ECCV 2026 Tutorial: Post-Training Alignment and Enhancement for Diffusion Models
RisingSayak · x · 2026-09-01
A half-day tutorial at ECCV 2026 titled "Post-Training Diffusion Models: Enhancing Capabilities, Control, and Alignment" will be presented by Sayak Paul (Hugging Face) and Linoy Tsaban (Black Forest Labs). Drawing on experiences from training impactful models like Flux, the tutorial covers state-of-the-art post-training methods to adapt large pre-trained models, focusing on Alignment, Enhancement, and Control.
More from Multimodal
- Luma Labs Launches FLUX Video Upscale for 2K and 4K Video — LudovicCreator · 2026-09-01
- VLM Run Launches Unified API Gateway for 21 Visual Models — spillai · 2026-09-01
- AI Image: Renzo Piano Unconvinced Under Blue Sky — misovalko · 2026-09-01
- fal enables faster-than-realtime video gen, 300M+ asset search — noahsolomon · 2026-09-01
- LTX 2.5 Bbox IC LoRA enables spatial control — linoy_tsaban · 2026-09-01
- The best part of building a facial auto-rigger: everything looks like a face — andrew_n_carr · 2026-09-01