What's New in v0.17.0
Features
- Always fetch Hugging Face model card in llamacpp-tuner Step 3 (#151)
- Optional Step 5.5 benchmark for first-token latency and throughput (#148)
- Fetch 9router models during setup (#142, #143)
Bug Fixes
- Stop and restart llama-server when switching models (#149, #150)
- Reset base URL when Local picked after remote env export (#145)
Other Changes
- Split engine lifecycle scripts by engine (#144)
Full Changelog: v0.16.0...v0.17.0