Chart Reasoning

Train a vision-language model on the ChartVerse-RL-40K dataset - 40K challenging chart QA samples for RL training.

Paste these prompts into Claude Code

Claude Code Prompt

Train a chart reasoning model on the ChartVerse-RL-40K dataset:
https://huggingface.co/datasets/opendatalab/ChartVerse-RL-40K

Use the chartverse template with:
- Model: Qwen/Qwen3-VL-30B-A3B-Instruct
- Reward: math_verify accuracy

Process the entire train split (40K samples).

Before starting:
1. Ensure rnow is installed: uv pip install rnow
2. Run: rnow init --template chartverse
3. Verify skills are loaded (check for rnow-* skills)
4. If skills not loaded, run: rnow init --template blank (to copy skills)