",
"fullname": "Hasan Abed Al Kader Hammoud",
"isPro": true,
"type": "user",
"user": "hammh0a"
}
},
{
"_id": "67b44f44620ae0bad17d669b",
"hidden": false,
"name": "Bernard Ghanem",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-16T11:46:23 | Towards Data-Efficient Pretraining for Atomic Property Prediction | This paper challenges the recent paradigm in atomic property prediction that
links progress to growing dataset sizes and computational resources. We show
that pretraining on a carefully selected, task-relevant dataset can match or
even surpass large-scale pretraining, while using as little as 1/24th of the
computational cost. We introduce the Chemical Similarity Index (CSI), a novel
metric inspired by computer vision's Fr\'echet Inception Distance, for
molecular graphs which quantifies the alignment between upstream pretraining
datasets and downstream tasks. By selecting the most relevant dataset with
minimal CSI distance, we show that models pretrained on a smaller, focused
dataset consistently outperform those pretrained on massive, mixed datasets
such as JMP, even when those larger datasets include the relevant dataset.
Counterintuitively, we also find that indiscriminately adding more data can
degrade model performance when the additional data poorly aligns with the task
at hand. Our findings highlight that quality often outperforms quantity in
pretraining for atomic property prediction. | 3 | 67b44f45620ae0bad17d66b0 | null | null |
|
2025-02-18T03:53:47.570000 | PhysReason: A Comprehensive Benchmark towards Physics-Based Reasoning | 2 | {
"_id": "6602548a68d519ed324b47c5",
"avatarUrl": "/avatars/5ab411f87440cc2a98c7a1c6a3ed5548.svg",
"followerCount": 4,
"fullname": "ChengyouJia",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "ChengyouJia",
"type": "user"
} | true | null | 2502.12054 | [
{
"_id": "67b44a6888813676da9f8239",
"hidden": false,
"name": "Xinyu Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44a6888813676da9f823a",
"hidden": false,
"name": "Yuxuan Dong",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44a6888813676da9f823b",
"hidden": false,
"name": "Yanrui Wu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44a6888813676da9f823c",
"hidden": false,
"name": "Jiaxing Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44a6888813676da9f823d",
"hidden": false,
"name": "Chengyou Jia",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:30:47.313Z",
"user": {
"_id": "6602548a68d519ed324b47c5",
"avatarUrl": "/avatars/5ab411f87440cc2a98c7a1c6a3ed5548.svg",
"fullname": "ChengyouJia",
"isPro": false,
"type": "user",
"user": "ChengyouJia"
}
},
{
"_id": "67b44a6888813676da9f823e",
"hidden": false,
"name": "Basura Fernando",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44a6888813676da9f823f",
"hidden": false,
"name": "Mike Zheng Shou",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44a6888813676da9f8240",
"hidden": false,
"name": "Lingling Zhang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b44a6888813676da9f8241",
"hidden": false,
"name": "Jun Liu",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T17:24:14 | PhysReason: A Comprehensive Benchmark towards Physics-Based Reasoning | Large language models demonstrate remarkable capabilities across various
domains, especially mathematics and logic reasoning. However, current
evaluations overlook physics-based reasoning - a complex task requiring physics
theorems and constraints. We present PhysReason, a 1,200-problem benchmark
comprising knowledge-based (25%) and reasoning-based (75%) problems, where the
latter are divided into three difficulty levels (easy, medium, hard). Notably,
problems require an average of 8.1 solution steps, with hard requiring 15.6,
reflecting the complexity of physics-based reasoning. We propose the Physics
Solution Auto Scoring Framework, incorporating efficient answer-level and
comprehensive step-level evaluations. Top-performing models like Deepseek-R1,
Gemini-2.0-Flash-Thinking, and o3-mini-high achieve less than 60% on
answer-level evaluation, with performance dropping from knowledge questions
(75.11%) to hard problems (31.95%). Through step-level evaluation, we
