[AGPT-698] Update cli commands and add groq integration

Significant-Gravitas / AutoGPT-Code-Ability

🖥️ AutoGPT's Coding Ability - empowering everyone to build software using AI

MIT License

109 stars 34 forks source link

[AGPT-698] Update cli commands and add groq integration #229

Closed Swiftyos closed 4 months ago

Swiftyos commented 4 months ago

Allows configuration of codex-url so we can use resume on gcp envs
Adds a quick an easy way to switch to groq and override chat model and max tokens
Added rate limiting based on requests per minute (10_000) and number of concurrent requests (1_000) defaulted to our openai limits

linear[bot] commented 4 months ago

AGPT-698

Torantulino commented 4 months ago

/review auto_approve

github-actions[bot] commented 4 months ago

Auto-approval option for PR-Agent is disabled. You can enable it via a configuration file

Torantulino commented 4 months ago

/reflect_and_review

github-actions[bot] commented 4 months ago

Questions to better understand the PR: 1) Was the removal of port configuration in CLI commands replaced entirely by the base URL configuration, or is there another mechanism to specify the port now? 2) Does the GROQ integration default to using the "llama3-70b-8192" model only when no other model is specified, or is it a fixed setting? 3) How are the new rate limits configured in the OpenAIChatClient class intended to interact with the existing system's throughput and performance?

Please respond to the questions above in the following format:

/answer 1) ... 2) ... ...

Swiftyos commented 4 months ago

/review

Swiftyos commented 4 months ago

/answer

Yes we now use url only as that includes the port
We use llama 3 when no model is specified for groq
Its meant to limit the existing system, though defaults match our limits with openai

ntindle commented 4 months ago

We are throwing errors on their side using pro

Swiftyos commented 4 months ago

/review