Learn how to use Datadog’s integration with the Ragas framework to evaluate your LLM application. For more information about the Ragas integration, including a detailed setup guide, see Ragas Evaluations.

1. Install required dependencies:

   pip install ragas==0.1.21 openai ddtrace>=3.0.0
   

2. Create a file named quickstart.py with the following code:

   import os
   from ddtrace.llmobs import LLMObs
   from ddtrace.llmobs.utils import Prompt
   from openai import OpenAI
   
   LLMObs.enable(
       ml_app="test-rag-app",
       agentless_enabled=True,
   )
   
   oai_client = OpenAI(api_key=os.environ.get("OPENAI_API_KEY"))
   
   rag_context = "The First AFL–NFL World Championship Game was an American football game played on January 15, 1967, at the Los Angeles Memorial Coliseum in Los Angeles"
   
   with LLMObs.annotation_context(
       prompt=Prompt(variables={"context": rag_context}),
   ):
       completion = oai_client.chat.completions.create(
           model="gpt-3.5-turbo",
           messages=[
               {"role": "system", "content": "Answer the user's question given the following context information {}".format(rag_context)},
               {"role": "user", "content": "When was the first superbowl?"},
           ],
       )
   

3. Run the script with the Ragas Faithfulness evaluation enabled:

   DD_LLMOBS_EVALUATORS=ragas_faithfulness DD_ENV=dev DD_API_KEY=<YOUR-DD-API-KEY> DD_SITE=datadoghq.com python quickstart.py
   

4. View your results in Datadog at https://<YOUR-DATADOG-SITE-URL>/llm/traces?query=%40ml_app%3Atest-rag-app.

Further Reading

Documentation, liens et articles supplémentaires utiles:

Ragas EvaluationsDOCUMENTATION