TSAIL group
- 888 followers
- FIT Building, Tsinghua University, Beijing, China
- https://ml.cs.tsinghua.edu.cn
PinnedLoading
- SageAttention
SageAttention PublicQuantized Attention that achieves speedups of 2.1-3.1x and 2.7-5.1x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.
- unidiffuser
unidiffuser PublicCode and models for the paper "One Transformer Fits All Distributions in Multi-Modal Diffusion"
- prolificdreamer
prolificdreamer PublicProlificDreamer: High-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillation (NeurIPS 2023 Spotlight)
Repositories
- SpargeAttn Public
SpargeAttention: A training-free sparse attention that can accelerate any model inference.
thu-ml/SpargeAttn’s past year of commit activity - DiffusionBridge Public
Official codebase for "Diffusion Bridge Implicit Models" (ICLR 2025) and "Consistency Diffusion Bridge Models" (NeurIPS 2024)
thu-ml/DiffusionBridge’s past year of commit activity - MMTrustEval Public
A toolbox for benchmarking trustworthiness of multimodal large language models (MultiTrust, NeurIPS 2024 Track Datasets and Benchmarks)
thu-ml/MMTrustEval’s past year of commit activity - TetraJet-MXFP4Training Public
Pytorch implementation of "Oscillation-Reduced MXFP4 Training for Vision Transformers" on DeiT Model Pre-training
thu-ml/TetraJet-MXFP4Training’s past year of commit activity - SageAttention Public
Quantized Attention that achieves speedups of 2.1-3.1x and 2.7-5.1x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.
thu-ml/SageAttention’s past year of commit activity