v0.2.3

What's Changed

• Bump protobuf from 3.20.0 to 3.20.2 by @dependabot in #66
• Adding information about evaluating quantized models with GPUs by @veekaybee in #65
• Trigger publishing on releases by @sfriedowitz in #67
• Update TestPyPI to only occur on manual workflow dispatch by @sfriedowitz in #68

New Contributors

• @dependabot made their first contribution in #66

Full Changelog: v0.2.2...v0.2.3

v0.2.2

What's Changed

• RD2024-134: Setuptools build by @sfriedowitz in #64

Full Changelog: v0.2.1...v0.2.2

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