15 89 374

alkinun

AtAndDev

AI & ML interests

LLMs, Alignment, Merging, Unsloth, DPO, SFT, ORPO, SPIN..

Recent Activity

liked a model 16 days ago

OpenGVLab/InternVL3-14B

upvoted a paper 16 days ago

InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models

View all activity

Organizations

AtAndDev's activity

reacted to hesamation's post with 👍🔥❤️ 16 days ago

Post

2892

this paper has been blowing up
they train an open-source multimodal LLM (InternVL3) that can compete with GPT-4o and Claude 3.5 Sonnet by:
> training text and vision on a single stage
> a novel V2PE positional encoding
> SFT & mixed preference optimization
Paper: InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models (2504.10479)
> test-time scaling

replied to Steven10429's post 23 days ago

It seems like they randomly filter out some ppl for some reason...

reacted to AdinaY's post with 👍 23 days ago

Post

1827

MAYE🎈a from-scratch RL framework for Vision Language Models, released by GAIR - an active research group from the Chinese community.

✨Minimal & transparent pipeline with standard tools
✨Standardized eval to track training & reflection
✨Open Code & Dataset

Code:
https://github.com/GAIR-NLP/MAYE?tab=readme-ov-file
Dataset:
ManTle/MAYE
Paper:
Rethinking RL Scaling for Vision Language Models: A Transparent, From-Scratch Framework and Comprehensive Evaluation Scheme (2504.02587)

1 reply

reacted to merterbak's post with ❤️👀🔥 23 days ago

Post

4739

Qwen 3 can launch very soon. 👀

https://github.com/ggml-org/llama.cpp/pull/12828

3 replies

replied to their post 24 days ago

No one has

posted an update 26 days ago

Post

2939

Llama 4 is out...

3 replies

reacted to BestWishYsh's post with 👀🔥 27 days ago

Post

2672

🚨 Hot Take: GPT-4o might NOT be a purely autoregressive model! 🚨

There’s a high chance it has a diffusion head. 🤯 If true, this could be a game-changer for AI architecture. What do you think? 🤔👇

Code: https://github.com/PicoTrex/GPT-ImgEval
Dataset: Yejy53/GPT-ImgEval
Paper: GPT-ImgEval: A Comprehensive Benchmark for Diagnosing GPT4o in Image Generation (2504.02782)

reacted to clem's post with 🔥 27 days ago

Post

1983

Llama models (arguably the most successful open AI models of all times) just represented 3% of total model downloads on Hugging Face in March.

People and media like stories of winner takes all & one model/company to rule them all but the reality is much more nuanced than this!

Kudos to all the small AI builders out there!

2 replies

reacted to Jaward's post with 🔥 27 days ago

Post

2336

Amazing work👏
Introduces Dream 7B - a discrete diffusion reasoning model, fully opensourced with weights on 🤗
- it outperforms existing non-autoregressive models and matches or beats frontier autoregressive of similar size on reasoning tasks.
Models:
- base: Dream-org/Dream-v0-Base-7B
- SFT: Dream-org/Dream-v0-Instruct-7B
Code: https://github.com/HKUNLP/Dream
Project: https://hkunlp.github.io/blog/2025/dream/

1 reply

reacted to openfree's post with ❤️👀🚀🔥 about 1 month ago

Post

8286

🔥 'Open Meme Studio': Your Creative Meme Factory 🎭✨

Hello everyone! Today I'm introducing 'Open Meme Studio', an amazing space where you can easily create and transform fun and original meme images. 🚀

VIDraft/Open-Meme-Studio

🎯 Taking Meme Creation to the Next Level!
This application leverages the powerful Kolors model and IP-Adapter-Plus to upgrade your meme-making abilities. Go beyond simple image editing and experience a completely new meme world powered by AI!

🛠️ Features You'll Love

📸 Transform and reinterpret existing meme templates
🎭 Freely change expressions and poses
👓 Add props (sunglasses, hats, etc.)
🏞️ Change backgrounds and composite characters
🎨 Apply various artistic styles

💪 Why 'Open Meme Studio' is So Effective

Fast Meme Generation: High-quality memes completed in seconds
Unlimited Creativity: Completely different results just by changing prompts
User-Friendly Interface: Simple prompt input and image upload is all you need
Fine-tuned Control: Adjust how much of the original image characteristics to preserve
Advanced User Options: Freely set seed values, resolution, number of steps, and more

🚀 Streamlined Meme Creation Process
Tasks that previously required complex tools like Photoshop can now be accomplished with just a few simple prompts. Experience intuitive image manipulation through text commands.

🌈 Effective Prompt Examples

😎 "sunglass" - Add cool sunglasses to your character
🏔️ "background alps" - Change the background to Alpine mountains
💃 "dancing" - Transform your character into a dancing pose
😁 "smile" - Change to a smiling expression
🎮 "with Pikachu" - Create a scene with Pikachu
🎨 "3d style" - Convert to 3D style

🔗 Join Our Community
For more meme creation tips and interaction with other users, join our Discord!
https://discord.gg/openfreeai

Start creating unique memes that will shake up social media with 'Open Meme Studio' right now! 🚀💯 It's time for your meme

3 replies

reacted to burtenshaw's post with 🚀❤️ about 1 month ago

Post

3114

NEW UNIT in the Hugging Face Reasoning course. We dive deep into the algorithm behind DeepSeek R1 with an advanced and hands-on guide to interpreting GRPO.

🔗

reasoning-course

This unit is super useful if you’re tuning models with reinforcement learning. It will help with:

- interpreting loss and reward progression during training runs
- selecting effective parameters for training
- reviewing and defining effective reward functions

This unit also works up smoothly toward the existing practical exercises form @mlabonne and Unsloth.

📣 Shout out to @ShirinYamani who wrote the unit. Follow for more great content.

1 reply