Open issues for triton-inference-server (round 2)
Opened this issue · 2 comments
kpedro88 commented
Tracking the second round of issues submitted to triton-inference-server:
- triton-inference-server/server#2018: Control number of threads used by CPU server
- triton-inference-server/server#2019: Stricter model versioning
- triton-inference-server/server#2020: Ability to disable or redirect cout/cerr
- triton-inference-server/server#2021: Support input/output compression
- triton-inference-server/server#2158: Ragged batching support for ML backends
- triton-inference-server/server#2159: Support different network protocols
- triton-inference-server/server#3141: CMake problems for client library in v2.11.0
kpedro88 commented
Model checksum support has been added: https://github.com/triton-inference-server/checksum_repository_agent
kpedro88 commented
I/O compression is available and implemented in CMSSW in cms-sw/cmssw#34508