Merry Christmas! 🎄🎁🎅
MemGPT version 0.2.10 includes:
- Improvements to local/open LLM performance
- Includes two new model wrappers that increase MemGPT "proactiveness" (when using local/open LLMs)
chatml-hintsandchatml-noforce-hints
- Use them by specifying them in
memgpt configureor adding them tomemgpt run- eg
memgpt run --model-wrapper chatml-noforce-hints
- eg
- Includes two new model wrappers that increase MemGPT "proactiveness" (when using local/open LLMs)
- Better visuals in the MemGPT CLI UI
- Various patches (
quickstartcommand, AutoGen, ...)
✍️ What's Changed
- docs: Added a new docs page describing how to run custom LLM parameters by @cpacker in #688
- feat: improve CLI appearance by @cpacker in #687
- fix: moved configs for hosted to https, patched bug in embedding creation by @cpacker in #685
- fix: allow edge case of quickstart before run on first install by @cpacker in #684
- fix: Remove match/case to support python <3.10 by @cpacker in #691
- fix: typo in Dockerfile comment by @tombedor in #690
- fix: memgpt agent ignores user messages by @javiersastre in #679
- fix: Better errors on over length persona/human files by @cpacker in #695
- feat: set a default temperature in the common local llm settings by @cpacker in #696
- feat: added basic heartbeat override heuristics by @cpacker in #621
- docs: updated readme for quickstart by @cpacker in #698
- fix: misc fixes by @cpacker in #700
- feat: added new 'hint' wrappers that inject hints into the pre-prefix by @cpacker in #707
Full Changelog: 0.2.9...0.2.10
👋 New Contributors
- @tombedor made their first contribution in #690
- @javiersastre made their first contribution in #679