identified four key bottlenecks: Physics Theorem Application, Physics Process
Understanding, Calculation, and Physics Condition Analysis. These findings
position PhysReason as a novel and comprehensive benchmark for evaluating
physics-based reasoning capabilities in large language models. Our code and
data will be published at https:/dxzxy12138.github.io/PhysReason. | 5 | 67b44a6988813676da9f82d0 | null | null |
|
2025-02-18T02:26:18.856000 | Large Language Models and Mathematical Reasoning Failures | 3 | {
"_id": "6033e34a9aa44495c80dd043",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1614079701740-6033e34a9aa44495c80dd043.jpeg",
"followerCount": 39,
"fullname": "Birger Moell",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "birgermoell",
"type": "user"
} | true | null | 2502.11574 | [
{
"_id": "67b435c29e5685b308a8edac",
"hidden": false,
"name": "Johan Boye",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-18T07:24:50.956Z",
"user": {
"_id": "65bcbc01d6d0ffbceb8b2e6e",
"avatarUrl": "/avatars/73edb2d6b7b11208439ac88b365079e8.svg",
"fullname": "Johan Boye",
"isPro": false,
"type": "user",
"user": "jboye"
}
},
{
"_id": "67b435c29e5685b308a8edad",
"hidden": false,
"name": "Birger Moell",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:30:49.328Z",
"user": {
"_id": "6033e34a9aa44495c80dd043",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1614079701740-6033e34a9aa44495c80dd043.jpeg",
"fullname": "Birger Moell",
"isPro": false,
"type": "user",
"user": "birgermoell"
}
}
]
| 2025-02-17T09:07:32 | Large Language Models and Mathematical Reasoning Failures | This paper investigates the mathematical reasoning capabilities of large
language models (LLMs) using 50 newly constructed high-school-level word
problems. Unlike prior studies that focus solely on answer correctness, we
rigorously analyze both final answers and solution steps to identify reasoning
failures. Evaluating eight state-of-the-art models - including Mixtral, Llama,
Gemini, GPT-4o, and OpenAI's o1 variants - we find that while newer models
(e.g., o3-mini, deepseek-r1) achieve higher accuracy, all models exhibit errors
in spatial reasoning, strategic planning, and arithmetic, sometimes producing
correct answers through flawed logic. Common failure modes include unwarranted
assumptions, over-reliance on numerical patterns, and difficulty translating
physical intuition into mathematical steps. Manual analysis reveals that models
struggle with problems requiring multi-step deduction or real-world knowledge,
despite possessing broad mathematical knowledge. Our results underscore the
importance of evaluating reasoning processes, not just answers, and caution
against overestimating LLMs' problem-solving proficiency. The study highlights
persistent gaps in LLMs' generalization abilities, emphasizing the need for
targeted improvements in structured reasoning and constraint handling. | 3 | 67b435c29e5685b308a8edf1 | null | null |
|
2025-02-18T02:23:29.869000 | Language Complexity Measurement as a Noisy Zero-Shot Proxy for Evaluating LLM Performance | 2 | {
"_id": "6033e34a9aa44495c80dd043",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1614079701740-6033e34a9aa44495c80dd043.jpeg",
"followerCount": 39,
"fullname": "Birger Moell",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "birgermoell",
"type": "user"
} | true | null | 2502.11578 | [
{
"_id": "67b435475bff5f34c1ebee1b",
"hidden": false,
"name": "Birger Moell",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:30:52.639Z",
"user": {
"_id": "6033e34a9aa44495c80dd043",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1614079701740-6033e34a9aa44495c80dd043.jpeg",
"fullname": "Birger Moell",
"isPro": false,
"type": "user",
"user": "birgermoell"
}
},
{
"_id": "67b435475bff5f34c1ebee1c",
"hidden": false,
"name": "Johan Boye",
"status": "extracted_pending",
"statusLastChangedAt": "2025-02-18T07:22:48.554Z",
"user": {
"_id": "65bcbc01d6d0ffbceb8b2e6e",
"avatarUrl": "/avatars/73edb2d6b7b11208439ac88b365079e8.svg",
"fullname": "Johan Boye",
"isPro": false,
"type": "user",
"user": "jboye"
}
}
]
| 2025-02-17T09:09:58 | Language Complexity Measurement as a Noisy Zero-Shot Proxy for
