logan7000/mllm-cogrpo-heter-qwen25vl-7b-x-internvl35-8b-mmr1-groupB-internvl35-8b-endpoint

q1716523669/mllm-cogrpo-heter-qwen25vl-7b-x-internvl35-8b-mmr1-groupB-internvl35-8b-endpoint

Cross-family co-learning (co-GRPO-DP): trained jointly with peer Qwen/Qwen2.5-VL-7B-Instruct via majority-vote cross-labeling — no ground-truth labels are used.

• —Base: OpenGVLab/InternVL3_5-8B-HF · Peer: Qwen/Qwen2.5-VL-7B-Instruct
• —Method: co-GRPO-DP (heterogeneous co-learning, rendezvous cross-pseudo-labels)
• —Dataset: mmr1 (~8k), 1 epoch = 722 steps
• —Checkpoint: endpoint (checkpoint-722)
• —4-bench avg: 53.85 (MathVision / MathVerse / MathVista / We-Math; prompt=answer, greedy T=0, mathruler)
• —WandB project: co-rl-mllm-mmr1