1 4 2

谢集

sanaka87

https://horizonwind2004.github.io/

AI & ML interests

Image Generation

Recent Activity

upvoted a paper about 10 hours ago

OmniEdit: Building Image Editing Generalist Models Through Specialist Supervision

upvoted a paper 2 days ago

SynthLight: Portrait Relighting with Diffusion Model by Learning to Re-render Synthetic Faces

reacted to m-ric's post with 👀 2 days ago

𝗠𝗶𝗻𝗶𝗠𝗮𝘅'𝘀 𝗻𝗲𝘄 𝗠𝗼𝗘 𝗟𝗟𝗠 𝗿𝗲𝗮𝗰𝗵𝗲𝘀 𝗖𝗹𝗮𝘂𝗱𝗲-𝗦𝗼𝗻𝗻𝗲𝘁 𝗹𝗲𝘃𝗲𝗹 𝘄𝗶𝘁𝗵 𝟰𝗠 𝘁𝗼𝗸𝗲𝗻𝘀 𝗰𝗼𝗻𝘁𝗲𝘅𝘁 𝗹𝗲𝗻𝗴𝘁𝗵 💥 This work from Chinese startup @MiniMax-AI introduces a novel architecture that achieves state-of-the-art performance while handling context windows up to 4 million tokens - roughly 20x longer than current models. The key was combining lightning attention, mixture of experts (MoE), and a careful hybrid approach. 𝗞𝗲𝘆 𝗶𝗻𝘀𝗶𝗴𝗵𝘁𝘀: 🏗️ MoE with novel hybrid attention: ‣ Mixture of Experts with 456B total parameters (45.9B activated per token) ‣ Combines Lightning attention (linear complexity) for most layers and traditional softmax attention every 8 layers 🏆 Outperforms leading models across benchmarks while offering vastly longer context: ‣ Competitive with GPT-4/Claude-3.5-Sonnet on most tasks ‣ Can efficiently handle 4M token contexts (vs 256K for most other LLMs) 🔬 Technical innovations enable efficient scaling: ‣ Novel expert parallel and tensor parallel strategies cut communication overhead in half ‣ Improved linear attention sequence parallelism, multi-level padding and other optimizations achieve 75% GPU utilization (that's really high, generally utilization is around 50%) 🎯 Thorough training strategy: ‣ Careful data curation and quality control by using a smaller preliminary version of their LLM as a judge! Overall, not only is the model impressive, but the technical paper is also really interesting! 📝 It has lots of insights including a great comparison showing how a 2B MoE (24B total) far outperforms a 7B model for the same amount of FLOPs. Read it in full here 👉 https://huggingface.co/papers/2501.08313 Model here, allows commercial use <100M monthly users 👉 https://huggingface.co/MiniMaxAI/MiniMax-Text-01

View all activity

Organizations

None yet

sanaka87's activity

upvoted a paper about 10 hours ago

OmniEdit: Building Image Editing Generalist Models Through Specialist Supervision

Paper • 2411.07199 • Published Nov 11, 2024 • 47

upvoted a paper 2 days ago

SynthLight: Portrait Relighting with Diffusion Model by Learning to Re-render Synthetic Faces

Paper • 2501.09756 • Published 3 days ago • 15

upvoted a paper 4 days ago

3DIS: Depth-Driven Decoupled Instance Synthesis for Text-to-Image Generation

Paper • 2410.12669 • Published Oct 16, 2024 • 1

upvoted a paper 5 days ago

3DIS-FLUX: simple and efficient multi-instance generation with DiT rendering

Paper • 2501.05131 • Published 10 days ago • 32