Implementation:Wandb Weave Weave Publish

Knowledge Sources	Wandb Weave Weave Docs
Domains	Versioning, Object_Storage
Last Updated	2026-02-14 00:00 GMT

Overview

Concrete tool for persisting and versioning Python objects and retrieving them by reference provided by the Wandb Weave library.

Description

weave.publish() saves any Python object (Model, Dataset, Prompt, Op, etc.) to the Weave backend with content-addressable versioning. If the object's content hash differs from the latest version, a new version is created. weave.ref() creates a reference to an existing object without retrieving it, and calling .get() on the reference materializes the object.

Usage

Use weave.publish() to save artifacts that should be versioned and retrievable. Use weave.ref() to create lightweight references for passing to other Weave APIs or for later retrieval.

Code Reference

Source Location

Repository: wandb/weave
File: weave/trace/api.py
Lines: L118-178 (publish), L181-207 (ref)

Signature

def publish(obj: Any, name: str | None = None) -> ObjectRef:
    """Save and version a Python object.

    Args:
        obj: The object to save and version.
        name: The name to save the object under.

    Returns:
        A Weave Ref to the saved object.
    """

def ref(location: str) -> ObjectRef:
    """Creates a Ref to an existing Weave object.

    Args:
        location: A Weave Ref URI, or "name:version" or "name".

    Returns:
        A Weave Ref to the object.
    """

Import

import weave

I/O Contract

Inputs (publish)

Name	Type	Required	Description
obj	Any	Yes	Python object to save (Model, Dataset, Prompt, etc.)
name	None	No	Name to save under (defaults to obj.name or class name)

Outputs (publish)

Name	Type	Description
return	ObjectRef	Immutable reference with entity, project, name, digest

Inputs (ref)

Name	Type	Required	Description
location	str	Yes	URI or short form ("name" or "name:version")

Outputs (ref)

Name	Type	Description
return	ObjectRef	Reference to the object (call .get() to materialize)

Usage Examples

Publish and Retrieve

import weave

weave.init("my-team/my-project")

# Publish a dataset
dataset = weave.Dataset(
    name="my_dataset",
    rows=[{"input": "hello", "expected": "world"}],
)
ref = weave.publish(dataset)
print(ref)  # weave:///my-team/my-project/object/my_dataset:<digest>

# Retrieve the dataset
retrieved = weave.ref("my_dataset").get()
print(retrieved.rows[0])

Version-Specific Retrieval

import weave

# Get a specific version by digest
specific_ref = weave.ref("my_dataset:abc123")
dataset_v1 = specific_ref.get()

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment