You can deploy this model and start earning money today!

q1716523669/
unmaj-qwen3-1p7b-math345-end

Name: unmaj-qwen3-1p7b-math345-end
Availability: InStock

reinforcement-learning

q1716523669/MATH-Level345

Qwen/Qwen3-1.7B-Base

finetune

apache-2.0

Views

No views yet

unmaj-qwen3-1p7b-math345 · end checkpoint (step 136)

Fine-tuned from Qwen/Qwen3-1.7B-Base with TTRL / un-GRPO-maj: no labels, trains against the model's own self-consistency majority over 12 rollouts.

Training data: q1716523669/MATH-Level345 (8,740 problems, MATH levels 3-5)
Eval: MATH-500 (500 prompts, pass@1, temperature 0.6)
This checkpoint: step 136 of 136 (2 epochs) — the end one
Its score: 0.6548

best is the step with the highest eval; end is step 136. Both are published because they answer different questions and the gap between them is a result in itself. The sibling repo is unmaj-qwen3-1p7b-math345-best.

Hyperparameters

Effective batch 128 prompts per optimizer step, lr 3e-6, 2 epochs, 12 generations per prompt, 3072-token completions, temperature 1.0 for rollouts and 0.6 for eval, beta 0, bnpo loss, scale_rewards group, seed 42.

Eval curve

step	MATH-500 pass@1
0	0.5437
10	0.6091
20	0.6329
30	0.6389
40	0.6587
50	0.6429
60	0.6627
70	0.6647
80	0.6667
90	0.6448
100	0.6647
110	0.6567
120	0.6369
130	0.6548

Files for audit

train.log — the complete training log this checkpoint came from
eval_curve.csv — the table above, machine-readable
run_config.json — resolved config as the trainer saw it

Code: williamium3000/trl-projects, branch n3-interaction-modes, projects/co-grpo-dp/.

q1716523669/unmaj-qwen3-1p7b-math345-end

unmaj-qwen3-1p7b-math345 · end checkpoint (step 136)

Hyperparameters

Eval curve

Files for audit

q1716523669/
unmaj-qwen3-1p7b-math345-end