Evaluating LLM Performance | Large Language Models (LLMs) have made significant strides in natural
language generation but often face challenges in tasks requiring precise
calculations and structural analysis. This paper investigates the performance
of state-of-the-art LLMs on language complexity measurement tasks, through the
computation of the LIX readability metric and Average Dependency Distance
(ADD). Using Swedish high school and university-level essays, we evaluate the
models' abilities to compute LIX scores and perform dependency parsing,
comparing their results to established ground truths. Our findings reveal that
while all models demonstrate some capacity for these tasks, ChatGPT-o1-mini
performs most consistently, achieving the highest accuracy in both LIX
computation and dependency parsing. Additionally, we observe a strong
significant correlation -0.875 p 0.026 (N=6) between the models' accuracy in
computing LIX and their overall performance on the Massive Multitask Language
Understanding (MMLU) benchmark. These results suggest that language complexity
measurement abilities can serve as a noisy zero-shot proxies for assessing the
general capabilities of LLMs, providing a practical method for model evaluation
without the need for extensive benchmarking datasets. | 0 | 67b435485bff5f34c1ebee52 | null | null |
|
2025-02-18T01:45:36.359000 | System Message Generation for User Preferences using Open-Source Models | 2 | {
"_id": "64587be872b60ae7a3817858",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64587be872b60ae7a3817858/BbdOOxOCEzWTvEpkWp8MM.png",
"followerCount": 3,
"fullname": "Minbyul Jeong",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Minbyul",
"type": "user"
} | true | null | 2502.11330 | [
{
"_id": "67b42c5632929e97a92dee90",
"hidden": false,
"name": "Minbyul Jeong",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T09:04:45.723Z",
"user": {
"_id": "64587be872b60ae7a3817858",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64587be872b60ae7a3817858/BbdOOxOCEzWTvEpkWp8MM.png",
"fullname": "Minbyul Jeong",
"isPro": false,
"type": "user",
"user": "Minbyul"
}
},
{
"_id": "67b42c5632929e97a92dee91",
"hidden": false,
"name": "Jungho Cho",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-21T09:59:57.458Z",
"user": {
"_id": "6596a87480a4560b8f9b9532",
"avatarUrl": "/avatars/33f39ee01c1648f1daca41a40a4964fb.svg",
"fullname": "Christopher Cho",
"isPro": false,
"type": "user",
"user": "ChristopherCho"
}
},
{
"_id": "67b42c5632929e97a92dee92",
"hidden": false,
"name": "Minsoo Khang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b42c5632929e97a92dee93",
"hidden": false,
"name": "Dawoon Jung",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:48:50.413Z",
"user": {
"_id": "6268ec3bd4aa92e53a66d028",
"avatarUrl": "/avatars/af7fd3b4a49aca8b71617f3f17673227.svg",
"fullname": "dawoon jung",
"isPro": false,
"type": "user",
"user": "noowad"
}
},
{
"_id": "67b42c5632929e97a92dee94",
"hidden": false,
"name": "Teakgyu Hong",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:48:42.592Z",
"user": {
"_id": "6551e62ca12178dd75f4592f",
"avatarUrl": "/avatars/f4213f9cb95847182bcca3281b1a042c.svg",
"fullname": "Teakgyu Hong",
"isPro": false,
"type": "user",
"user": "tghong"
}
}
]
| 2025-02-17T01:05:31 | System Message Generation for User Preferences using Open-Source Models | System messages play a crucial role in interactions with large language
models (LLMs), often serving as prompts to initiate conversations. Through
system messages, users can assign specific roles, perform intended tasks,
incorporate background information, specify various output formats and
communication styles. Despite such versatility, publicly available data are
often lack system messages and subject to strict license constraints in the
industry field. Manual labeling of publicly available data with system messages
that align with user instructions demands significant resources. In view of
such challenges, our work introduces SysGen, a pipeline for generating system
messages with better aligned assistant responses from the supervised
fine-tuning dataset without system messages. Training on SysGen data has
