Skip to main content

Artifact

Flexible and lightweight building block for dataset and model versioning.

Artifact(
name: str,
type: str,
description: Optional[str] = None,
metadata: Optional[Dict[str, Any]] = None,
incremental: bool = (False),
use_as: Optional[str] = None
) -> None

Constructs an empty artifact whose contents can be populated using its add family of functions. Once the artifact has all the desired files, you can call wandb.log_artifact() to log it.

Arguments
nameA human-readable name for this artifact, which is how you can identify this artifact in the UI or reference it in use_artifact calls. Names can contain letters, numbers, underscores, hyphens, and dots. The name must be unique across a project.
typeThe type of the artifact, which is used to organize and differentiate artifacts. Common types include dataset or model, but you can use any string containing letters, numbers, underscores, hyphens, and dots.
descriptionFree text that offers a description of the artifact. The description is markdown rendered in the UI, so this is a good place to place tables, links, etc.
metadataStructured data associated with the artifact, for example class distribution of a dataset. This will eventually be queryable and plottable in the UI. There is a hard limit of 100 total keys.
Returns
An Artifact object.

Examples:

Basic usage:

wandb.init()

artifact = wandb.Artifact("mnist", type="dataset")
artifact.add_dir("mnist/")
wandb.log_artifact(artifact)
Attributes
aliasesThe aliases associated with this artifact. The list is mutable and calling save() will persist all alias changes.
collectionThe collection this artifact was retrieved from. If this artifact was retrieved from a portfolio / linked collection, that collection will be returned rather than the source sequence.
commit_hashThe hash returned when this artifact was committed.
created_atThe time at which the artifact was created.
descriptionThe artifact description. Free text that offers a user-set description of the artifact.
digestThe logical digest of the artifact. The digest is the checksum of the artifact's contents. If an artifact has the same digest as the current latest version, then log_artifact is a no-op.
entityThe name of the entity of the secondary (portfolio) artifact collection.
file_countThe number of files (including references).
idThe artifact's ID.
manifestThe artifact's manifest. The manifest lists all of its contents, and can't be changed once the artifact has been logged.
metadataUser-defined artifact metadata. Structured data associated with the artifact.
nameThe artifact name and version in its secondary (portfolio) collection. A string with the format {collection}:{alias}. Before the artifact is saved, contains only the name since the version is not yet known.
projectThe name of the project of the secondary (portfolio) artifact collection.
qualified_nameThe entity/project/name of the secondary (portfolio) collection.
sizeThe total size of the artifact in bytes. Includes any references tracked by this artifact.
source_collectionThe artifact's primary (sequence) collection.
source_entityThe name of the entity of the primary (sequence) artifact collection.
source_nameThe artifact name and version in its primary (sequence) collection. A string with the format {collection}:{alias}. Before the artifact is saved, contains only the name since the version is not yet known.
source_projectThe name of the project of the primary (sequence) artifact collection.
source_qualified_nameThe entity/project/name of the primary (sequence) collection.
source_versionThe artifact's version in its primary (sequence) collection. A string with the format "v{number}".
stateThe status of the artifact. One of: "PENDING", "COMMITTED", or "DELETED".
ttlTime To Live (TTL). The artifact will be deleted shortly after TTL since its creation. None means the artifact will never expire. If TTL is not set on an artifact, it will inherit the default for its collection.
typeThe artifact's type.
updated_atThe time at which the artifact was last updated.
versionThe artifact's version in its secondary (portfolio) collection.

Methods

add

View source

add(
obj: data_types.WBValue,
name: StrPath
) -> ArtifactManifestEntry

Add wandb.WBValue obj to the artifact.

Arguments
objThe object to add. Currently support one of Bokeh, JoinedTable, PartitionedTable, Table, Classes, ImageMask, BoundingBoxes2D, Audio, Image, Video, Html, Object3D
nameThe path within the artifact to add the object.
Returns
The added manifest entry
Raises
ArtifactFinalizedErrorif the artifact has already been finalized

Examples:

Basic usage:

artifact = wandb.Artifact("my_table", type="dataset")
table = wandb.Table(
columns=["a", "b", "c"], data=[(i, i * 2, 2**i) for i in range(10)]
)
artifact.add(table, "my_table")

wandb.log_artifact(artifact)

Retrieve an object:

artifact = wandb.use_artifact("my_table:latest")
table = artifact.get("my_table")

add_dir

View source

add_dir(
local_path: str,
name: Optional[str] = None
) -> None

Add a local directory to the artifact.

