Help on fine tuning llama 2 using the generated arrow dataset

[bp117]

Discussed in #401

^{Originally posted by bp117 May 1, 2024}
Hi,

@ShishirPatil I ran the raft.py on the pdf document using gpt-3.5 turbo model and it generated the output files as data-xx.arrow, state.json, dataset_info.json and .jsonl files, please let me know how can i fine tune llama 2 model using these files as these are not in the format that llama 2 understands