Skip to main content
W&B stores your script’s stdout and stderr as output.log (or multipart chunks under logs/). Where to retrieve it depends on whether the run finished, is still active, or crashed. For capture settings and console_multipart, see Console logs.

​
From the UI

  1. Open the run page.
  2. Click the Files tab.
  3. Find output.log (or files under logs/) and click the download icon.
By default, output.log uploads when the run finishes. It may not appear on Files while the run is still active unless you enabled multipart logging at init time.

​
Programmatically via the API

import wandb

api = wandb.Api()
run = api.run("entity/project/run_id")
run.file("output.log").download(replace=True)
For multipart runs, list and download individual files under logs/ the same way.

​
While the run is still running

Enable multipart console logging at wandb.init time. Use wandb v0.22.3 or later so the SDK can write and upload timestamped chunks under logs/ while the run is still active. Upload cadence cannot be changed after the run starts.

​
If the run crashed

Without multipart logging, a crashed run may have no output.log on the server. Chunks uploaded before the crash remain downloadable when multipart was enabled. Check the local copy at wandb/run-<timestamp>-<run-id>/logs/output.log if server-side files are missing.

​
Resuming a run

Without multipart logging, wandb.init(resume="allow", id=...) can overwrite a single output.log. With console_multipart=True (requires wandb v0.22.3 or later to enable multipart logging) and wandb v0.20.1 or later for correct chunk uploads when you resume, each session keeps its own chunks. See Console logs.

​
More lines than the Logs tab shows

The Logs tab shows at most 10,000 lines at a time for performance. Download from Files or the API for the full log. For capture issues, see Why is console output not captured for my run?.
Logs Runs