Can you release Finetune command with custom dataset?

[srn-source]

📚 The doc issue

I try to use this command but it dose not working. What should i do?

python -m olmocr.train.train --model.name_or_path Qwen/Qwen2-VL-2B --model.arch causal --train_data.sources allenai/olmOCR-mix-0225

Suggest a potential alternative/fix

No response

[aman-17]

Hey @srn-source, check out the branch jakep/new_trainer and use the olmocr/train/train.py script for fine-tuning.