rrvaswin
/

MT_1_to_16_step200

Model card Files Files and versions

MT_1_to_16_step200 / tokenizer_config.json

Commit History

Jul 27

08:06
MT(463q x 16 heuristics) + GRPO step 200 — compute-matched to vanilla GRPO (0.128B vs 0.136B generated tokens)

rrvaswin

Verified signature
499e9ef