demonstrated substantial improvements in the alignment of model responses with
system messages and user instructions, as demonstrated across various
open-source models on the Multifacet benchmark, while maintaining minimal
impact on other unseen benchmarks such as Open LLM Leaderboard 2. Our
qualitative analysis highlights the importance of diverse system messages to
ensure better adaptability across different contexts. | 15 | 67b42c5732929e97a92deed7 | null | null |
|
2025-02-18T01:02:25.236000 | How Do LLMs Acquire New Knowledge? A Knowledge Circuits Perspective on Continual Pre-Training | 6 | {
"_id": "620b3bbb0668e435407c8d0a",
"avatarUrl": "/avatars/e0fccbb2577d76088e09f054c35cffbc.svg",
"followerCount": 19,
"fullname": "Ningyu Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Ningyu",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/620b3bbb0668e435407c8d0a/_LGnwvwslWc3YDIirfOKS.png"
]
| 2502.11196 | [
{
"_id": "67b42223c2fe54b8d43efed6",
"hidden": false,
"name": "Yixin Ou",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:22:40.840Z",
"user": {
"_id": "6241749cf80bd930bd99f3dd",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1669210243382-6241749cf80bd930bd99f3dd.jpeg",
"fullname": "Ou Yixin",
"isPro": false,
"type": "user",
"user": "OE-Heart"
}
},
{
"_id": "67b42223c2fe54b8d43efed7",
"hidden": false,
"name": "Yunzhi Yao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:22:56.226Z",
"user": {
"_id": "6122fbe636a4c36a99dbea7b",
"avatarUrl": "/avatars/c0cd2c1ef58e315d9adda9d26000f625.svg",
"fullname": "Yunzhi Yao",
"isPro": false,
"type": "user",
"user": "cowTodd"
}
},
{
"_id": "67b42223c2fe54b8d43efed8",
"hidden": false,
"name": "Ningyu Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:04.227Z",
"user": {
"_id": "620b3bbb0668e435407c8d0a",
"avatarUrl": "/avatars/e0fccbb2577d76088e09f054c35cffbc.svg",
"fullname": "Ningyu Zhang",
"isPro": false,
"type": "user",
"user": "Ningyu"
}
},
{
"_id": "67b42223c2fe54b8d43efed9",
"hidden": false,
"name": "Hui Jin",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b42223c2fe54b8d43efeda",
"hidden": false,
"name": "Jiacheng Sun",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:23:08.677Z",
"user": {
"_id": "6591895250d39af7f431ad3d",
"avatarUrl": "/avatars/83c9717e7d8bea8cae6a640ee6455214.svg",
"fullname": "Sun Jiacheng",
"isPro": false,
"type": "user",
"user": "sunjc826"
}
},
{
"_id": "67b42223c2fe54b8d43efedb",
"hidden": false,
"name": "Shumin Deng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:23:17.528Z",
"user": {
"_id": "6441f1d2603214724ec0c1c2",
"avatarUrl": "/avatars/d3c4b759e6a5635e37ff715fae52e5ba.svg",
"fullname": "Shumin Deng",
"isPro": false,
"type": "user",
"user": "231sm"
}
},
{
"_id": "67b42223c2fe54b8d43efedc",
"hidden": false,
"name": "Zhenguo Li",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b42223c2fe54b8d43efedd",
"hidden": false,
"name": "Huajun Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:23:34.297Z",
"user": {
"_id": "64931296137833d7ec7689cd",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64931296137833d7ec7689cd/TBihNdp1ZwIWjhfAWjRr6.jpeg",
"fullname": "Huajun Chen",
"isPro": false,
"type": "user",
"user": "huajunsir"
}
}
]
| 2025-02-16T16:55:43 | How Do LLMs Acquire New Knowledge? A Knowledge Circuits Perspective on
Continual Pre-Training | Despite exceptional capabilities in knowledge-intensive tasks, Large Language
Models (LLMs) face a critical gap in understanding how they internalize new
knowledge, particularly how to structurally embed acquired knowledge in their
neural computations. We address this issue through the lens of knowledge
circuit evolution, identifying computational subgraphs that facilitate
knowledge storage and processing. Our systematic analysis of circuit evolution
throughout continual pre-training reveals several key findings: (1) the
acquisition of new knowledge is influenced by its relevance to pre-existing
knowledge; (2) the evolution of knowledge circuits exhibits a distinct phase
shift from formation to optimization; (3) the evolution of knowledge circuits
follows a deep-to-shallow pattern. These insights not only advance our
theoretical understanding of the mechanisms of new knowledge acquisition in
