search

microsoft / teams-ai

SDK focused on building AI based applications and extensions for Microsoft Teams and other Bot Framework channels
MIT License
427 stars 183 forks source link

[Feature Request]: Specify Deployment Name for Azure OpenAI Embedding Models #1406

Open linjungz opened 8 months ago

linjungz commented 8 months ago

Scenario

In the js samples Microsoft Teams Conversational Bot with AI: Teams Chef, the deployment name for Azure OpenAI Embedding model is hardcoded as "embedding":

https://github.com/microsoft/teams-ai/blob/f9b1014dcb071b83353498964999663f863760bc/js/samples/04.ai.a.teamsChefBot/src/VectraDataSource.ts#L78-L88

Kindly refer to this issue: https://github.com/microsoft/teams-ai/issues/1162

Solution

Instead of hardcoding the deployment name, we should put this in .env and load it when initializing VectraDataSource

https://github.com/microsoft/teams-ai/blob/f9b1014dcb071b83353498964999663f863760bc/js/samples/04.ai.a.teamsChefBot/src/index.ts#L118-L127

Add one more option, say azureEmbedding, similar to what we do with apiKey or apiEndpoint

This should make it flexible for the user to update the correct deployment name for embedding model if they're not using embedding by default.

Additional Context

No response

singhk97 commented 7 months ago

I like the idea, feel free to open up a PR with the suggested changes.