Research Workspace
Experiments
Promote saved notebook code into quota-governed distributed PyTorchJobs, preserve run configuration, compare history, metrics, artifacts and estimated cost.
Experiments
0
Owned by you
Runs
0
Selected experiment
Active
0
Submitted/running
Recorded cost
₩0
Estimated first-hour compute
Experiment
No experiments. Create one from My Notebooks.
Run Distributed
Saved files in the notebook workspace are mounted into the PyTorchJob. Multi-node runs require an RWX workspace such as CephFS/Lustre.
Baseline vs Optimized Run · Best Run
Compare successful runs with run-scoped telemetry. Failed runs, critical-error runs and NO-DATA runs are excluded from automatic Best Run selection.
Analyze the optimization lineage to rank eligible Baseline and Optimized runs.
Best Run → Research Result Registry
Open Research ResultsPromotion preserves an immutable metadata snapshot for benchmark/model reuse. It does not overwrite a production Job Template or copy external artifact bytes.
Analyze the lineage and select an eligible Best Run first.
Artifacts
Registered artifacts
No registered artifacts for this run.
Run history & comparison
| Run | Status | Runtime | Resources | Parameters / Metrics | Cost | Evidence |
|---|---|---|---|---|---|---|
| No runs yet. | ||||||