Including OpenAI and Azure OpenAI, there are quite a lot of LLM models user can use for their AI scenarios. Sometimes they might fine tune a minor model and want to apply it to their Teams bot app. In such scenario, shall we support user-defined LLM models rather than only OAI and AzOAI?
since we export the PromptCompletionModel interface devs can implement their own models in a compatible way the same way we do. An example of this is our LLaMaModel.
Scenario
Including OpenAI and Azure OpenAI, there are quite a lot of LLM models user can use for their AI scenarios. Sometimes they might fine tune a minor model and want to apply it to their Teams bot app. In such scenario, shall we support user-defined LLM models rather than only OAI and AzOAI?
Solution
Make Python SDK support user-defined LLM models
Additional Context
No response