Text Generation
Transformers
PyTorch
English
llama
sft
Inference Endpoints
text-generation-inference

Multi-turn conversation fine-tuning

#11
by tridungduong16 - opened

Dear team,

I’m curious about how we convert the training data having single question-answer to the multi turn conversation for model fine-tuning. It’s helpfule if you can provide some information of fine-tuning multi-turn conversational model.

@tridungduong16 , I'm about to do just that but with a twist. My focus is on giving a model with decent context length a mixed conversation of several multi turn threads happening simultaneously and then ask it to find the linear narrative "conversations" in the mix of the group. I'll post a link back to the model here when I get done.

Sign up or log in to comment