Skip to content

Update submission directory to align with new rules.#12

Draft
arav-agarwal2 wants to merge 2 commits into
mainfrom
Update-submission-directory
Draft

Update submission directory to align with new rules.#12
arav-agarwal2 wants to merge 2 commits into
mainfrom
Update-submission-directory

Conversation

@arav-agarwal2

Copy link
Copy Markdown
Collaborator

Do not merge - this is just for internal review.

@github-actions

github-actions Bot commented May 28, 2026

Copy link
Copy Markdown

MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅

for (system_id, model), runs in groups.items():
all_system_runs = runs_by_system[system_id]
if system_id not in written_systems:
max_concurrency = max(_extract_concurrency(r["config"]) for r in all_system_runs)

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is incorrect. The max concurrency is itself a field within the submission structure. I failed to notice that before, but I'll make that change when we get clarification.

if _normalize_division(division) == "Standardized":
(submission_dir / "src").mkdir(exist_ok=True)
_write_run_entries(submission_dir, system_id, model, runs, max_concurrency)
_write_model_sweep_stubs(submission_dir, system_id, model)

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This function definitely shouldn't exist. We'll need to confirm what the csv files are, and generate them on the fly, probably.

json.dumps(run["result_summary"], indent=2), encoding="utf-8"
)
(result_dir / "mlperf_endpoints_log_detail.json").write_text(
(run_dir / "mlperf_endpoints_log_detail.json").write_text(

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This line is also suspect. I can't believe I missed this earlier, but it's worrying that we build a submission with an empty directory here. We need to use real data.

json.dumps(
{
"serving_framework": run["system_info"].get("framework", ""),
"parallelism": {},

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same here.

src_dir.mkdir(parents=True, exist_ok=True)
(src_dir / ".gitkeep").write_text("", encoding="utf-8")

# Per-run accuracy placeholder

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same here. This should be sourced from our database. If that isn't the case, we need to change things to match.



def _write_accuracy_placeholders(
def _write_model_sweep_stubs(

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Noting this again - we'll need to change this to an actual csv creator.

]

_DEFAULT_BASE_URL = "http://localhost:8080"
_DEFAULT_BASE_URL = "https://mlperf-endpoints-api-50577619532.us-central1.run.app"

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note to self: See if there's a better way to store this instead of in the source code. I don't think there will be, but better to check.

Pulling new changes in.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant