license, library_name, pipeline_tag, tags

license	library_name	pipeline_tag	tags
apache-2.0	transformers	text-generation	qwen3 reinforcement-learning grpo

qwen3-8b-chemistry-grpo

Fine-tuned from Qwen/Qwen3-8B with GRPO on the chemistry split.

Validation Performance

Metric: val-aux/sciknoweval/reward/mean@16 from 10-step validation logs.

best mean@16	best step	final mean@16	final step
79.49%	100	79.49%	100

step	mean@16
10	45.54%
20	59.64%
30	67.20%
40	69.05%
50	72.11%
60	73.90%
70	74.58%
80	76.52%
90	76.43%
100	79.49%

Files included with this repo:

• metrics.json: parsed validation summary
• eval_mean16.csv: step-level validation curve data
• eval_mean16.png: validation curve plot

Important: the uploaded weights are the final global_step_100/actor checkpoint. If best step is earlier than 100, the best validation point is reported for tracking, but the corresponding actor weights may not be retained locally.

Checkpoint source: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-GRPO-Qwen-Qwen3-8B-mbs8-train64-rollout8-lr1e-6-vllm0.8

W&B run: run-20260630_103227-flhubvrx

This upload uses global_step_100/actor converted from VERL FSDP shards to Hugging Face format.