Skip to content

Changelog

v0.1.10¤

View full release on GitHub and PyPi

Streaming, Docstring Templates, and System Message Fixes

New features around streaming for chat tasks and the use of Jinja templates in docstrings, plus a crucial fix for system messages in chat.

Features:

  • Streaming for chat tasks.
  • Support for Jinja templates in docstrings.

Fixes:

  • System message correctly set when passed through the chat decorator.

Changes:

  • Added streaming support with middleware for chat tasks as well. Closes #112.
  • Typos in README fixed.
  • Jinja templates in docstrings supported. Closes #118.
  • Fixed issue with passing system messages in chat decorator. Closes #121.

Contributors:

  • @matankley

v0.1.9¤

View full release on GitHub and PyPi

Streaming, Middleware, and API Refactoring

Introducing the support for streaming responses and middlewares in Declarai chats, along with API improvements.

Features:

  • Streaming responses for Declarai tasks enabled.
  • Middleware support for Declarai chats.

Changes:

  • Introduced middleware support to chat. Closes #105.
  • Added streaming for OpenAI tasks. Closes #107.
  • API flattened and refactored, thanks to @giladbarnea.

New Contributors:

  • @giladbarnea made their first contribution in #104

Contributors:

  • @giladbarnea
  • @matankley

v0.1.8¤

View full release on GitHub and PyPi

Changes:

  • Dual support for DECLARAI_OPEN_AI_KEY and OPENAI_API_KEY.
  • Introduced support for pydantic description field.
  • Enabled Azure OpenAI LLM. Closes #67 and #72.
  • Default Azure API version fixed.

Contributors:

  • @helmanofer
  • @matankley
  • @shobhit9957

v0.1.7¤

View full release on GitHub and PyPi

Improvements and Refactoring

A major focus on refining API Reference docs and refactoring the internal structure of the project.

Changes:

  • Introduction of gpt-3.5-turbo-16k support.
  • Internal structure refactored and API reference generation in the docs.
  • Fixes and updates in documentation.

Contributors:

  • @helmanofer
  • @matankley

v0.1.6¤

View full release on GitHub and PyPi

Bug Fix Release

Changes:

  • Fixed chat greeting duplication bug. Closes #84.

Contributors:

  • @matankley

v0.1.5¤

View full release on GitHub and PyPi

Extended Database Support

In this update, Declarai extends its support to PostgreSQL, Redis, and MongoDB databases for saving chat message history.

Changes:

  • Support for PostgreSQL, Redis, and MongoDB databases to save chat messages. Closes #77.

Contributors:

  • @matankley

v0.1.4¤

View full release on GitHub and PyPi

Introduction of Chat Memory

Declarai now supports retaining message history across chat sessions, ensuring continuity and context.

Changes:

  • Removal of operator inheritance from llm params type.
  • Added chat memory to preserve message history between sessions. Closes #62.

Contributors:

  • @matankley

v0.1.3¤

View full release on GitHub and PyPi

Enhanced Control Over LLM Params

In this version, users gain the ability to finetune the parameters of the Language Model like temperature and max tokens.

Changes:

  • Control the llm params such as temperature, max tokens, and more. Closes #54.
  • Deployment example with FastAPI for Declarai tasks added. Closes #63.
  • Support for passing LLM parameters directly to tasks. Closes #70.
  • Comprehensive documentation added for LLMParams control.
  • Updated package version.

Contributors:

  • @matankley

v0.1.2¤

View full release on GitHub and PyPi

Minor bug fixes

Changes:

  • Updates to documentation
  • Updates to dependencies with reported vulnerabilities
  • Fix typing and improve support for IDE autocompletion
  • Fix issue with initialization failing when passed the openai_token at runtime.

v0.1.1¤

View full release on GitHub and PyPi

Announcing the first release of Declarai! 🥳 🥳

Declarai was born out of the awe and excitement of LLMs, along with our passion for excellent engineering and real-world applications at scale.

We hope this project will help introduce more developers into the world of LLMs and enable them to more easily and reliably integrate these amazing capabilities into their production systems.

Main features:

  • Task interface
  • Chat interface
  • Middlewares
  • Exhaustive documentation
  • OpenAI support
  • LLM prompt best practices