LLMs, but also provide potential implications for improving continual
pre-training strategies to enhance model performance. Code and data will be
available at https://github.com/zjunlp/DynamicKnowledgeCircuits. | 22 | 67b42225c2fe54b8d43eff9b | null | null |
|
2025-02-18T01:01:24.331000 | SURGE: On the Potential of Large Language Models as General-Purpose Surrogate Code Executors | 2 | {
"_id": "650267e7e751d03da933a24a",
"avatarUrl": "/avatars/f047a047d1de304cd97027463541bdf3.svg",
"followerCount": 1,
"fullname": "Bohan22",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Bohan22",
"type": "user"
} | true | null | 2502.11167 | [
{
"_id": "67b4221bbc387d2eda6f8637",
"hidden": false,
"name": "Bohan Lyu",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:06.388Z",
"user": {
"_id": "650267e7e751d03da933a24a",
"avatarUrl": "/avatars/f047a047d1de304cd97027463541bdf3.svg",
"fullname": "Bohan22",
"isPro": false,
"type": "user",
"user": "Bohan22"
}
},
{
"_id": "67b4221bbc387d2eda6f8638",
"hidden": false,
"name": "Siqiao Huang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b4221bbc387d2eda6f8639",
"hidden": false,
"name": "Zichen Liang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:08.469Z",
"user": {
"_id": "67286718746a95c09d04cb1d",
"avatarUrl": "/avatars/317efa8459cca08c2ff56c3ab116e15c.svg",
"fullname": "Zichen Liang",
"isPro": false,
"type": "user",
"user": "zcliang22"
}
}
]
| 2025-02-16T15:38:19 | SURGE: On the Potential of Large Language Models as General-Purpose
Surrogate Code Executors | Large language models (LLMs) have demonstrated remarkable capabilities in
code-related tasks, such as code understanding and code generation. However, an
equally important yet underexplored question is whether LLMs can serve as
general-purpose surrogate code executors, to predict the output and behavior of
a program without actually running it. To systematically investigate this
capability, we introduce SURGE, a comprehensive benchmark covering eight key
aspects: multi-language programming tasks, competition-level programming
problems, repository-level code analysis, high-cost scientific computing,
time-complexity-intensive algorithms, buggy code analysis, programs dependent
on specific compilers or execution environments, and formal mathematical proof
verification. We evaluate multiple open-source and proprietary LLMs on SURGE
and conduct a scaling study to analyze the impact of model size and training
data scale on surrogate execution accuracy. Additionally, we categorize model
prediction errors and explore potential areas for improvement. Our findings
indicate that while LLMs can predict code execution results in certain cases,
they exhibit limitations in general-purpose surrogate execution. This study
provides empirical insights into the feasibility of using LLMs as surrogate
code executors. Code and dataset are released at
https://github.com/Imbernoulli/SURGE. | 10 | 67b4221ebc387d2eda6f8717 | null | null |
|
2025-02-18T00:58:24.094000 | ReLearn: Unlearning via Learning for Large Language Models | 2 | {
"_id": "620b3bbb0668e435407c8d0a",
"avatarUrl": "/avatars/e0fccbb2577d76088e09f054c35cffbc.svg",
"followerCount": 19,
"fullname": "Ningyu Zhang",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "Ningyu",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/620b3bbb0668e435407c8d0a/A4YB7t6hDVty6QrvLN0a7.png"
]
| 2502.11190 | [
{
"_id": "67b420dfb2528c023491f455",
"hidden": false,
"name": "Haoming Xu",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b420dfb2528c023491f456",
"hidden": true,
"name": "Ningyuan Zhao",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:20:38.762Z",
"user": {
"_id": "6641f74c3bd2f40eb41fb0d8",
"avatarUrl": "/avatars/b0167a521b6e4aea27e245dd9e026ef3.svg",
"fullname": "zhaoningyuan",
"isPro": false,
"type": "user",
"user": "zhaoningyuan"
}
},
{
"_id": "67b420dfb2528c023491f457",
"hidden": false,
"name": "Liming Yang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b420dfb2528c023491f458",
"hidden": false,
"name": "Sendong Zhao",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b420dfb2528c023491f459",
"hidden": false,
"name": "Shumin Deng",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:20:59.377Z",
"user": {
"_id": "6441f1d2603214724ec0c1c2",
"avatarUrl": "/avatars/d3c4b759e6a5635e37ff715fae52e5ba.svg",
"fullname": "Shumin Deng",
"isPro": false,
"type": "user",
"user": "231sm"
}
},
{
"_id": "67b420dfb2528c023491f45a",
"hidden": false,
"name": "Mengru Wang",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:21:08.761Z",
"user": {
"_id": "64bf898d979949d2e2585c9a",
"avatarUrl": "/avatars/da77c856ec997e2b812c06272a01c8b2.svg",
"fullname": "mengruwang",
"isPro": false,
"type": "user",
"user": "mengru"
}
},
{
"_id": "67b420dfb2528c023491f45b",
"hidden": false,
"name": "Bryan Hooi",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:21:14.704Z",
"user": {
"_id": "651d8032c50012d33e914f2f",
"avatarUrl": "/avatars/0a44c9f51fc50ce86582e328c361ea00.svg",
"fullname": "Bryan Hooi",
"isPro": false,
"type": "user",
"user": "bhooi"
}
},
{
"_id": "67b420dfb2528c023491f45c",
"hidden": false,
"name": "Nay Oo",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b420dfb2528c023491f45d",
"hidden": false,
"name": "Huajun Chen",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:21:32.744Z",
"user": {
"_id": "64931296137833d7ec7689cd",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/64931296137833d7ec7689cd/TBihNdp1ZwIWjhfAWjRr6.jpeg",
"fullname": "Huajun Chen",
"isPro": false,
"type": "user",
"user": "huajunsir"
}
},
{
"_id": "67b420dfb2528c023491f45e",
"hidden": false,
"name": "Ningyu Zhang",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:11.243Z",
"user": {
"_id": "620b3bbb0668e435407c8d0a",
"avatarUrl": "/avatars/e0fccbb2577d76088e09f054c35cffbc.svg",
"fullname": "Ningyu Zhang",
"isPro": false,
"type": "user",
"user": "Ningyu"
}
}
]
| 2025-02-16T16:31:00 | ReLearn: Unlearning via Learning for Large Language Models | Current unlearning methods for large language models usually rely on reverse
optimization to reduce target token probabilities. However, this paradigm
disrupts the subsequent tokens prediction, degrading model performance and
linguistic coherence. Moreover, existing evaluation metrics overemphasize
contextual forgetting while inadequately assessing response fluency and
relevance. To address these challenges, we propose ReLearn, a data augmentation
and fine-tuning pipeline for effective unlearning, along with a comprehensive
evaluation framework. This framework introduces Knowledge Forgetting Rate (KFR)
and Knowledge Retention Rate (KRR) to measure knowledge-level preservation, and
Linguistic Score (LS) to evaluate generation quality. Our experiments show that
ReLearn successfully achieves targeted forgetting while preserving high-quality
output. Through mechanistic analysis, we further demonstrate how reverse
optimization disrupts coherent text generation, while ReLearn preserves this
essential capability. Code is available at https://github.com/zjunlp/unlearn. | 29 | 67b420e2b2528c023491f506 | null | null |
|
2025-02-18T00:49:53.124000 | Learning Getting-Up Policies for Real-World Humanoid Robots | 3 | {
"_id": "6201fc5d91d53938a6432fbf",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6201fc5d91d53938a6432fbf/VLs8ZYaZrop4KBpZn53fH.jpeg",
"followerCount": 3,
"fullname": "Runpei Dong",
"isHf": false,
"isMod": false,
"isPro": false,
"name": "RunpeiDong",
"type": "user"
} | true | [
"https://cdn-uploads.huggingface.co/production/uploads/6201fc5d91d53938a6432fbf/x35BuXOhc6ubukxLfiVzt.mp4"
]
| 2502.12152 | [
{
"_id": "67b41ed52867282b4eb37ce4",
"hidden": false,
"name": "Xialin He",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b41ed52867282b4eb37ce5",
"hidden": false,
"name": "Runpei Dong",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-18T09:31:13.178Z",
"user": {
"_id": "6201fc5d91d53938a6432fbf",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6201fc5d91d53938a6432fbf/VLs8ZYaZrop4KBpZn53fH.jpeg",
"fullname": "Runpei Dong",
"isPro": false,
"type": "user",
"user": "RunpeiDong"
}
},
{
"_id": "67b41ed52867282b4eb37ce6",
"hidden": false,
"name": "Zixuan Chen",
"status": "claimed_verified",
"statusLastChangedAt": "2025-02-19T15:44:49.892Z",
"user": {
"_id": "64a02b4c1396c67ac07798bb",
