v0.0.8
What's Changed
- test: add E2E correctness canary test for vLLM integration by @xiaguan in #44
- fix(connector): use correct torch device for CUDA sync in TP mode by @xiaguan in #46
- feat: add vllm-async-loading-debug skill by @xiaguan in #47
- Feat/connector preemption and logging by @xiaguan in #48
- chore: bump version to 0.0.8 by @xiaguan in #49
Full Changelog: v0.0.7...v0.0.8