Submitted by akhaliq 117 Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling · 8 authors 6
Submitted by etomoscow 83 SynthDetoxM: Modern LLMs are Few-Shot Parallel Detoxification Data Annotators · 5 authors 2
Submitted by vanilla1116 47 Exploring the Limit of Outcome Reward for Learning Mathematical Reasoning · 17 authors 4
Submitted by bidiptas 29 Training Language Models for Social Deduction with Multi-Agent Reinforcement Learning · 4 authors 3
Submitted by ashraful 20 CODESIM: Multi-Agent Code Generation and Problem Solving through Simulation-Driven Planning and Debugging · 3 authors 3
Submitted by Lingaaaaaaa 17 ReasonFlux: Hierarchical LLM Reasoning via Scaling Thought Templates · 4 authors 3
Submitted by zhijie3 16 Show-o Turbo: Towards Accelerated Unified Multimodal Understanding and Generation · 6 authors 2
Submitted by Jiabin99 13 MetaChain: A Fully-Automated and Zero-Code Framework for LLM Agents · 3 authors 2
Submitted by zomss 13 Lossless Acceleration of Large Language Models with Hierarchical Drafting based on Temporal Locality in Speculative Decoding · 9 authors 3
Submitted by akhaliq 11 Lumina-Video: Efficient and Flexible Video Generation with Multi-scale Next-DiT · 19 authors 2
Submitted by Paranioar 11 EVEv2: Improved Baselines for Encoder-Free Vision-Language Models · 9 authors 2
Submitted by ztwang 11 The Hidden Life of Tokens: Reducing Hallucination of Large Vision-Language Models via Visual Information Steering · 10 authors 3
Submitted by akhaliq 8 CustomVideoX: 3D Reference Attention Driven Dynamic Adaptation for Zero-Shot Customized Video Diffusion Transformers · 12 authors 2
Submitted by PY007 7 Efficient-vDiT: Efficient Video Diffusion Transformers With Attention Tile · 7 authors 2
Submitted by zhenglin 5 DreamDPO: Aligning Text-to-3D Generation with Human Preferences via Direct Preference Optimization · 6 authors 2
Submitted by Hanyuezhuohua 5 APE: Faster and Longer Context-Augmented Generation via Adaptive Parallel Encoding · 3 authors 4
Submitted by aaabiao 4 Steel-LLM:From Scratch to Open Source -- A Personal Journey in Building a Chinese-Centric LLM · 4 authors 2
Submitted by Hhaiduo 3 Jakiro: Boosting Speculative Decoding with Decoupled Multi-Head via MoE · 10 authors 2
Submitted by dnoever 1 Forbidden Science: Dual-Use AI Challenge Benchmark and Scientific Refusal Tests · 2 authors 2