Tutorial: App versioning ↗

Summary: Learn how to use Weave Model to track and version your application and its parameters

Original Documentation

Documentation Index#
Fetch the complete documentation index at: https://docs.wandb.ai/llms.txt Use this file to discover all available pages before exploring further.

Learn how to use Weave Model to track and version your application and its parameters

Tracking the inputs, outputs, metadata as well as data flowing through your app is critical to understanding the performance of your system. However versioning your app over time is also critical to understand how modifications to your code or application parameters change your outputs. Weave’s Model class is how these changes can be tracked in Weave.

In this tutorial you’ll learn:

How to use Weave Model to track and version your application and its parameters.
How to export, modify and re-use a Weave Model already logged.

Using `weave.Model`#

The weave.Model class is currently only supported in Python.

Using Weave Models means that parameters such as model vendor ids, prompts, temperature, and more are stored and versioned when they change.

To create a Model in Weave, you need the following:

a class that inherits from weave.Model
type definitions on all class fields
a typed invoke function with the @weave.op() decorator

When you change the class fields or the code that defines your model, these changes will be logged and the version will be updated. This ensures that you can compare the generations across different versions of your app.

In the example below, the model name, temperature and system prompt will be tracked and versioned:

    import json
    from openai import OpenAI

    import weave

    @weave.op()
    def extract_dinos(wmodel: weave.Model, sentence: str) -> dict:
        response = wmodel.client.chat.completions.create(
            model=wmodel.model_name,
            temperature=wmodel.temperature,
            messages=[
                {
                    "role": "system",
                    "content": wmodel.system_prompt
                },
                {
                    "role": "user",
                    "content": sentence
                }
                ],
                response_format={ "type": "json_object" }
            )
        return response.choices[0].message.content

    # Sub-class with a weave.Model
    class ExtractDinos(weave.Model):
        client: OpenAI = None
        model_name: str
        temperature: float
        system_prompt: str

        # Ensure your function is called `invoke` or `predict`
        @weave.op()
        def invoke(self, sentence: str) -> dict:
            dino_data  = extract_dinos(self, sentence)
            return json.loads(dino_data)
    ```
  <span class="tab-end"></span>

  <span class="tab-start" data-tab-title="TypeScript"></span>
```plaintext
    This feature is not available in TypeScript yet.  Stay tuned!
    ```
  <span class="tab-end"></span>
<span class="tab-group-end"></span>

Now you can instantiate and call the model with `invoke`:

<span class="tab-group-start"></span>
  <span class="tab-start" data-tab-title="Python"></span>
```python
    weave.init('jurassic-park')
    client = OpenAI()

    system_prompt = """Extract any dinosaur `name`, their `common_name`, \
    names and whether its `diet` is a herbivore or carnivore, in JSON format."""

    dinos = ExtractDinos(
        client=client,
        model_name='gpt-4o',
        temperature=0.4,
        system_prompt=system_prompt
    )

    sentence = """I watched as a Tyrannosaurus rex (T. rex) chased after a Triceratops (Trike), \
    both carnivore and herbivore locked in an ancient dance. Meanwhile, a gentle giant \
    Brachiosaurus (Brachi) calmly munched on treetops, blissfully unaware of the chaos below."""

    result = dinos.invoke(sentence)
    print(result)
    ```
  <span class="tab-end"></span>

  <span class="tab-start" data-tab-title="TypeScript"></span>
```plaintext
    This feature is not available in TypeScript yet.  Stay tuned!
    ```
  <span class="tab-end"></span>
<span class="tab-group-end"></span>

Now after calling `.invoke` you can see the trace in Weave **now tracks the model parameters as well as the code** for the model functions that have been decorated with `weave.op()`. You can see the model is also versioned, "v21" in this case, and if you click on the model **you can see all of the calls** that have used that version of the model

<img src="https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_invoke3.png?fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=313be15ded432fa94e016c9cf3353f50" alt="Re-using a weave model" data-og-width="1664" width="1664" data-og-height="1292" height="1292" data-path="images/tutorial-model_invoke3.png" data-optimize="true" data-opv="3" srcset="https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_invoke3.png?w=280&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=ba3e9dbf3b0c2abbdf1fdbd1e4837c6b 280w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_invoke3.png?w=560&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=2710f8084a66042e5ea82c6a808b509b 560w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_invoke3.png?w=840&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=8b345b5add0697cfce685eed9a41135b 840w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_invoke3.png?w=1100&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=0ad244c148788c7a0ce9f9b8bc94f865 1100w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_invoke3.png?w=1650&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=a232b065cd2d10aa48d93078646bc6df 1650w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_invoke3.png?w=2500&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=ba0e1d95b25273c398461918e4417cd0 2500w" />

**A note on using `weave.Model`:**

* You can use `predict` instead of `invoke` for the name of the function in your Weave `Model` if you prefer.
* If you want other class methods to be tracked by weave they need to be wrapped in `weave.op()`
* Parameters starting with an underscore are ignored by weave and won't be logged

## Exporting and re-using a logged `weave.Model`

Because Weave stores and versions Models that have been invoked, it is possible to export and re-use these models.

**Get the Model ref**
In the Weave UI you can get the Model ref for a particular version

**Using the Model**
Once you have the URI of the Model object, you can export and re-use it. Note that the exported model is already initialised and ready to use:

<span class="tab-group-start"></span>
  <span class="tab-start" data-tab-title="Python"></span>
```python
    # the exported weave model is already initialised and ready to be called
    new_dinos = weave.ref("weave://morgan/jurassic-park/object/ExtractDinos:ey4udBU2MU23heQFJenkVxLBX4bmDsFk7vsGcOWPjY4").get()

    # set the client to the openai client again
    new_dinos.client = client

    new_sentence = """I also saw an Ankylosaurus grazing on giant ferns"""
    new_result = new_dinos.invoke(new_sentence)
    print(new_result)
    ```
  <span class="tab-end"></span>

  <span class="tab-start" data-tab-title="TypeScript"></span>
```plaintext
    This feature is not available in TypeScript yet.  Stay tuned!
    ```
  <span class="tab-end"></span>
<span class="tab-group-end"></span>

Here you can now see the name Model version (v21) was used with the new input:

<img src="https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_re-use.png?fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=0a70382652bdfd4d06a715cd13553dff" alt="Re-using a weave model" data-og-width="1260" width="1260" data-og-height="1120" height="1120" data-path="images/tutorial-model_re-use.png" data-optimize="true" data-opv="3" srcset="https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_re-use.png?w=280&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=ecfc9a6d08187fe51b83b5e6aee1658a 280w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_re-use.png?w=560&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=9ae60db7006e7ecfd72256f761299481 560w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_re-use.png?w=840&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=77394d3120b0b1622670d5af0adf72af 840w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_re-use.png?w=1100&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=c6aa898956bf76bef25dddebc3459ac1 1100w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_re-use.png?w=1650&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=463b1c17729578a7203e32ef1b00475f 1650w, https://mintcdn.com/wb-21fd5541/aRvhhwVWqlxBzke5/images/tutorial-model_re-use.png?w=2500&fit=max&auto=format&n=aRvhhwVWqlxBzke5&q=85&s=84cc6009c1f926a2923d65a82e37b950 2500w" />

## What's next?

* Follow the [Build an Evaluation pipeline tutorial](/weave/tutorial-eval) to start iteratively improving your applications.

Link last verified June 7, 2026. View original ↗

Source: Weights & Biases Docs

Link last verified: 2026-03-04

Tutorial: App versioning ↗

Original Documentation

Documentation Index#

Using weave.Model#

Using `weave.Model`#