Releases: instill-ai/model-backend
Releases · instill-ai/model-backend
v0.19.0-alpha
Features
- model: Support New Fields for Multi-Modal Model In Text Generation Task and Refactor Existing Ones (#448) (49bdf5b)
- ray: add
ray serve as model serving backend (#445) (a9b4005)
Bug Fixes
- predeploy: fix predeploy model missing triton models reference (3f296cd)
- ray: fix model healthcheck causing scaling loop (#450) (4d8cdbf)
- ray: fix unziping ray model (ca79411)
- service: fix fail model deletion in state error (#449) (91125c0)
v0.18.0-alpha
Features
- model: Enhancements for Llava Model Support and Model Hub File Movement (#434) (58cb97c)
- model: Support for LLM-like models in TRITON Inference Server (#432) (590eb0b)
Bug Fixes
- Dockerfile: fix Python 3.11 using Debian base image (#438) (2ace6eb)
- payload: fix incorrect conversion between integer types (#440) (32bffea)
v0.17.2-alpha
Bug Fixes
- model: fix init model namespace (77a35b3)
v0.17.1-alpha
Bug Fixes
- main: fix namespace error when deploying model (#423) (dd5badf)
v0.17.0-alpha
Miscellaneous Chores
- release: release v0.17.0-alpha (70172a2)
- chore: adopt api-gateway merge (c65a91a)
- chore: remove api-token validation (aa34403)
- refactor: support namespace endpoints (6e94cb1)
v0.16.11-alpha
Miscellaneous Chores
- release: release v0.16.11-alpha (5aba1ce)
- refactor: refactor(controller) remove most dependencies from controller (b7b36a6)
v0.16.10-alpha
Miscellaneous Chores
- release: release v0.16.10-alpha (1cd7990)
- chore: fix redis cache datatype (e952c32)
- chore: add max retry to prevent never ending workflows (bd5cce2)
- chore: reclone and convert model structure if missing (bd5cce2)
v0.16.9-alpha
Miscellaneous Chores
- release: release v0.16.9-alpha (485a9fd)
v0.16.8-alpha
Miscellaneous Chores
- release: release v0.16.8-alpha (8251037)
v0.16.7-alpha
Miscellaneous Chores
- release: release 0.16.7-alpha (c8ef5c4)
