此页面为英文。我们正在努力将其翻译成您的语言。
IBM WatsonX Integration
Lunary has partnered with IBM to provide a seamless integration for monitoring WatsonX calls in your Python app.
Our Python SDK includes automatic integration with IBM WatsonX's foundation models using Lunary.
1
Setup the SDK
First, ensure you have installed the IBM WatsonX SDK and Lunary. Set your environment variables for IBM authentication.
pip install ibm-watsonx-ai lunary
Configure your environment variables:
IBM_API_KEY: your IBM API keyIBM_PROJECT_ID: your IBM project id
2
Monitor IBM WatsonX calls
Wrap your WatsonX model instance with Lunary's monitor method to automatically track your calls.
import os
from ibm_watsonx_ai import Credentials
from ibm_watsonx_ai.foundation_models import ModelInference
import lunary
model = ModelInference(
model_id="meta-llama/llama-3-1-8b-instruct",
credentials=Credentials(
api_key=os.environ.get("IBM_API_KEY"),
url="https://us-south.ml.cloud.ibm.com"
),
project_id=os.environ.get("IBM_PROJECT_ID")
)
lunary.monitor(model)
messages = [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Who won the world series in 2020?"}
]
response = model.chat(messages=messages)
3
Tag requests and identify users
Optionally, pass extra parameters to track details such as tags and user information by including additional arguments to the chat call.
response = model.chat(messages=messages, tags=["baseball"], user_id="1234", user_props={"name": "Alice"})
