Could you publish the loss curves for both pretraining and finetuning?

Thank you for your excellent work. I am attempting to replicate the pretraining and lora finetuning processes, and I would like to confirm whether my model's convergence is reasonable.

Review the repository's pretraining and LoRA finetuning materials and determine where the corresponding loss outputs are recorded. Document or publish representative curves for both processes so users can compare their model convergence against the project's results.

Written by the indexing model from the issue text.