Skip to content

Acceptance criteria for training (RetinaNet) accelerator utilization #513

Description

@rasar07

We've noticed a discrepancy between the documentation and a configuration file.
https://github.com/mlcommons/storage/blob/main/configs/dlio/workload/retinanet_b200.yaml states
metric:
au: 0.85

whereas the table in the Benchmark Overview section of https://github.com/mlcommons/storage/blob/main/training/README.md lists Minimum AU% as 90%.

Could you please clarify?

Also, the link to configurations in the Benchmark Overview section does not work:

The benchmark suite provides workload configurations that simulate the I/O patterns of selected workloads listed in Table 1.

Thank you!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions