v0.2.1
v0.2.1 - 2024-03-19
This release includes a critical fix to revert the default inference max sequence length back to 200 as it was in v0.1.0. A commit in v0.2.0 accidentally changes the default max sequence length to 20.
Changelog
Features 🌈
Bug Fixes 🐞
- ed345d6 fix: Protect secret with environment (#300)
- 4c4e803 fix: Update default params and add associated UTs (#294)
- 268675c fix: update manifest and helm charts (#278)
Continuous Integration 💜
- c704f84 ci: fix 1ES pool label name (#301)
- 63ff6cf ci: Update supported_models.yaml (#296)
- ab88635 ci: Add environment for pipelines (#290)
- dd59ef3 ci: Use 1ES runner for kaito workspace workflow jobs that push to ACR (#283)
Documentation 📘
- d58d22f docs: Add gpu-provisioner github repo (#267)
- 13c77e3 docs: update README.md for new models (#279)
Maintenance 🔧
- 9b33f33 chore: bump peter-evans/repository-dispatch from 1 to 3 (#269)
- c54a32b chore: bump azure/setup-helm from 3 to 4 (#270)
- 40a6e03 chore: bump actions/checkout from 3 to 4 (#271)
Security Fix 🛡️
- d5d3c57 Fix protobuf to address CVE-2024-24786
- ebf2f46 Fix fastapi to address CVE-2024-24762