7 304 593

xziayro

xziayro

AI & ML interests

None yet

Recent Activity

liked a Space about 10 hours ago

gen6scp/sana-zero

liked a model about 10 hours ago

ostris/Flex.1-alpha

reacted to mkurman's post with ❤️ about 14 hours ago

Blurred-Thoughts Supervised Fine-Tuning (BT-SFT) 🤖 Can we teach a model to think completely on its own without reinforcement learning? Actually, yes. We can do straightforward supervised fine-tuning using a relatively simple trick: blurring a part of CoT thoughts. But why is this effective? We observed that various models differ in their thinking processes, and fine-tuning one model on another model’s thoughts (CoT) can sometimes be inefficient—often resulting in the model simply memorizing reasoning rather than learning how to actually think. I discovered that this process can still be efficient if we clearly indicate when the model should start and stop thinking and uncover only a part of CoT and the expected answer, blurring the other part of CoT. This approach allows the model to learn only a portion of the thought process while still arriving at an expected answer. To demonstrate this, you can watch my experimental BT-SFT on meditsolutions/Llama-3.2-SUN-2.5B-chat model, which was fine-tuned on 151 million tokens from the Magpie-Align/Magpie-Reasoning-V2-250K-CoT-Deepseek-R1-Llama-70B dataset. Enjoy! 🚀 PS. If you were curious enough to read this, leave me a comment. It's always nice to chat with open-minded and intelligent ppl.

View all activity

Organizations

None yet

xziayro's activity

upvoted an article 4 days ago

Article

Open-R1: a fully open reproduction of DeepSeek-R1

4 days ago

• 508

upvoted 2 papers 7 days ago

One-Prompt-One-Story: Free-Lunch Consistent Text-to-Image Generation Using a Single Prompt

Paper • 2501.13554 • Published 8 days ago • 9

Can We Generate Images with CoT? Let's Verify and Reinforce Image Generation Step by Step

Paper • 2501.13926 • Published 8 days ago • 31

upvoted 2 papers 9 days ago

VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding

Paper • 2501.13106 • Published 9 days ago • 76

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

Paper • 2501.12948 • Published 9 days ago • 279

upvoted 3 papers 10 days ago

Go-with-the-Flow: Motion-Controllable Video Diffusion Models Using Real-Time Warped Noise

Paper • 2501.08331 • Published 17 days ago • 20

Hunyuan3D 2.0: Scaling Diffusion Models for High Resolution Textured 3D Assets Generation

Paper • 2501.12202 • Published 10 days ago • 31

SEAL: Entangled White-box Watermarks on Low-Rank Adaptation

Paper • 2501.09284 • Published 16 days ago • 10

upvoted 2 papers 16 days ago

FramePainter: Endowing Interactive Image Editing with Video Diffusion Priors

Paper • 2501.08225 • Published 17 days ago • 18

MangaNinja: Line Art Colorization with Precise Reference Following

Paper • 2501.08332 • Published 17 days ago • 55

upvoted a paper 19 days ago

LlamaV-o1: Rethinking Step-by-step Visual Reasoning in LLMs

Paper • 2501.06186 • Published 21 days ago • 59

upvoted a paper 23 days ago

LLaVA-Mini: Efficient Image and Video Large Multimodal Models with One Vision Token

Paper • 2501.03895 • Published 24 days ago • 48

upvoted a collection 23 days ago

Sana

Collection

⚡️Sana: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer • 19 items • Updated 23 days ago • 87

upvoted 2 papers 24 days ago

Diffusion as Shader: 3D-aware Video Diffusion for Versatile Video Generation Control

Paper • 2501.03847 • Published 24 days ago • 23

Cosmos World Foundation Model Platform for Physical AI

Paper • 2501.03575 • Published 25 days ago • 68

upvoted 3 papers 25 days ago

STAR: Spatial-Temporal Augmentation with Text-to-Video Models for Real-World Video Super-Resolution

Paper • 2501.02976 • Published 25 days ago • 52

TransPixar: Advancing Text-to-Video Generation with Transparency

Paper • 2501.03006 • Published 25 days ago • 23

SLAM3R: Real-Time Dense Scene Reconstruction from Monocular RGB Videos

Paper • 2412.09401 • Published Dec 12, 2024 • 2

upvoted 2 papers 27 days ago

Nested Attention: Semantic-aware Attention Values for Concept Personalization

Paper • 2501.01407 • Published 29 days ago • 11

SeedVR: Seeding Infinity in Diffusion Transformer Towards Generic Video Restoration

Paper • 2501.01320 • Published 29 days ago • 11