actions-marketplace-validations/Weights & Biases

Get Runs From Weights & Biases

Trust Signals

Scorecard Score: not yet scored
Maintenance Recency: Stalelast commit Apr 14, 2020
License: MIT

Pinned Snippet

workflow.ymlSHA-pinned

uses: actions-marketplace-validations/machine-learning-apps_wandb-action@d5f0e285b4826f89e1110e1fb9e16f12deebabbb # no releases — HEAD as of 2026-07-10

tags can be moved; commit SHAs can't. why a SHA?

Inputs

name	description	required	default
WANDB_API_KEY	your W&B api key.	yes	—
PROJECT_NAME	The entity/project name associated with your wandb project. Example - 'github/predict-issue-labels'	yes	—
RUN_ID	the run id, which can be found in the url https://app.wandb.ai/{entity_name}/{project_name}/runs/{run_ID}. When supplying this input, FILTER_GITHUB_SHA and FILTER_SECONDARY_SHA are ignored and only the run corresponding to this id (along with any baselines corresponding to the input BASELINE_TAGS) are returned.	no	—
FILTER_GITHUB_SHA	The git SHA that you want to filter runs by. This assumes you have a logged a configuration variable named 'github_sha' to your runs. A common usage pattern is to supply the built-in environment variable $GITHUB_SHA, to get the commit SHA that triggered the workflow. Note that this argument is ignored if RUN_ID is specified.	—	—
FILTER_SECONDARY_SHA	This is an optional field you can filter your runs by. This assumes you have logged a configuration variable named 'secondary_sha' to your model runs. You might use this field for data versioning. Note that this argument is ignored if RUN_ID is specified.	—	""
BASELINE_TAGS	A list of tags that correspond to runs you want to retrieve in addition to those that correspond to the FILTER_GITHUB_SHA. You would typically use this field to obtain baseline runs to compare your current runs against. Example - "['baseline']"	—	[]
DISPLAY_METRICS	A list of summary metrics you want to retain for the csv file that is written to the actions environment. Example - "['acc', 'loss', 'val_acc', 'val_loss']"	—	[]
DISPLAY_CONFIG_VARS	A list of configuration variables you want to retain for the csv file written to the actions environment. Example - "['learning_rate', 'num_layers']"	—	[]
DEBUG	Setting this variable to any value will turn debug mode on.	—	""

Outputs

name	description
BOOL_COMPLETE	True if there is at least 1 finished run and no runs that have a state of 'running' else False
BOOL_SINGLE_RUN	True if there is only 1 run returned from the query else False
NUM_FINISHED	The number of non-baseline runs with a state of 'finished'
NUM_RUNNING	The number of non-baseline runs with a state of 'running'
NUM_CRASHED	The number of non-baseline runs with a state of 'crashed'
NUM_ABORTED	The number of non-baseline runs with a state of 'aborted'
NUM_BASELINES	The number of baseline runs returned.