Documentation
Search…
wandb.Run
A unit of computation logged by wandb. Typically, this is an ML experiment.
Run(
settings: Settings,
config: Optional[Dict[str, Any]] = None,
sweep_config: Optional[Dict[str, Any]] = None,
launch_config: Optional[Dict[str, Any]] = None
) -> None
Create a run with wandb.init():
import wandb
run = wandb.init()
There is only ever at most one active wandb.Run in any process, and it is accessible as wandb.run:
import wandb
assert wandb.run is None
wandb.init()
assert wandb.run is not None
anything you log with wandb.log will be sent to that run.
If you want to start more runs in the same script or notebook, you'll need to finish the run that is in-flight. Runs can be finished with wandb.finish or by using them in a with block:
import wandb
wandb.init()
wandb.finish()
assert wandb.run is None
with wandb.init() as run:
pass # log data here
assert wandb.run is None
See the documentation for wandb.init for more on creating runs, or check out our guide to wandb.init.
In distributed training, you can either create a single run in the rank 0 process and then log information only from that process, or you can create a run in each process, logging from each separately, and group the results together with the group argument to wandb.init. For more details on distributed training with W&B, check out our guide.
Currently, there is a parallel Run object in the wandb.Api. Eventually these two objects will be merged.
Attributes
Text
summary
(Summary) Single values set for each wandb.log() key. By default, summary is set to the last value logged. You can manually set summary to the best value, like max accuracy, instead of the final value.
config
Returns the config object associated with this run.
dir
Returns the directory where files associated with the run are saved.
entity
Returns the name of the W&B entity associated with the run. Entity can be a user name or the name of a team or organization.
group
Returns the name of the group associated with the run. Setting a group helps the W&B UI organize runs in a sensible way. If you are doing a distributed training you should give all of the runs in the training the same group. If you are doing crossvalidation you should give all the crossvalidation folds the same group.
id
Returns the identifier for this run.
mode
For compatibility with 0.9.x and earlier, deprecate eventually.
name
Returns the display name of the run. Display names are not guaranteed to be unique and may be descriptive. By default, they are randomly generated.
notes
Returns the notes associated with the run, if there are any. Notes can be a multiline string and can also use markdown and latex equations inside $, like $x + 3$.
path
Returns the path to the run. Run paths include entity, project, and run ID, in the format entity/project/run_id.
project
Returns the name of the W&B project associated with the run.
resumed
Returns True if the run was resumed, False otherwise.
settings
Returns a frozen copy of run's Settings object.
start_time
Returns the unix time stamp, in seconds, when the run started.
starting_step
Returns the first step of the run.
step
Returns the current value of the step. This counter is incremented by wandb.log.
sweep_id
Returns the ID of the sweep associated with the run, if there is one.
tags
Returns the tags associated with the run, if there are any.
url
Returns the W&B url associated with the run.

Methods

alert

alert(
title: str,
text: str,
level: Union[str, 'AlertLevel'] = None,
wait_duration: Union[int, float, timedelta, None] = None
) -> None
Launch an alert with the given title and text.
Arguments
Text
title
(str) The title of the alert, must be less than 64 characters long.
text
(str) The text body of the alert.
level
(str or wandb.AlertLevel, optional) The alert level to use, either: INFO, WARN, or ERROR.
wait_duration
(int, float, or timedelta, optional) The time to wait (in seconds) before sending another alert with this title.

define_metric

define_metric(
name: str,
step_metric: Union[str, wandb_metric.Metric, None] = None,
step_sync: bool = None,
hidden: bool = None,
summary: str = None,
goal: str = None,
overwrite: bool = None,
**kwargs
) -> wandb_metric.Metric
Define metric properties which will later be logged with wandb.log().
Arguments
Text
name
Name of the metric.
step_metric
Independent variable associated with the metric.
step_sync
Automatically add step_metric to history if needed. Defaults to True if step_metric is specified.
hidden
Hide this metric from automatic plots.
summary
Specify aggregate metrics added to summary. Supported aggregations: "min,max,mean,best,last,none" Default aggregation is copy Aggregation best defaults to goal==minimize
goal
Specify direction for optimizing the metric. Supported directions: "minimize,maximize"
Returns
Text
A metric object is returned that can be further specified.

detach

detach() -> None

display

display(
height: int = 420,
hidden: bool = (False)
) -> bool
Displays this run in jupyter.

