The training sample includes 2053 conversations;
The test sample includes 229 conversations;
topic - medical topic
context - user-ai message history
user_question - last user question
assistant_answer - ai answer according the context and topic
prompt - ready prompt for fincetuning instruct model (adapted for using with unsloth… See the full description on the dataset page:
https://huggingface.co/datasets/Mykes/rus_med_dialogues.
