Training setup
- Prompts:
U-WIN/equational-theory-lean2nl-references-gpt-approved-v2-rubric @ fd284c793204 (2182 problems, all TRUE; 54 finite-carrier).
- Eval:
U-WIN/sair-stage2-nl-verified-v2-rubric @ ed7fb4630dfe (205 problems, 99 TRUE / 106 FALSE), 4 samples per problem every 20 rollouts.
- Reward: rubric grading by a frozen local Qwen3.8-27B judge (policy id
v2rubric-verdict-neutral-user-rubric-20261104-v3): generic policy + two-field JSON output (assessment, score_0_to_7), reward = score/7, dedicated verdict lines removed from the candidate, no reference solution or truth label shown to the judge.
- Sampling: temperature 1.0, 32,768 response tokens, GRPO with group filtering of unresolved grades; learning rate see config.
- Weights: bf16. Only the language model is trained; the vision tower (
model.visual.*) and the multi-token-prediction head (mtp.*) are copied unchanged from the initializer (shards named model-untrained-*).
Branches
Table with columns: branch, Megatron iteration, trained tensors, carried over unchanged| branch | Megatron iteration | trained tensors | carried over unchanged |
|---|
checkpoint-10 | iter_0000009 | 427 | 348 |
checkpoint-20 | iter_0000019 | 427 | 348 |
checkpoint-30 | iter_0000029 | 427 | 348 |
checkpoint-40 | iter_0000039 | 427 | 348 |
checkpoint-50 | iter_0000049 | 427 | 348 |
checkpoint-60 | iter_0000059 | 427 | 348 |
checkpoint-70 | iter_0000069 | 427 | 348 |
checkpoint-80 | iter_0000079 | 427 | 348 |
checkpoint-90 | iter_0000089 | 427 | 348 |
checkpoint-100 | iter_0000099 | 427 | 348 |
checkpoint-110 | iter_0000109 | 427 | 348 |
checkpoint-120 | iter_0000119 | 427 | 348 |
checkpoint-130 | iter_0000129 | 427 | 348 |
checkpoint-140 | iter_0000139 | 427 | 348 |
checkpoint-150 | iter_0000149 | 427 | 348 |
checkpoint-160 | iter_0000159 | 427 | 348 |
checkpoint-170 | iter_0000169 | 427 | 348 |
checkpoint-180 | iter_0000179 | 427 | 348 |
checkpoint-190 | iter_0000189 | 427 | 348 |
checkpoint-200 | iter_0000199 | 427 | 348 |
Eval during training (sair-stage2 205 problems × 4 samples, score/7)
Table with columns: rollout, overall, TRUE problems, FALSE problems| rollout | overall | TRUE problems | FALSE problems |
|---|
| 0 | 0.420 | 0.560 | 0.288 |
| 20 | 0.469 | 0.592 | 0.354 |
| 40 | 0.465 | 0.607 | 0.333 |
| 60 | 0.460 | 0.652 | 0.282 |
| 80 | 0.497 | 0.676 | 0.329 |
| 100 | 0.497 | 0.686 | 0.320 |
| 120 | 0.505 | 0.736 | 0.289 |
| 140 | 0.527 | 0.748 | 0.321 |
| 160 | 0.487 | 0.718 | 0.271 |
| 180 | 0.492 | 0.767 | 0.235 |
Eval points are taken at rollout ids (the weights in use are at most one update behind). Conversion: miles-upstream/tools/convert_torch_dist_to_hf.py functions with the parser namespace rebuilt from the recorded training command; each branch carries conversion_manifest.json with per-shard sha256 and the max |Δ| against the initializer and the previous checkpoint.