[PROBLEM] Multi-turn conversations: Contradiction between fine-tuning documentation and cookbook

[R-seny]

trafficstars

This file says that multi-turn conversations are collapsed and only the last message is used during fine-tuning: https://github.com/openai/openai-cookbook/blob/main/examples/How_to_finetune_chat_models.ipynb

While the documentation here contradicts that, saying that multi-turn conversations will be split into multiple examples, and that the behavior can be controlled via setting the weights.

A simple solution would be to adjust the cookbook (which, I assume, is outdated) to reflect the newly added functionality described in the documentation. Or vice-versa.

[github-actions[bot]]

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[github-actions[bot]]

This issue was closed because it has been stalled for 10 days with no activity.