process-reward

Here are 5 public repositories matching this topic...

liuxiaotong / knowlyr-gym

Gymnasium-style RL framework for LLM agent training — MDP environments, three-layer process reward & SFT/DPO/GRPO policy optimization. CLI + MCP ready.

python cli reinforcement-learning mcp gymnasium trajectory sft dpo llm ai-data-pipeline process-reward knowlyr

Updated Mar 15, 2026
Python

waybarrios / crystal

Star

CRYSTAL: Beyond Final Answers: Benchmark for Transparent Multimodal Reasoning Evaluation | arXiv 2603.13099

benchmark reinforcement-learning computer-vision deep-learning evaluation vqa dartmouth reasoning multimodal chain-of-thought mllm llm-evaluation vision-language-models multimodal-reasoning grpo process-reward step-level-evaluation

Updated Jun 26, 2026
Python

19PINE-AI / rlvp

Star

Penalize the Path, Reward the Outcome — verifiable per-action penalties as a dense channel for deployable, sample-efficient agentic RL (GRPO). Paper: arXiv:2607.07435

machine-learning reinforcement-learning llm agentic-ai grpo rlvr verifiable-rewards process-reward