GitHub - BerriAI/reliableGPT: Get 100% uptime, reliability from OpenAI. Handle Rate Limit, Timeout, API, Keys Errors

🚀 Manage failed requests in Large Language Models seamlessly with GitHub - BerriAI/reliableGPT! 🤖💡 Retries with alternate models, caches responses, handles API key errors, and offers easy integration. Get 100% uptime & reliability from OpenAI. #AI #NLProc #GitHub

BerriAI/reliableGPT is a tool for managing failed requests in Large Language Models (LLMs).
It retries failed requests with alternate models like GPT-4, GPT3.5, etc.
It handles context window errors by using models with larger context windows.
Offers caching as a backup for retries and overloaded queues.
Allows users to specify fallback strategies and backup tokens for OpenAI keys.
Supports handling overloaded servers with cached responses stored in a Supabase table.
Handles rotated keys by rotating through stored keys in case of OpenAI API key errors.
Provides community support through Discord and email.
Offers easy integration with OpenAI, Azure, Langchain, and LlamaIndex.
Provides a simple two-step installation process and 1-line core package integration.
Offers advanced options like setting maximum threads and user email for alerts.
Allows for defining custom fallback strategies and model limits for rate-limited accounts.
Enables developers to specify Azure fallback strategies and maximum token/request capacities.
Supports troubleshooting with version-specific installation if needed.