merry merlin
Hi I have added the azureAssistant endpoint in my .env and in my local I am able to use the assitants also properly, as in the picture, but when I have deployed the application in the azure app service then the endpoint- azureAssistant is not being picked up, can some one tell me what is the correct way of adding the assistant for azure in the environment variables of app service so that I can fetch it properly?like ASSISTANTS_API_KEY? (BUT THIS DID NOT WORK)or what?
merry merlin
Hi used the above doc in local and it’s working
But in azure app service deployed using terraform the Assistant is not showing up as in picture from local
Any specific env variable to be added in the app service on portal??
merry merlin
Please help
torn hare
Idk if the terraform script was updated, but it was made well before the config moved to librechat.yaml, so it wouldn't pick up the config. I did my own PaaS deployment from scratch using azure app containers (which are a lot more modern and easier to manage than app service), and mounted a storage account to the main librechat image, which maps the librechat.yaml file. Then I start the image with a custom command that copies that file into ./ so it can find it upon startup.
smoky magnet
@torn hare do you have azure assistants working in your deployment? Assistants are partialy working for me. I can create new assistants and query them as a chat bot but when I try to upload files I get the error below.
"2024-06-03T18:45:10.341Z error: [/files] Error processing file: 404 Resource not found"
Here is what my librechat.yaml file looks like.
`version: 1.0.9
cache: true
interface:
privacyPolicy:
externalUrl: 'https://librechat.ai/privacy-policy'
openNewTab: true
termsOfService:
externalUrl: 'https://librechat.ai/tos'
openNewTab: true
registration:
socialLogins: ['openid']
endpoints:
azureOpenAI:
# Endpoint-level configuration
titleModel: "gpt-3.5-turbo"
plugins: true
assistants: true
summarize: true
azureAssistants:
capabilities: ["code_interpreter", "actions", "tools"]
groups:
- group: "Redacted-AIChat-EastUS"
apiKey: "${Redacted-AIchat-eastus_API_KEY}"
instanceName: "${AZURE_OPENAI_INSTANCE}"
# baseURL: "https://${INSTANCE_NAME}.openai.azure.com/openai/deployments/${DEPLOYMENT_NAME}"
assistants: true
models:
gpt-4o:
deploymentName: "AIChat-gpt-4o"
version: "2024-05-01-preview"
gpt-3.5-turbo:
deploymentName: "AIChat-gpt-35-turbo"
version: "2024-05-01-preview"
gpt-4:
deploymentName: "AIChat-gpt-4"
version: "2024-05-01-preview"`
smoky magnet
If I upload a file from the main change intercace it seems to work and I see the embedding happen in the logs but I get this error.
error: [/assistants/chat/] 400 Unknown parameter: 'file_ids'.
smoky magnet
The files show up in the Librachat Mange files interface with the Azure logo but I can not delete them.
2024-06-03 20:23:13 error: [/files] Error deleting files: {"type":"no_user_key"}
smoky magnet
@torn hare do you have azure assistants working in your deployment? If so is there any chance you can share your redacted config files with me so I can compare them against mine?
torn hare
Did you add all the config for the rag api btw?
I use this config in the .env file (in my container deployment I give these variables to the rag api image directly):
RAG_AZURE_OPENAI_API_KEY=yourapikey RAG_AZURE_OPENAI_ENDPOINT=https://instancename.openai.azure.com/ DEBUG_RAG_API=true EMBEDDINGS_PROVIDER=azure EMBEDDINGS_MODEL=text-embedding-ada-002 RAG_AZURE_OPENAI_API_VERSION=2024-03-01-preview
And then this is my librechat.yaml:
`version: 1.0.6
cache: true
interface:
privacyPolicy:
externalUrl: 'https://librechat.ai/privacy-policy'
openNewTab: true
termsOfService:
externalUrl: 'https://librechat.ai/tos'
openNewTab: true
registration:
socialLogins: ['github', 'google', 'discord', 'openid', 'facebook']
endpoints:
assistants:
disableBuilder: true
pollIntervalMs: 750
timeoutMs: 180000
capabilities: ["code_interpreter", "actions", "tools", "image_vision"]
azureOpenAI:
titleModel: "gpt-3.5-turbo-1106"
plugins: true
assistants: true
summarize: true
summaryModel: "gpt-3.5-turbo-1106"
titleConvo: true
titleMethod: "functions"
groups:
- group: "myresourcegroup"
apiKey: "myapikey"
instanceName: "myinstancename"
version: "2024-04-01-preview"
assistants: true
models:
gpt-4-turbo-2024-04-09:
deploymentName: gpt4turbo
gpt-3.5-turbo-1106:
deploymentName: gpt-35-turbo`
plush patrol
