Tags: antonioleonardo/req_llm
Tags
Release version 1.0.0-rc.2 with significant enhancements and fixes - Added a model metadata guide and a local patching system for model synchronization. - Introduced a centralized `ReqLLM.Keys` module for unified API key management. - Updated bang methods to return naked values, improving API usability. - Enhanced provider architecture with a new `translate_options/3` callback for model-specific parameter handling. - Improved documentation across various files for clarity and accuracy. - Integrated ExCoveralls for test coverage reporting and established automated dependency management with Dependabot. This release focuses on improving flexibility, usability, and documentation within the ReqLLM library.
Refactor request preparation and response handling for providers - Improved the `prepare_request` function in the `ReqLLM.Providers.Anthropic` module to handle unsupported operations more gracefully. - Enhanced the `ReqLLM.Generation` module by formatting the request preparation for better readability. - Streamlined the `decode_object_stream` function in the `ReqLLM.Response` module to simplify the decoding process. - Updated various provider implementations to ensure consistent handling of options and parameters. - Refined tests for object generation and streaming to validate the new request handling logic and ensure robust response structures.