Chen

Chen on Hugging Face Daily Papers: 15 papers, 1 in the top 3 of their day, 560 upvotes.

  1. Sol-Attn: Accelerating Video Generation Inference via On-the-Fly Attention Sparsification 35 upvotes, #8 of 2026-07-28
  2. SANA-Video 2.0: Hybrid Linear Attention with Attention Residuals for Efficient Video Generation 39 upvotes, #5 of 2026-07-24
  3. SANA-Streaming: Real-time Streaming Video Editing with Hybrid Diffusion Transformer 36 upvotes, #12 of 2026-06-01
  4. SANA-WM: Efficient Minute-Scale World Modeling with Hybrid Linear Diffusion Transformer 80 upvotes, #4 of 2026-05-15
  5. FP4 Explore, BF16 Train: Diffusion Reinforcement Learning via Efficient Rollout Scaling 34 upvotes, #5 of 2026-04-09
  6. MHLA: Restoring Expressivity of Linear Attention via Token-Level Multi-Head 46 upvotes, #4 of 2026-01-13
  7. DC-VideoGen: Efficient Video Generation with Deep Compression Video Autoencoder 33 upvotes, #8 of 2025-10-01
  8. SANA-Video: Efficient Video Generation with Block Linear Diffusion Transformer 38 upvotes, #7 of 2025-09-30
  9. DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space 6 upvotes, #48 of 2025-09-30
  10. SANA-Sprint: One-Step Diffusion with Continuous-Time Consistency Distillation 24 upvotes, #12 of 2025-03-14
  11. SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer 16 upvotes, #9 of 2025-01-31
  12. HART: Efficient Visual Generation with Hybrid Autoregressive Transformer 12 upvotes, #10 of 2024-10-21
  13. PixArt-Σ: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generation 30 upvotes, #4 of 2024-03-08
  14. PIXART-δ: Fast and Controllable Image Generation with Latent Consistency Models 47 upvotes, #1 of 2024-01-11
  15. Fast Training of Diffusion Transformer with Extreme Masking for 3D Point Clouds Generation 7 upvotes, #15 of 2023-12-13

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.