Skip to content

Conversation

@daveey
Copy link
Contributor

@daveey daveey commented Dec 8, 2025

cp

feat: add skydeck experiment dashboard

Skydeck is a web dashboard for tracking and managing ML experiments with:

  • SQLite-backed experiment state management
  • W&B integration for experiment syncing
  • Flag-based experiment configuration
  • Web UI for monitoring experiment status

Update pyproject.toml and uv.lock

Commit skydeck files

Copy link
Contributor Author

daveey commented Dec 8, 2025

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

This stack of pull requests is managed by Graphite. Learn more about stacking.

@daveey daveey changed the title cp skydeck Dec 8, 2025
@daveey daveey force-pushed the daveey-cvc-diversity branch from aec5381 to 4eab0c8 Compare December 8, 2025 04:49
@daveey daveey force-pushed the daveey-cvc-diversity branch from 4eab0c8 to c965d51 Compare December 8, 2025 07:44
@daveey daveey force-pushed the daveey-skytower branch 2 times, most recently from acc2427 to 842df15 Compare December 8, 2025 08:41
@daveey daveey force-pushed the daveey-cvc-diversity branch 2 times, most recently from 805624d to 4863989 Compare December 8, 2025 08:56
@daveey daveey force-pushed the daveey-skytower branch 3 times, most recently from 2d56b2c to cfbb106 Compare December 8, 2025 09:17
Skydeck is a web dashboard for tracking and managing ML experiments with:
- SQLite-backed experiment state management
- W&B integration for experiment syncing
- Flag-based experiment configuration
- Web UI for monitoring experiment status
@daveey daveey force-pushed the daveey-cvc-diversity branch from 4863989 to b52bd87 Compare December 8, 2025 20:11
@daveey daveey marked this pull request as ready for review December 10, 2025 21:57
@openhands-ai
Copy link

openhands-ai bot commented Dec 10, 2025

Looks like there are a few issues preventing this PR from being merged!

  • GitHub Actions are failing:
    • Validate pyproject.toml
    • Validate pyproject.toml

If you'd like me to help, just leave a comment, like

@OpenHands please fix the failing actions on PR #4237 at branch `daveey-skytower`

Feel free to include any additional details that might help me get this PR into a better state.

You can manage your notification settings

Copy link

@chatgpt-codex-connector chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

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

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment on lines +142 to +143
# Always add run name using experiment ID
parts.append(f"run={self.id}")

Choose a reason for hiding this comment

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

P1 Badge Honor run_name when building experiment command

The command builder always appends run={self.id} and there is no hook for the run_name field even though the schema and tables include a run_name column and tests construct experiments with a separate run name. Whenever a caller supplies a custom run_name (e.g., to match existing W&B runs), it is silently ignored and the command always uses the experiment id instead, so generated runs cannot be aligned with the requested name.

Useful? React with 👍 / 👎.

Comment on lines +249 to +257
async def update_experiment_state(self, experiment_id: str, current_state: JobStatus):
"""Update experiment current state."""
await self._conn.execute(
"""
UPDATE experiments
SET current_state = ?, updated_at = ?
WHERE id = ?
""",
(current_state.value, datetime.utcnow().isoformat(), experiment_id),

Choose a reason for hiding this comment

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

P1 Badge Persist current_job_id when updating experiment state

The update_experiment_state method only takes a status and updates current_state, ignoring the current_job_id column that exists in the schema and is passed by callers/tests. Invocations that provide a job id raise a TypeError, and even two-arg calls never record which job produced the state, so the database cannot track the active job for an experiment.

Useful? React with 👍 / 👎.

@daveey daveey marked this pull request as draft December 14, 2025 19:34
@daveey daveey closed this Dec 19, 2025
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.

2 participants