---
base_model: Qwen/Qwen3-VL-4B-Thinking
library_name: transformers
pipeline_tag: image-text-to-text
tags:
  - video
  - vseek
---

# VSeek-EM

VSeek-EM is a Qwen3-VL-4B-Thinking checkpoint trained with an exact-match reward for tag-summary video question answering.

This repository contains the Hugging Face weights only.