Hugging Face Journal Club derives the math behind score centering for off-policy RL
Hugging Face · youtube · 2026-10-26
Kashif Rasul from the Hugging Face research team gives an in-depth mathematical derivation of the score centering paper (2609.20807), which stabilizes off-policy reinforcement learning, followed by team discussion. Slides show score centering keeping training stable across low-precision combos like FP8/FP4 and INT8/INT4 where baselines collapse.
More from Research
- RCT: ChatGPT temporarily boosts consultants' data science skills, gains vanish without it — EricTopol · 2026-10-26
- FocusVLA routes robot camera views by predicted interaction intent across task stages — iscas-cloud · 2026-10-26
- Transformers' biggest audio overhaul: unified AudioProcessor architecture — LysandreJik · 2026-10-26
- Free 195-page primer: The Little Book of Generative AI Foundations — caglar_ee · 2026-10-26
- SPADER from Zhejiang and Peking University Tops All Four Multi-Answer QA Benchmarks, Accepted at EMNLP — jiqizhixin · 2026-10-26
- Purdue's GPS-free autonomous drone tracks runners through dense forests using on-chip LiDAR AI — KordingLab · 2026-10-26