finish

finish(
exit_code: int = None,
quiet: Optional[bool] = None
) -> None
Marks a run as finished, and finishes uploading all data.
This is used when creating multiple runs in the same process. We automatically call this method when your script exits or if you use the run context manager.
Arguments
Text
exit_code
Set to something other than 0 to mark a run as failed
quiet
Set to true to minimize log output

finish_artifact

finish_artifact(
artifact_or_path: Union[wandb_artifacts.Artifact, str],
name: Optional[str] = None,
type: Optional[str] = None,
aliases: Optional[List[str]] = None,
distributed_id: Optional[str] = None
) -> wandb_artifacts.Artifact
Finishes a non-finalized artifact as output of a run.
Subsequent "upserts" with the same distributed ID will result in a new version.
Arguments
Text
artifact_or_path
(str or Artifact) A path to the contents of this artifact, can be in the following forms: - /local/directory - /local/directory/file.txt - s3://bucket/path You can also pass an Artifact object created by calling wandb.Artifact.
name
(str, optional) An artifact name. May be prefixed with entity/project. Valid names can be in the following forms: - name:version - name:alias - digest This will default to the basename of the path prepended with the current run id if not specified.
type
(str) The type of artifact to log, examples include dataset, model
aliases
(list, optional) Aliases to apply to this artifact, defaults to ["latest"]
distributed_id
(string, optional) Unique string that all distributed jobs share. If None, defaults to the run's group name.
Returns
Text
An Artifact object.

get_project_url

get_project_url() -> Optional[str]
Returns the url for the W&B project associated with the run, if there is one.
Offline runs will not have a project url.

get_sweep_url

get_sweep_url() -> Optional[str]
Returns the url for the sweep associated with the run, if there is one.

get_url

get_url() -> Optional[str]
Returns the url for the W&B run, if there is one.
Offline runs will not have a url.

join

join(
exit_code: int = None
) -> None
Deprecated alias for finish() - please use finish.
link_artifact(
artifact: Union[public.Artifact, Artifact],
target_path: str,
aliases: Optional[List[str]] = None
) -> None
Links the given artifact to a portfolio (a promoted collection of artifacts).
The linked artifact will be visible in the UI for the specified portfolio.
Arguments
Text
artifact
the (public or local) artifact which will be linked
target_path
str - takes the following forms: {portfolio}, {project}/{portfolio}, or {entity}/{project}/{portfolio}
aliases
List[str] - optional alias(es) that will only be applied on this linked artifact inside the portfolio. The alias "latest" will always be applied to the latest version of an artifact that is linked.
Returns
Text
None

log

log(
data: Dict[str, Any],
step: Optional[int] = None,
commit: Optional[bool] = None,
sync: Optional[bool] = None
) -> None
Logs a dictonary of data to the current run's history.
Use wandb.log to log data from runs, such as scalars, images, video, histograms, plots, and tables.
See our guides to logging for live examples, code snippets, best practices, and more.
The most basic usage is wandb.log({"train-loss": 0.5, "accuracy": 0.9}). This will save the loss and accuracy to the run's history and update the summary values for these metrics.
Visualize logged data in the workspace at wandb.ai, or locally on a self-hosted instance of the W&B app, or export data to visualize and explore locally, e.g. in Jupyter notebooks, with our API.
In the UI, summary values show up in the run table to compare single values across runs. Summary values can also be set directly with wandb.run.summary["key"] = value.
Logged values don't have to be scalars. Logging any wandb object is supported. For example wandb.log({"example": wandb.Image("myimage.jpg")}) will log an example image which will be displayed nicely in the W&B UI. See the reference documentation for all of the different supported types or check out our guides to logging for examples, from 3D molecular structures and segmentation masks to PR curves and histograms. wandb.Tables can be used to logged structured data. See our guide to logging tables for details.
Logging nested metrics is encouraged and is supported in the W&B UI. If you log with a nested dictionary like wandb.log({"train": {"acc": 0.9}, "val": {"acc": 0.8}}), the metrics will be organized into train and val sections in the W&B UI.
wandb keeps track of a global step, which by default increments with each call to wandb.log, so logging related metrics together is encouraged. If it's inconvenient to log related metrics together calling wandb.log({"train-loss": 0.5}, commit=False) and then wandb.log({"accuracy": 0.9}) is equivalent to calling wandb.log({"train-loss": 0.5, "accuracy": 0.9}).
wandb.log is not intended to be called more than a few times per second. If you want to log more frequently than that it's better to aggregate the data on the client side or you may get degraded performance.
Arguments
Text
data
(dict, optional) A dict of serializable python objects i.e str, ints, floats, Tensors, dicts, or any of the wandb.data_types.
commit
(boolean, optional) Save the metrics dict to the wandb server and increment the step. If false wandb.log just updates the current metrics dict with the data argument and metrics won't be saved until wandb.log is called with commit=True.
step
(integer, optional) The global step in processing. This persists any non-committed earlier steps but defaults to not committing the specified step.
sync
(boolean, True) This argument is deprecated and currently doesn't change the behaviour of wandb.log.

