Metrics: official VQA accuracy (VQAv2/TextVQA), normalized EM (GQA/CLEVR); harness cross-checked against lmms-eval. Full tables, configs, and per-run JSONs: https://github.com/guangboyu/vqa-rlvr.

Usage

Short template: 83.0 / 63.6 / 92.8 / 85.0
Reasoning template: 70.2 / 50.8 / 96.2 / 75.4

python

from peft import PeftModel
from transformers import AutoModelForImageTextToText

base = AutoModelForImageTextToText.from_pretrained("Qwen/Qwen3-VL-8B-Instruct", dtype="bfloat16")
model = PeftModel.from_pretrained(base, "omnifish123/vqa-rlvr-sft-8b").merge_and_unload()

Training config is in run_config.json in this repo.