Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Reorganize the project content and add/updates features #280

Open
wants to merge 6 commits into
base: main
Choose a base branch
from

Conversation

elvis-mdevo
Copy link

@elvis-mdevo elvis-mdevo commented Aug 12, 2024

UPDATES

  • reorganize project structure
  • config.yaml file is now located in configs/config.yaml
  • renforce uniqueness of roleAssignments name in bicep resources
  • [TO CHECK] we disable connector param in ClientSession instance to resolve this issue: AIOHTTP client issue with async DNS #275
    #connector=TCPConnector(resolver=AsyncResolver()),
  • [NOT COMPLETED] add the ability to disable in config.yaml unused llm tools
llm:
  excluded_llm_tools: 
    - send_sms
    - search_document # not stable / still investigating
  slow: {}
  fast: {}
  • add the ability to disable Azure content safety during the bicep deployment
make deploy-bicep deploy-post name=my-rg-name enable_content_filter=false
  • [NOT FINALIZED] add the ability to specify in config.yaml the content vector field in the search index
ai_search:
  access_key: xxx
  endpoint: xxx
  index: trainings
  vector_content_field: "contentVector"
  • add the ability to disable language choice in config.yaml at the begining of the call with enable_language_choice
conversation:
  initiate:
    task: "xxx"
    agent_phone_number: "xxx"
    bot_company: xxx
    bot_name: xxx
    enable_language_choice: false
  • [NOT COMPLETED] add the ability to show debug info during the bicep deployment
make deploy-bicep deploy-post name=my-rg-name log_type=--debug
  • [NOT COMPLETED] add the ability to exclude unused llm tools

Observations (for developers)

  • Make sure to stop the function app before function deployment
  • Please make sure you have Application Insights configured with your function app if you want to view function app logs: APPINSIGHTS_INSTRUMENTATIONKEY

@elvis-mdevo
Copy link
Author

@microsoft-github-policy-service agree

@Bergdoktor
Copy link

Hi @elvis-mdevo,
I recently managed to set up the original project after successfully overcoming various speed bumps on my way there (features not being available in my prefered regions, parameters no longer being available or being required like the content safety, openai quota exceeded, and so on).

Now that it's finally running I'm trying to improve/customize on similar things that you are mentioning in your pull request. Disabling IVR, disabling SMS feature, etc.
Love the improvements that you are mentioning/making and looking forward to them being included in the main branch!

Also just a quick question: Are you using synthetic-rag-index for supplying knowledge to AI Search? Is your vector size 0 bytes as well?

@Levis0045
Copy link

Levis0045 commented Aug 21, 2024

Hi @Bergdoktor,

Thanks you for your feedbacks. Hope that this version will be great for you.

For your question:

  • In my use case, I don't need to use synthetic-rag-index to build data injection pipeline. I just add the ability to configure ourself an index and set it to the project configuration.

It there an incidence with this pull request ?

@clemlesne clemlesne force-pushed the main branch 2 times, most recently from 88b6d54 to 9a164fb Compare August 23, 2024 20:32
@clemlesne clemlesne added the enhancement New feature or request label Aug 23, 2024
@clemlesne
Copy link
Collaborator

What already have been implemented:

  • Content Safety as a feature flag in the deployment
  • Reorganization with app and cicd folders

Would you be able to rebase your changes and send dedicated PR for each of your functional changes:

  • Folder for configuration named configs
  • Reliability for RBAC assignments (was not included in the commits if I'm not wrong)
  • Feature flag to disable the language selection in the initiate configuration

My 2 cents:

  • Have you been able to reproduce the async dns issue?
  • Is it really helpful for anyone to customize the vector field in AI Search? Data model customization is totally legit but in most cases many columns would requires to be added (ie security groups, business context, etc). In that case, just fork and edit the code.
  • What was the rational showing debug logs during bicep deployment?

Thank you by advance!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants