Hi,
I use the openai llm model in production, and we would like to incorporate conversational history with the ChatML approach, with the User/Assistant alternating streams that openai supports:
[{"role": "user", "content": messge1},{"role": "assistant", "content": messge2},{"role": "user", "content": messge3},{"role": "assistant", "content": messge4},]
Such that we could provide a list of dicts and update the sent request by appending before the request with the current prompt
|
{"messages": [{"role": "user", "content": prompt}]} |
So I was wondering if that parameter could be added to the OPENAI rest model
Hi,
I use the openai llm model in production, and we would like to incorporate conversational history with the ChatML approach, with the User/Assistant alternating streams that openai supports:
[{"role": "user", "content": messge1},{"role": "assistant", "content": messge2},{"role": "user", "content": messge3},{"role": "assistant", "content": messge4},]Such that we could provide a list of dicts and update the sent request by appending before the request with the current prompt
spacy-llm/spacy_llm/models/rest/openai/model.py
Line 129 in 117f689
So I was wondering if that parameter could be added to the OPENAI rest model