"avatarUrl": "/avatars/9b1b4319edbac5faeb7586a4933791d2.svg",
"fullname": "Eric Chen",
"isPro": false,
"type": "user",
"user": "zxuannn"
}
},
{
"_id": "67b41ed52867282b4eb37ce7",
"hidden": false,
"name": "Saurabh Gupta",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T18:59:06 | Learning Getting-Up Policies for Real-World Humanoid Robots | Automatic fall recovery is a crucial prerequisite before humanoid robots can
be reliably deployed. Hand-designing controllers for getting up is difficult
because of the varied configurations a humanoid can end up in after a fall and
the challenging terrains humanoid robots are expected to operate on. This paper
develops a learning framework to produce controllers that enable humanoid
robots to get up from varying configurations on varying terrains. Unlike
previous successful applications of humanoid locomotion learning, the
getting-up task involves complex contact patterns, which necessitates
accurately modeling the collision geometry and sparser rewards. We address
these challenges through a two-phase approach that follows a curriculum. The
first stage focuses on discovering a good getting-up trajectory under minimal
constraints on smoothness or speed / torque limits. The second stage then
refines the discovered motions into deployable (i.e. smooth and slow) motions
that are robust to variations in initial configuration and terrains. We find
these innovations enable a real-world G1 humanoid robot to get up from two main
situations that we considered: a) lying face up and b) lying face down, both
tested on flat, deformable, slippery surfaces and slopes (e.g., sloppy grass
and snowfield). To the best of our knowledge, this is the first successful
demonstration of learned getting-up policies for human-sized humanoid robots in
the real world. Project page: https://humanoid-getup.github.io/ | 36 | 67b41edb2867282b4eb37ddf | null | null |
|
2025-02-18T00:28:31.293000 | SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering? | 5 | {
"_id": "60f1abe7544c2adfd699860c",
"avatarUrl": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg",
"followerCount": 6280,
"fullname": "AK",
"isHf": true,
"isMod": false,
"isPro": false,
"name": "akhaliq",
"type": "user"
} | false | null | 2502.12115 | [
{
"_id": "67b41a72a38d04cc6148d80e",
"hidden": false,
"name": "Samuel Miserendino",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b41a72a38d04cc6148d80f",
"hidden": false,
"name": "Michele Wang",
"status": null,
"statusLastChangedAt": null,
"user": null
},
{
"_id": "67b41a72a38d04cc6148d810",
"hidden": false,
"name": "Tejal Patwardhan",
"status": "admin_assigned",
"statusLastChangedAt": "2025-02-19T15:15:20.891Z",
"user": {
"_id": "64d4f504887f55fb6eedec74",
"avatarUrl": "/avatars/054fb826890adcb330f0e4cbca3ef7c4.svg",
"fullname": "Tejal Patwardhan",
"isPro": false,
"type": "user",
"user": "tejalp"
}
},
{
"_id": "67b41a72a38d04cc6148d811",
"hidden": false,
"name": "Johannes Heidecke",
"status": null,
"statusLastChangedAt": null,
"user": null
}
]
| 2025-02-17T18:41:16 | SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance
Software Engineering? | We introduce SWE-Lancer, a benchmark of over 1,400 freelance software
engineering tasks from Upwork, valued at \1 million USD total in real-world
payouts. SWE-Lancer encompasses both independent engineering tasks--ranging
from 50 bug fixes to \$32,000 feature implementations--and managerial tasks,
where models choose between technical implementation proposals. Independent
tasks are graded with end-to-end tests triple-verified by experienced software
engineers, while managerial decisions are assessed against the choices of the
original hired engineering managers. We evaluate model performance and find
that frontier models are still unable to solve the majority of tasks. To
facilitate future research, we open-source a unified Docker image and a public
evaluation split, SWE-Lancer Diamond
(https://github.com/openai/SWELancer-Benchmark). By mapping model performance
to monetary value, we hope SWE-Lancer enables greater research into the
economic impact of AI model development. | 42 | 67b41a74a38d04cc6148d84b | null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.