About Original Dataset File

[Heaven-zhw]

Hello! Could you clarify which original train and test files in JSONL format should be used for the experiments?

In MMTab (https://huggingface.co/datasets/SpursgoZmy/MMTab/tree/main
), I could not find the appropriate test file to replace TEST_PATH.

Also, I noticed that you have already released the processed train/valid data, but I am wondering which file should be used as the input_file in create_mixed_dataset.py.

Thank you!