Examples:

For more and more detailed examples, see our guides to logging.

Basic usage

import wandb
wandb.init()
wandb.log({"accuracy": 0.9, "epoch": 5})

Incremental logging

import wandb
wandb.init()
wandb.log({"loss": 0.2}, commit=False)
# Somewhere else when I'm ready to report this step:
wandb.log({"accuracy": 0.8})

Histogram

import numpy as np
import wandb
# sample gradients at random from normal distribution
gradients = np.random.randn(100, 100)
wandb.init()
wandb.log({"gradients": wandb.Histogram(gradients)})

Image from numpy

import numpy as np
import wandb
wandb.init()
examples = []
for i in range(3):
pixels = np.random.randint(low=0, high=256, size=(100, 100, 3))
image = wandb.Image(pixels, caption=f"random field {i}")
examples.append(image)
wandb.log({"examples": examples})

Image from PIL

import numpy as np
from PIL import Image as PILImage
import wandb
wandb.init()
examples = []
for i in range(3):
pixels = np.random.randint(low=0, high=256, size=(100, 100, 3), dtype=np.uint8)
pil_image = PILImage.fromarray(pixels, mode="RGB")
image = wandb.Image(pil_image, caption=f"random field {i}")
examples.append(image)
wandb.log({"examples": examples})

Video from numpy

import numpy as np
import wandb
wandb.init()
# axes are (time, channel, height, width)
frames = np.random.randint(low=0, high=256, size=(10, 3, 100, 100), dtype=np.uint8)
wandb.log({"video": wandb.Video(frames, fps=4)})

Matplotlib Plot

from matplotlib import pyplot as plt
import numpy as np
import wandb
wandb.init()
fig, ax = plt.subplots()
x = np.linspace(0, 10)
y = x * x
ax.plot(x, y) # plot y = x^2
wandb.log({"chart": fig})

PR Curve

wandb.log({"pr": wandb.plots.precision_recall(y_test, y_probas, labels)})

3D Object

wandb.log(
{
"generated_samples": [
wandb.Object3D(open("sample.obj")),
wandb.Object3D(open("sample.gltf")),
wandb.Object3D(open("sample.glb")),
]
}
)
Raises
Text
wandb.Error
if called before wandb.init
ValueError
if invalid data is passed

log_artifact

log_artifact(
artifact_or_path: Union[wandb_artifacts.Artifact, str],
name: Optional[str] = None,
type: Optional[str] = None,
aliases: Optional[List[str]] = None
) -> wandb_artifacts.Artifact
Declare an artifact as an output of a run.
Arguments
Text
artifact_or_path
(str or Artifact) A path to the contents of this artifact, can be in the following forms: - /local/directory - /local/directory/file.txt - s3://bucket/path You can also pass an Artifact object created by calling wandb.Artifact.
name
(str, optional) An artifact name. May be prefixed with entity/project. Valid names can be in the following forms: - name:version - name:alias - digest This will default to the basename of the path prepended with the current run id if not specified.
type
(str) The type of artifact to log, examples include dataset, model
aliases
(list, optional) Aliases to apply to this artifact, defaults to ["latest"]
Returns
Text
An Artifact object.

log_code

log_code(
root: str = ".",
name: str = None,
include_fn: Callable[[str], bool] = _is_py_path,
exclude_fn: Callable[[str], bool] = filenames.exclude_wandb_fn
) -> Optional[Artifact]
Saves the current state of your code to a W&B Artifact.
By default, it walks the current directory and logs all files that end with .py.
Arguments
Text
root
The relative (to os.getcwd()) or absolute path to recursively find code from.
name
(str, optional) The name of our code artifact. By default, we'll name the artifact source-$PROJECT_ID-$ENTRYPOINT_RELPATH. There may be scenarios where you want many runs to share the same artifact. Specifying name allows you to achieve that.
include_fn
A callable that accepts a file path and returns True when it should be included and False otherwise. This defaults to: lambda path: path.endswith(".py")
exclude_fn
A callable that accepts a file path and returns True when it should be excluded and False otherwise. This defaults to: lambda path: False

Examples:

Basic usage
run.log_code()
Advanced usage
run.log_code(
"../", include_fn=lambda path: path.endswith(".py") or path.endswith(".ipynb")
)
Returns
Text
An Artifact object if code was logged

mark_preempting

mark_preempting() -> None
Marks this run as preempting.
Also tells the internal process to immediately report this to server.

