MARTI
[ICLR 2026] A Framework for LLM-based Multi-Agent Reinforced Training and Inference
- 575
- Python
- MIT
- 3 mo ago
[ICLR 2026] A Framework for LLM-based Multi-Agent Reinforced Training and Inference
Offical Code of MICCAI'25 Best-Paper-Shortlist paper "MedGround-R1: Advancing Medical Image Grounding via Spatial-Semantic Rewarded Group Relative Policy Optimization"
Code for the 2025 ACL publication "Fine-Tuning on Diverse Reasoning Chains Drives Within-Inference CoT Refinement in LLMs"
Code to reproduce the experimental results from the EMNLP 2026 paper From Leaky Thoughts to Private Reasoning: Controlling What LRMs Say to Themselves