Modelpublic

SeongryongJung/qwen3-4b-biology-rlsd-ema005

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face

0likes14downloads

Model Card

qwen3-4b-biology-rlsd-ema005

Fine-tuned from Qwen/Qwen3-4B with RLSD (EMA 0.05) on the biology 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
53.50%	90	41.50%	100

[image]

step	mean@16
10	35.75%
20	36.25%
30	39.25%
40	43.88%
50	51.25%
60	52.75%
70	45.12%
80	48.25%
90	53.50%
100	41.50%

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/biology/qwen3gen-biology-RLSD-Qwen-Qwen3-4B-mbs8-decay0-ema0.05-train64-rollout8-lr1e-6-vllm0.8

W&B run: run-20260629_225016-1l107kq8

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