This dataset is a transformed version of the GAIR/LIMA dataset, specifically formatted for training with TRL (Transformers Reinforcement Learning).
This dataset contains conversations from the GAIR/LIMA dataset, transformed into a message-based format suitable for conversational AI training and reinforcement learning.
Converted from Q&A format to multi-turn conversation format
Applied… See the full description on the dataset page:
https://huggingface.co/datasets/essobi/trl_gair_lima.
