- Start Triton server
- Update Triton GRPC host and port in config.env
- Set the preferred concurrency number in config.env
- Run
python preprocessing.py - Run
make app - To benchmark run
curl -v localhost:8000/benchmark -d '{}' - Update the
input_filevariable (on line 87) in postprocessing.py to the output file path from step 4 - Run
python postprocessing.pyto generate thecleaned_output.csvfile that has accuracy metrics
-
Notifications
You must be signed in to change notification settings - Fork 1
theBeginner86/triton-perf
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
About
A distributed performance benchmark engine for ASR workloads on Triton Inference Servers
Topics
Resources
Stars
Watchers
Forks
Packages 0
No packages published