alibaba / alibaba/ROLL

ROLL now supports the RLVR pipeline for Qwen2.5 VL

Open
#34 5 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
3.4k
Forks
312
Avg merge
1h 2m
Merged PRs (30d)
2

Description

Hello everyone,
Thank you for your interest in ROLL.

We're excited to share some news — ROLL has now released support for the **Qwen2.5 VL** model using the **RLVR pipeline**.

You can find a verified experimental configuration in the `./examples/qwen2.5-vl-7B-math` directory.

The example dataset is [leonardPKU/GEOQA_R1V_Train_8K](https://huggingface.co/datasets/leonardPKU/GEOQA_R1V_Train_8K). Simply download the data files to the `./data/geoqa_data` directory and you can run the demo with one click.

We welcome everyone to try it out, and we sincerely invite you to join the discussion and provide feedback.

Contributor guide

No contributing guide indexed for this repository

Research direction

The announcement points to the verified configuration in ./examples/qwen2.5-vl-7B-math and the dataset location ./data/geoqa_data; start by reviewing those paths and the linked GEOQA_R1V_Train_8K data. No requested change or acceptance criteria are stated, so the issue does not define what completion would look like.

Written by the indexing model from the issue text.

Assessment

Tech stack
huggingface, python
Domain
ai, machine-learning
Issue type
Documentation
Difficulty
1/5
Estimated time
Under an hour
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.