Observability for IBM watsonx with Opik

watsonx is a next generation enterprise studio for AI builders to train, validate, tune and deploy AI models.

Account Setup

Comet provides a hosted version of the Opik platform, simply create an account and grab your API Key.

You can also run the Opik platform locally, see the installation guide for more information.

Getting Started

Installation

To start tracking your watsonx LLM calls, you can use our LiteLLM integration. You’ll need to have both the opik and litellm packages installed. You can install them using pip:

$ pip install opik litellm

Configuring Opik

Configure the Opik Python SDK for your deployment type. See the Python SDK Configuration guide for detailed instructions on:

CLI configuration: opik configure
Code configuration: opik.configure()
Self-hosted vs Cloud vs Enterprise setup
Configuration files and environment variables

If you’re unable to use our LiteLLM integration with watsonx, please open an issue

Configuring watsonx

In order to configure watsonx, you will need to have:

The endpoint URL: Documentation for this parameter can be found here
Watsonx API Key: Documentation for this parameter can be found here
Watsonx Token: Documentation for this parameter can be found here
(Optional) Watsonx Project ID: Can be found in the Manage section of your project.

Once you have these, you can set them as environment variables:

1 import os
2 
3 os.environ["WATSONX_URL"] = ""  # (required) Base URL of your WatsonX instance
4 # (required) either one of the following:
5 os.environ["WATSONX_API_KEY"] = ""  # IBM cloud API key
6 os.environ["WATSONX_TOKEN"] = ""  # IAM auth token
7 # optional - can also be passed as params to completion() or embedding()
8 # os.environ["WATSONX_PROJECT_ID"] = "" # Project ID of your WatsonX instance
9 # os.environ["WATSONX_DEPLOYMENT_SPACE_ID"] = "" # ID of your deployment space to use deployed models

Logging LLM calls

In order to log the LLM calls to Opik, you will need to create the OpikLogger callback. Once the OpikLogger callback is created and added to LiteLLM, you can make calls to LiteLLM as you normally would:

1 from litellm.integrations.opik.opik import OpikLogger
2 import litellm
3 import os
4 
5 os.environ["OPIK_PROJECT_NAME"] = "watsonx-integration-demo"
6 
7 opik_logger = OpikLogger()
8 litellm.callbacks = [opik_logger]
9 
10 prompt = """
11 Write a short two sentence story about Opik.
12 """
13 
14 response = litellm.completion(
15     model="watsonx/ibm/granite-13b-chat-v2",
16     messages=[{"role": "user", "content": prompt}]
17 )
18 
19 print(response.choices[0].message.content)

Advanced Usage

Using with the `@track` decorator

If you have multiple steps in your LLM pipeline, you can use the @track decorator to log the traces for each step. If WatsonX is called within one of these steps, the LLM call will be associated with that corresponding step:

1 from opik import track
2 from opik.opik_context import get_current_span_data
3 import litellm
4 
5 @track
6 def generate_story(prompt):
7     response = litellm.completion(
8         model="watsonx/ibm/granite-13b-chat-v2",
9         messages=[{"role": "user", "content": prompt}],
10         metadata={
11             "opik": {
12                 "current_span_data": get_current_span_data(),
13             },
14         },
15     )
16     return response.choices[0].message.content
17 
18 @track
19 def generate_topic():
20     prompt = "Generate a topic for a story about Opik."
21     response = litellm.completion(
22         model="watsonx/ibm/granite-13b-chat-v2",
23         messages=[{"role": "user", "content": prompt}],
24         metadata={
25             "opik": {
26                 "current_span_data": get_current_span_data(),
27             },
28         },
29     )
30     return response.choices[0].message.content
31 
32 @track
33 def generate_opik_story():
34     topic = generate_topic()
35     story = generate_story(topic)
36     return story
37 
38 # Execute the multi-step pipeline
39 generate_opik_story()