Highlights
- Support files in GPTAssistants. Notebook: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_oai_assistant_retrieval.ipynb
- Richer choice of speaker selector and robustness improvement in GroupChat.
- Support custom text formats and recursive in RAG.
- Fix bugs in caching and filter checking.
- Robustness improvement in code block extraction.
- Sample web application: https://github.com/microsoft/autogen/tree/main/samples/apps/autogen-assistant.
- Adding first version of AgentEval -- a framework for assessing task utility for LLM-powered applications. Blogpost: https://microsoft.github.io/autogen/blog/2023/11/20/AgentEval.
Thanks to @IANTHEREAL @thinkall @afourney @gagb @victordibia @julianakiseleva @Narabzad @DearVa and all the other contributors!
What's Changed
- Fix typos in my affiliation by @gagb in #667
- Update index.mdx with version number by @gagb in #670
- Issue-560 by @olaoluwasalami in #578
- Sets the umask before executing the task in Docker. by @afourney in #593
- Allows users to specify a different requirements.txt file to install … by @afourney in #671
- Addresses issue 635, relating to newlines in Windows. by @afourney in #678
- Added an 'auto' mode to last_n_messages by @afourney in #693
- fix assistant creating without file by @IANTHEREAL in #689
- Fix typos in website blog by @Hao-tian-Zheng in #696
- Add RAG gptassistant example notebook by @gagb in #694
- Fix some type annotations and edge cases by @s-cerevisiae in #572
- Makes select_speaker more robust by checking for mentions anywhere. by @afourney in #669
- update side bar by @skzhang1 in #702
- Update speaker selector in GroupChat and update some notebooks by @thinkall in #688
- Update FAQ.md with api_base vs base_url question by @gagb in #672
- Re-added completion logging when using older versions of Autogen. by @afourney in #701
- add relevant dependency to the oai test workflow by @qingyun-wu in #712
- bump version to 0.2.0b6 by @sonichi in #710
- fix typos in function helper text by @shresthasurav in #503
- Sample Web Application Built with AutoGen by @victordibia in #695
- Fixes cache issue from 703 and 679 by @afourney in #707
- Support custom text formats and recursive by @thinkall in #496
- Adding first version of AgentEval -- a framework for assessing task utility for LLM-powered applications by @julianakiseleva in #681
- improve CODE_BLOCK_PATTERN for a more robust code match by @DearVa in #571
New Contributors
- @Hao-tian-Zheng made their first contribution in #696
- @s-cerevisiae made their first contribution in #572
- @skzhang1 made their first contribution in #702
- @shresthasurav made their first contribution in #503
- @julianakiseleva made their first contribution in #681
- @DearVa made their first contribution in #571
Full Changelog: 0.2.0b5...v0.2.0b6