This repository has been archived by the owner on Sep 24, 2024. It is now read-only.
v0.2.1
First real release, tracks what's on PyPI. 🚀
Autogenerated notes:
What's Changed
- Port lm-tuner to Flamingo CLI by @sfriedowitz in #1
- Folder layout by @sfriedowitz in #2
- Add more structured configurations for jobs by @sfriedowitz in #3
- RD2024-9: Create and use W&B artifacts in training/evaluation runs by @sfriedowitz in #4
- RD2024-10: Dev workflow setup and guides by @sfriedowitz in #5
- RD2024-12: Replace relative w/ absolute imports in init files by @sfriedowitz in #6
- RD2024-18: Helper for creating dataset artifacts from a directory by @sfriedowitz in #7
- RD2024-50: Add structured representation to HuggingFace repos by @sfriedowitz in #9
- Create LICENSE.md by @binaryaaron in #10
- RD2024-45: Setup test structure for ITs by @sfriedowitz in #13
- Small fixes by @aittalam in #12
- RD2024-44: Structured AdapterConfig class by @sfriedowitz in #15
- CLI folder and schema generation by @sfriedowitz in #16
- Misc improvements after halting on RD2024-24 by @sfriedowitz in #19
- Adding vLLM inference entrypoints to Flamingo by @veekaybee in #18
- RD2029-79: Use Pydantic V2 by @sfriedowitz in #20
- Enable vLLM inference in Flamingo Part 2 by @veekaybee in #23
- moving all dev setup to contributing by @binaryaaron in #11
- change licence to apache 2.0 by @binaryaaron in #24
- Add support for artifact linking in lm-harness inference server by @sfriedowitz in #25
- Fully enabling inference server evaluation with vLLM in Flamingo by @veekaybee in #27
- RD2024-25: Remove ray.remote from lm-harness job by @sfriedowitz in #26
- Bump ruff to 0.2.0 by @veekaybee in #28
- Fix torch.dtype validations by @sfriedowitz in #30
- adding publishing info for public and test pypi by @binaryaaron in #31
- Adding documentation to Flamingo after adding vLLM by @veekaybee in #29
- RD2024-88: Artifact loader interface by @sfriedowitz in #33
- RD2024-89: Use artifact loader in finetuning by @sfriedowitz in #34
- RD2024-90: Make test resources even smaller by @sfriedowitz in #36
- Initial pass at Sphinx-based docs by @veekaybee in #35
- Dev branch for artifact loader interface by @sfriedowitz in #37
- RD2024-59: lm-buddy naming by @sfriedowitz in #39
- Cache Poetry + Split up Jobs by @chainlink in #38
- Several small CONTRIBUTING fixes and notebook clarifications by @veekaybee in #41
- RD2024-91: Pass over README and CONTRIBUTING by @sfriedowitz in #42
- Minor tweak to README.md by @sfriedowitz in #43
- Disclaimer for alpha development by @sfriedowitz in #52
- Adds first PyPI publishing workflow by @binaryaaron in #51
- RD2024-124: Job runner method by @sfriedowitz in #54
- RD2024-132: Reorganize layout of jobs directory by @sfriedowitz in #55
- Updated github workflows to do manual triggers from web UI for both test and prod by @binaryaaron in #53
- Consolidate workflow jobs into one job by @binaryaaron in #56
- RD2024-133: Example notebook for direct job runner by @sfriedowitz in #57
- Adding Actions workflow to build Sphinx docs by @veekaybee in #58
- RD2024-131: Job runner interface by @sfriedowitz in #59
- Fix publishing by sourcing venv differently by @sfriedowitz in #60
- Update env var name for Poetry PyPI token by @sfriedowitz in #62
- Update readme after publishing by @sfriedowitz in #63
New Contributors
- @sfriedowitz made their first contribution in #1
- @binaryaaron made their first contribution in #10
- @aittalam made their first contribution in #12
- @veekaybee made their first contribution in #18
Full Changelog: https://github.com/mozilla-ai/lm-buddy/commits/v0.2.1