plot_table

@staticmethod
plot_table(
vega_spec_name: str,
data_table: "wandb.Table",
fields: Dict[str, Any],
string_fields: Optional[Dict[str, Any]] = None
) -> CustomChart
Creates a custom plot on a table.
Arguments
Text
vega_spec_name
the name of the spec for the plot
table_key
the key used to log the data table
data_table
a wandb.Table object containing the data to be used on the visualization
fields
a dict mapping from table keys to fields that the custom visualization needs
string_fields
a dict that provides values for any string constants the custom visualization needs

project_name

project_name() -> str

restore

restore(
name: str,
run_path: Optional[str] = None,
replace: bool = (False),
root: Optional[str] = None
) -> Union[None, TextIO]
Downloads the specified file from cloud storage.
File is placed into the current directory or run directory. By default, will only download the file if it doesn't already exist.
Arguments
Text
name
the name of the file
run_path
optional path to a run to pull files from, i.e. username/project_name/run_id if wandb.init has not been called, this is required.
replace
whether to download the file even if it already exists locally
root
the directory to download the file to. Defaults to the current directory or the run directory if wandb.init was called.
Returns
Text
None if it can't find the file, otherwise a file object open for reading
Raises
Text
wandb.CommError
if we can't connect to the wandb backend
ValueError
if the file is not found or can't find run_path

save

save(
glob_str: Optional[str] = None,
base_path: Optional[str] = None,
policy: "PolicyName" = "live"
) -> Union[bool, List[str]]
Ensure all files matching glob_str are synced to wandb with the policy specified.
Arguments
Text
glob_str
(string) a relative or absolute path to a unix glob or regular path. If this isn't specified the method is a noop.
base_path
(string) the base path to run the glob relative to
policy
(string) on of live, now, or end - live: upload the file as it changes, overwriting the previous version - now: upload the file once now - end: only upload file when the run ends

to_html

to_html(
height: int = 420,
hidden: bool = (False)
) -> str
Generates HTML containing an iframe displaying the current run.

unwatch

unwatch(
models=None
) -> None

upsert_artifact

upsert_artifact(
artifact_or_path: Union[wandb_artifacts.Artifact, str],
name: Optional[str] = None,
type: Optional[str] = None,
aliases: Optional[List[str]] = None,
distributed_id: Optional[str] = None
) -> wandb_artifacts.Artifact
Declare (or append to) a non-finalized artifact as output of a run.
Note that you must call run.finish_artifact() to finalize the artifact. This is useful when distributed jobs need to all contribute to the same artifact.
Arguments
Text
artifact_or_path
(str or Artifact) A path to the contents of this artifact, can be in the following forms: - /local/directory - /local/directory/file.txt - s3://bucket/path You can also pass an Artifact object created by calling wandb.Artifact.
name
(str, optional) An artifact name. May be prefixed with entity/project. Valid names can be in the following forms: - name:version - name:alias - digest This will default to the basename of the path prepended with the current run id if not specified.
type
(str) The type of artifact to log, examples include dataset, model
aliases
(list, optional) Aliases to apply to this artifact, defaults to ["latest"]
distributed_id
(string, optional) Unique string that all distributed jobs share. If None, defaults to the run's group name.
Returns
Text
An Artifact object.

use_artifact

use_artifact(
artifact_or_name: Union[str, public.Artifact, Artifact],
type: Optional[str] = None,
aliases: Optional[List[str]] = None,
use_as: Optional[str] = None
) -> Union[public.Artifact, Artifact]
Declare an artifact as an input to a run.
Call download or file on the returned object to get the contents locally.
Arguments
Text
artifact_or_name
(str or Artifact) An artifact name. May be prefixed with entity/project/. Valid names can be in the following forms: - name:version - name:alias - digest You can also pass an Artifact object created by calling wandb.Artifact
type
(str, optional) The type of artifact to use.
aliases
(list, optional) Aliases to apply to this artifact
use_as
(string, optional) Optional string indicating what purpose the artifact was used with. Will be shown in UI.
Returns
Text
An Artifact object.

watch

watch(
models, criterion=None, log="gradients", log_freq=100, idx=None,
log_graph=(False)
) -> None

__enter__

__enter__() -> "Run"

__exit__

__exit__(
exc_type: Type[BaseException],
exc_val: BaseException,
exc_tb: TracebackType
) -> bool
Copy link
Outline
Methods
alert
define_metric
detach
display
finish
finish_artifact
get_project_url
get_sweep_url
get_url
join
link_artifact
log
Basic usage
Incremental logging
Histogram
Image from numpy
Image from PIL
Video from numpy
Matplotlib Plot
PR Curve
3D Object
log_artifact
log_code
mark_preempting
plot_table
project_name
restore
save
to_html
unwatch
upsert_artifact
use_artifact
watch
__enter__
__exit__