Arguments
local_pathThe path to the directory being added.
nameThe path within the artifact to use for the directory being added. Defaults to the root of the artifact.
Raises
ArtifactFinalizedErrorif the artifact has already been finalized

Examples:

Add a directory without an explicit name:

# All files in `my_dir/` are added at the root of the artifact.
artifact.add_dir("my_dir/")

Add a directory and name it explicitly:

# All files in `my_dir/` are added under `destination/`.
artifact.add_dir("my_dir/", name="destination")

add_file

View source

add_file(
local_path: str,
name: Optional[str] = None,
is_tmp: Optional[bool] = (False)
) -> ArtifactManifestEntry

Add a local file to the artifact.

Arguments
local_pathThe path to the file being added.
nameThe path within the artifact to use for the file being added. Defaults to the basename of the file.
is_tmpIf true, then the file is renamed deterministically to avoid collisions.
Returns
The added manifest entry
Raises
ArtifactFinalizedErrorif the artifact has already been finalized

Examples:

Add a file without an explicit name:

# Add as `file.txt'
artifact.add_file("path/to/file.txt")

Add a file with an explicit name:

# Add as 'new/path/file.txt'
artifact.add_file("path/to/file.txt", name="new/path/file.txt")

add_reference

View source

add_reference(
uri: Union[ArtifactManifestEntry, str],
name: Optional[StrPath] = None,
checksum: bool = (True),
max_objects: Optional[int] = None
) -> Sequence[ArtifactManifestEntry]

Add a reference denoted by a URI to the artifact.

Unlike adding files or directories, references are NOT uploaded to W&B. However, artifact methods such as download() can be used regardless of whether the artifact contains references or uploaded files.

By default, W&B offers special handling for the following schemes:

  • http(s): The size and digest of the file will be inferred by the Content-Length and the ETag response headers returned by the server.
  • s3: The checksum and size will be pulled from the object metadata. If bucket versioning is enabled, then the version ID is also tracked.
  • gs: The checksum and size will be pulled from the object metadata. If bucket versioning is enabled, then the version ID is also tracked.
  • https, domain matching *.blob.core.windows.net (Azure): The checksum and size will be pulled from the blob metadata. If storage account versioning is enabled, then the version ID is also tracked.
  • file: The checksum and size will be pulled from the file system. This scheme is useful if you have an NFS share or other externally mounted volume containing files you wish to track but not necessarily upload.

For any other scheme, the digest is just a hash of the URI and the size is left blank.

Arguments
uriThe URI path of the reference to add. Can be an object returned from Artifact.get_path to store a reference to another artifact's entry.
nameThe path within the artifact to place the contents of this reference
checksumWhether or not to checksum the resource(s) located at the reference URI. Checksumming is strongly recommended as it enables automatic integrity validation, however it can be disabled to speed up artifact creation. (default: True)
max_objectsThe maximum number of objects to consider when adding a reference that points to directory or bucket store prefix. For S3 and GCS, this limit is 10,000 by default but is uncapped for other URI schemes. (default: None)
Returns
The added manifest entries.
Raises
ArtifactFinalizedErrorif the artifact has already been finalized.

Examples:

Add an HTTP link:

# Adds `file.txt` to the root of the artifact as a reference.
artifact.add_reference("http://myserver.com/file.txt")

Add an S3 prefix without an explicit name:

# All objects under `prefix/` will be added at the root of the artifact.
artifact.add_reference("s3://mybucket/prefix")

Add a GCS prefix with an explicit name:

# All objects under `prefix/` will be added under `path/` at the artifact
# root.
artifact.add_reference("gs://mybucket/prefix", name="path")

checkout

View source

checkout(
root: Optional[str] = None
) -> str

Replace the specified root directory with the contents of the artifact.

WARNING: This will DELETE all files in root that are not included in the artifact.

Arguments
rootThe directory to replace with this artifact's files.
Returns
The path to the checked out contents.
Raises
ArtifactNotLoggedErrorif the artifact has not been logged

delete

View source

delete(
delete_aliases: bool = (False)
) -> None

Delete an artifact and its files.

