Latest posts
-
Diffusion Language Models Might Have the Right Shape for Agents
Diffusion language models could give agent harnesses a useful interface for fast drafting, joint decisions, structured infilling, and continuously updated working memory.
-
From Math RLVR to Agent RL: A Practical Introduction
A practical bridge from math RLVR and GRPO to interactive agent training, covering credit assignment, exploration, rollout infrastructure, and policy freshness.
-
Reinforcement Learning for Diffusion Language Models
What is the policy when text is generated by denoising? A technical guide to the objectives, likelihood estimators, failure modes, and emerging design principles behind RL for dLLMs.
-
β-OPSD: Deriving with Policy Optimization, Training with Self-Distillation
A principled way to make on-policy self-distillation smoother, more stable, and more effective for reasoning language models.