Improvements to ANN Benchmark Python scripts and docs #1734
Conversation
divyegala
added
improvement
cjnolet
changed the title
| if args.configuration: | ||
| conf_filepath = args.configuration | ||
| else: | ||
| conf_filepath = os.path.join(scripts_path, "conf", f"{args.dataset}.json") |
There was a problem hiding this comment.
Can we enforce the constraint here that either the dataset or the configuration argument needs to be specified (one of them is required but both can't be specified).
There was a problem hiding this comment.
Since dataset has a default, that enforces that one of them is specified. Should I remove the default and throw an error instead if both are unspecified?
| @@ -34,6 +34,7 @@ dependencies: | |||
| - nccl>=2.9.9 | |||
| - ninja | |||
| - nlohmann_json>=3.11.2 | |||
| - pyyaml | |||
There was a problem hiding this comment.
Matplotlib doesn't seem to be getting installed as a dependency of the libraft-ann-bench conda package.
There was a problem hiding this comment.
Yes, it's part of the bench-ann environment file. It should not be a dependency of the libraft-ann-bench conda package because that package is just the executables
Co-authored-by: Corey J. Nolet <[email protected]>
|
/merge |
No description provided.