Arguments
delete_aliasesIf true, deletes all aliases associated with the artifact. Otherwise, this raises an exception if the artifact has existing aliases.
Raises
ArtifactNotLoggedErrorif the artifact has not been logged

Examples:

Delete all the "model" artifacts a run has logged:

runs = api.runs(path="my_entity/my_project")
for run in runs:
for artifact in run.logged_artifacts():
if artifact.type == "model":
artifact.delete(delete_aliases=True)

download

View source

download(
root: Optional[str] = None,
recursive: bool = (False),
allow_missing_references: bool = (False)
) -> FilePathStr

Download the contents of the artifact to the specified root directory.

NOTE: Any existing files at root are left untouched. Explicitly delete root before calling download if you want the contents of root to exactly match the artifact.

Arguments
rootThe directory in which to download this artifact's files.
recursiveIf true, then all dependent artifacts are eagerly downloaded. Otherwise, the dependent artifacts are downloaded as needed.
Returns
The path to the downloaded contents.
Raises
ArtifactNotLoggedErrorif the artifact has not been logged

file

View source

file(
root: Optional[str] = None
) -> StrPath

Download a single file artifact to dir specified by the root.

Arguments
rootThe root directory in which to place the file. Defaults to './artifacts/self.name/'.
Returns
The full path of the downloaded file.
Raises
ArtifactNotLoggedErrorif the artifact has not been logged
ValueErrorif the artifact contains more than one file

files

View source

files(
names: Optional[List[str]] = None,
per_page: int = 50
) -> ArtifactFiles

Iterate over all files stored in this artifact.

Arguments
namesThe filename paths relative to the root of the artifact you wish to list.
per_pageThe number of files to return per request
Returns
An iterator containing File objects
Raises
ArtifactNotLoggedErrorif the artifact has not been logged

finalize

View source

finalize() -> None

Mark this artifact as final, disallowing further modifications.

This happens automatically when calling log_artifact.

get

View source

get(
name: str
) -> Optional[data_types.WBValue]

Get the WBValue object located at the artifact relative name.

Arguments
nameThe artifact relative name to get
Raises
ArtifactNotLoggedErrorif the artifact isn't logged or the run is offline

Examples:

Basic usage:

# Run logging the artifact
with wandb.init() as r:
artifact = wandb.Artifact("my_dataset", type="dataset")
table = wandb.Table(
columns=["a", "b", "c"], data=[(i, i * 2, 2**i) for i in range(10)]
)
artifact.add(table, "my_table")
wandb.log_artifact(artifact)

# Run using the artifact
with wandb.init() as r:
artifact = r.use_artifact("my_dataset:latest")
table = artifact.get("my_table")

get_added_local_path_name

View source

get_added_local_path_name(
local_path: str
) -> Optional[str]

Get the artifact relative name of a file added by a local filesystem path.

Arguments
local_pathThe local path to resolve into an artifact relative name.
Returns
The artifact relative name.

Examples:

Basic usage:

artifact = wandb.Artifact("my_dataset", type="dataset")
artifact.add_file("path/to/file.txt", name="artifact/path/file.txt")

# Returns `artifact/path/file.txt`:
name = artifact.get_added_local_path_name("path/to/file.txt")

get_path

View source

get_path(
name: StrPath
) -> ArtifactManifestEntry

Get the entry with the given name.

Arguments
nameThe artifact relative name to get
Raises
ArtifactNotLoggedErrorif the artifact isn't logged or the run is offline
KeyErrorif the artifact doesn't contain an entry with the given name

Examples:

Basic usage:

# Run logging the artifact
with wandb.init() as r:
artifact = wandb.Artifact("my_dataset", type="dataset")
artifact.add_file("path/to/file.txt")
wandb.log_artifact(artifact)

# Run using the artifact
with wandb.init() as r:
artifact = r.use_artifact("my_dataset:latest")
path = artifact.get_path("file.txt")

# Can now download 'file.txt' directly:
path.download()

is_draft

View source

is_draft() -> bool

Whether the artifact is a draft, i.e. it hasn't been saved yet.

json_encode

View source

json_encode() -> Dict[str, Any]

View source

link(
target_path: str,
aliases: Optional[List[str]] = None
) -> None

Link this artifact to a portfolio (a promoted collection of artifacts).

