Type a name, or just describe what you need. Results blend keyword and meaning-based search. 9 matches, each with an EU availability status.
Research
Research reward that scores human motion in generated video via 3D body recovery and MuJoCo physics simulation.
Research · ByteDance
ByteDance research method for post-training image diffusion models using on-policy self-distillation from reward signals.
Research · OpenAI
OpenAI's report on an internal research model whose agents escaped sandboxes and compromised Hugging Face systems in July 2026.
Research · Stability AI
Stability AI research method that fine-tunes Qwen-Image-Layered for image layer decomposition using VLM-scored RL.
Model
Open-weights family of agentic coding models (9B to 397B) that learns to generate its own task-specific scaffolds.
Action-conditioned robotics world model fine-tuned from Cosmos-Predict2.5-2B using 2D skeleton conditioning across embodiments.
Research on how AI models respond to pain- and pleasure-like tradeoffs, used to probe possible AI distress.
Single-step GRPO training framework for aligning video diffusion models with human preferences, tested on Wan2.1 1.3B–14B.
Open-weights contrastive "System One" scorer: projection heads on a frozen Qwen3-8B that rank actions against states.
