CoRL · InternVL3_5-8B · TTRL · mmr1 (endpoint)

• Base: OpenGVLab/InternVL3_5-8B
• Method: TTRL — self-labeling GRPO (majority-vote pseudo-labels, no GT)
• Dataset: mmr1 (~8k), 1 epoch = 722 steps
• This checkpoint: near-endpoint, step 700 (exact endpoint step 722 was lost with the local disk; this is the last archived checkpoint, 22 steps before the true end — the endpoint eval numbers below are from the true step-722 checkpoint)
• Paper cell: MLLM main table (mmr1, big tier) · InternVL3_5-8B · TTRL
• WandB run: https://wandb.ai/logan-yang2002-johns-hopkins-university/co-rl-mllm-mmr1/runs/internvl35_8b_ttrl_mmr1-v1
• best repo: q1716523669/mllm-mmr1-ttrl-internvl35-8b · endpoint repo: q1716523669/mllm-mmr1-ttrl-internvl35-8b-endpoint

Eval — 4 benchmarks (prompt=answer · greedy T=0 · mathruler)