PixVL-m3-allcaption-selfverify

Public PixVL training export from exports/rl_branch_model3_large_dpo_20260723/m3_allcaption_selfverify_cyclepreweight_50k_step90_20260728. This repository contains the complete Hugging Face model export. Source code and evaluation wrappers are available at https://github.com/StuHude/PixVL.

The model is released for research use. It is initialized from Qwen3-VL-4B-SAMTok and follows the upstream model's license and usage restrictions.

Downloads last month
18
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support