Arguments
target_pathThe path to the portfolio. It must take the form {portfolio}, {project}/{portfolio} or {entity}/{project}/{portfolio}.
aliasesA list of strings which uniquely identifies the artifact inside the specified portfolio.
Raises
ArtifactNotLoggedErrorif the artifact has not been logged

logged_by

View source

logged_by() -> Optional[Run]

Get the run that first logged this artifact.

Raises
ArtifactNotLoggedErrorif the artifact has not been logged

new_draft

View source

new_draft() -> "Artifact"

Create a new draft artifact with the same content as this committed artifact.

The artifact returned can be extended or modified and logged as a new version.

Raises
ArtifactNotLoggedErrorif the artifact has not been logged

new_file

View source

@contextlib.contextmanager
new_file(
name: str,
mode: str = "w",
encoding: Optional[str] = None
) -> Generator[IO, None, None]

Open a new temporary file that will be automatically added to the artifact.

Arguments
nameThe name of the new file being added to the artifact.
modeThe mode in which to open the new file.
encodingThe encoding in which to open the new file.
Returns
A new file object that can be written to. Upon closing, the file will be automatically added to the artifact.
Raises
ArtifactFinalizedErrorif the artifact has already been finalized.

Examples:

artifact = wandb.Artifact("my_data", type="dataset")
with artifact.new_file("hello.txt") as f:
f.write("hello!")
wandb.log_artifact(artifact)

remove

View source

remove(
item: Union[StrPath, 'ArtifactManifestEntry']
) -> None

Remove an item from the artifact.

Arguments
itemthe item to remove. Can be a specific manifest entry or the name of an artifact-relative path. If the item matches a directory all items in that directory will be removed.
Raises
ArtifactFinalizedErrorif the artifact has already been finalized.
FileNotFoundErrorif the item isn't found in the artifact.

save

View source

save(
project: Optional[str] = None,
settings: Optional['wandb.wandb_sdk.wandb_settings.Settings'] = None
) -> None

Persist any changes made to the artifact.

If currently in a run, that run will log this artifact. If not currently in a run, a run of type "auto" will be created to track this artifact.

Arguments
projectA project to use for the artifact in the case that a run is not already in context
settingsA settings object to use when initializing an automatic run. Most commonly used in testing harness.

used_by

View source

used_by() -> List[Run]

Get a list of the runs that have used this artifact.

Raises
ArtifactNotLoggedErrorif the artifact has not been logged

verify

View source

verify(
root: Optional[str] = None
) -> None

Verify that the actual contents of an artifact match the manifest.

All files in the directory are checksummed and the checksums are then cross-referenced against the artifact's manifest.

NOTE: References are not verified.

Arguments
rootThe directory to verify. If None artifact will be downloaded to './artifacts/self.name/'
Raises
ArtifactNotLoggedErrorif the artifact has not been logged
ValueErrorIf the verification fails.

wait

View source

wait(
timeout: Optional[int] = None
) -> "Artifact"

Wait for this artifact to finish logging, if needed.

Arguments
timeoutWait up to this long.

__getitem__

View source

__getitem__(
name: str
) -> Optional[data_types.WBValue]

Get the WBValue object located at the artifact relative name.

Arguments
nameThe artifact relative name to get
Raises
ArtifactNotLoggedErrorif the artifact isn't logged or the run is offline

Examples:

Basic usage:

artifact = wandb.Artifact("my_table", type="dataset")
table = wandb.Table(
columns=["a", "b", "c"], data=[(i, i * 2, 2**i) for i in range(10)]
)
artifact["my_table"] = table

wandb.log_artifact(artifact)

Retrieving an object:

artifact = wandb.use_artifact("my_table:latest")
table = artifact["my_table"]

__setitem__

View source

__setitem__(
name: str,
item: data_types.WBValue
) -> ArtifactManifestEntry

Add item to the artifact at path name.

Arguments
nameThe path within the artifact to add the object.
itemThe object to add.
Returns
The added manifest entry
Raises
ArtifactFinalizedErrorif the artifact has already been finalized.

Examples:

Basic usage:

artifact = wandb.Artifact("my_table", type="dataset")
table = wandb.Table(
columns=["a", "b", "c"], data=[(i, i * 2, 2**i) for i in range(10)]
)
artifact["my_table"] = table

wandb.log_artifact(artifact)

Retrieving an object:

artifact = wandb.use_artifact("my_table:latest")
table = artifact["my_table"]
Was this page helpful?👍👎