​
Cerebras Integration

This guide demonstrates how to integrate Langtrace with Cerebras to gain deep observability into your AI workloads. Cerebras, a leader in AI hardware and cloud solutions, enables unprecedented performance for large language models. By combining Cerebras with Langtrace, you can monitor, debug, and optimize your AI pipelines effectively.

​
Prerequisites

Before you begin, make sure you have:
  1. A Cerebras API key
  2. Python 3.7 or later installed
  3. The Langtrace SDK installed and configured

​
Installation

Install the required packages: 
pip install langtrace-python-sdk cerebras-cloud-sdk python-dotenv

​
Configuration

Set up your environment variables: 
export CEREBRAS_API_KEY='your-cerebras-api-key'
export LANGTRACE_API_KEY='your-langtrace-api-key'
Or use a .env file: 
CEREBRAS_API_KEY=your-cerebras-api-key
LANGTRACE_API_KEY=your-langtrace-api-key

​
Usage

​
Direct Cerebras SDK Integration

Here’s how to use Langtrace with the Cerebras SDK: 
from langtrace_python_sdk import langtrace
from cerebras.cloud.sdk import Cerebras
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

# Initialize Langtrace
langtrace.init()

# Initialize Cerebras client
client = Cerebras()

def completion_example(stream=False):
    completion = client.chat.completions.create(
        messages=[{
            "role": "user",
            "content": "Why is fast inference important?",
        }],
        model="llama3.1-8b",
        stream=stream,
    )

    if stream:
        for chunk in completion:
            print(chunk)
    else:
        return completion

# Execute the completion
completion_example()

​
OpenAI-Compatible Endpoint

Cerebras also provides an OpenAI-compatible endpoint: 
from langtrace_python_sdk import langtrace
from dotenv import load_dotenv
from openai import OpenAI
import os

# Load environment variables
load_dotenv()

# Initialize Langtrace
langtrace.init()

# Initialize OpenAI client with Cerebras endpoint
openai_client = OpenAI(
    base_url="https://api.cerebras.ai/v1",
    api_key=os.getenv("CEREBRAS_API_KEY"),
)

def openai_cerebras_example(stream=False):
    completion = openai_client.chat.completions.create(
        messages=[{
            "role": "user",
            "content": "Why is fast inference important?",
        }],
        model="llama3.1-8b",
        stream=stream,
    )

    if stream:
        for chunk in completion:
            print(chunk)
    else:
        return completion

# Execute the completion
openai_cerebras_example()

​
Observability Features

Langtrace provides comprehensive observability for your Cerebras workloads:
  1. Performance Monitoring
    • Track latency and throughput metrics
    • Monitor resource utilization
    • Identify bottlenecks in your pipeline
  2. Error Tracking
    • Trace errors in distributed workloads
    • Debug complex pipelines
    • Monitor both synchronous and asynchronous operations
  3. Resource Optimization
    • Analyze hardware resource usage
    • Optimize model execution
    • Fine-tune performance parameters
  4. User Experience Metrics
    • Track completion times
    • Monitor streaming response performance
    • Measure end-to-end request latency

​
Best Practices

  1. Initialize Early Always initialize Langtrace before creating the Cerebras client to ensure all operations are traced.
  2. Use Environment Variables Store sensitive credentials in environment variables or .env files.
  3. Enable Streaming For long-running operations, use streaming to monitor progress in real-time.
  4. Monitor Resource Usage Regularly check resource utilization metrics to optimize performance.

​
Troubleshooting

Common issues and solutions:
  1. Authentication Errors
    • Verify your API keys are correctly set
    • Ensure environment variables are loaded
  2. Performance Issues
    • Check Langtrace dashboards for bottlenecks
    • Monitor resource utilization metrics
    • Verify model configuration settings

​
Additional Resources