Updated Cookbook: Example for Fetching Scores from Langfuse

[Sohammhatre10]

Description

This update provides an example of using the fetch_scores() function from Langfuse to retrieve evaluation metrics. The example integrates UpTrain and Ragas for model evaluation and demonstrates how to log and fetch scores within Langfuse as mentioned in langfuse/langfuse#3505

Key Features

1. Evaluation with UpTrain and Ragas:

   • Provides examples for evaluating context relevance, factual accuracy, response completeness, context precision, faithfulness, and answer relevancy..

2. Fetching Scores:

   • Shows how to retrieve and filter scores using fetch_scores_from_langfuse.

3. Correlation Matrix Visualization:

   • Adds a section that calculates and visualizes the correlation between UpTrain and Ragas evaluation scores using a heatmap.

---

Important

Adds an example for using Langfuse to fetch scores, evaluate models with UpTrain and Ragas, and visualize results using a correlation matrix.

• Behavior:
  • Adds example for using fetch_scores() from Langfuse to retrieve evaluation metrics.
  • Demonstrates integration with UpTrain and Ragas for model evaluation.
  • Shows how to log and fetch scores within Langfuse.
• Visualization:
  • Includes a section for calculating and visualizing correlation between evaluation scores using a heatmap.
• Misc:
  • Minor whitespace changes in dspy.md, instructor.md, example-javascript.md, example-python-langgraph.md, example-python-instrumentation-module.md, example-python.md, example-vercel-ai.md, example_external_evaluation_pipelines.md, integration_dspy.md, integration_instructor.md, integration_langgraph.md, integration_llama-index_instrumentation.md, integration_llama_index_posthog_mistral.md, integration_mirascope.md, integration_mistral_sdk.md, integration_ollama.md, integration_openai_structured_output.md, example-langchain.md, js_integration_langchain.md, js_tracing_example_vercel_ai_sdk.md, prompt_management_langchain.md.

^{This description was created by for 02ebc24. It will automatically update as commits are pushed.}

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[vercel]

@Sohammhatre10 is attempting to deploy a commit to the langfuse Team on Vercel.

A member of the Team first needs to authorize it.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
0 out of 2 committers have signed the CLA.

❌ Your Name
❌ Sohammhatre10

---

Your Name seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[ellipsis-dev]

👍 Looks good to me! Reviewed everything up to 02ebc24 in 42 seconds

More details

• Looked at 1680 lines of code in 26 files
• Skipped 3 files when reviewing.
• Skipped posting 1 drafted comments based on config settings.

1. pages/docs/integrations/dspy.md:242

• Draft comment:
  Remove trailing whitespace for cleaner code. This issue is present in multiple files, such as example-javascript.md, example-python-langgraph.md, example-python-instrumentation-module.md, example-python.md, example-vercel-ai.md, external-evaluation-pipelines.md, integration_dspy.md, integration_instructor.md, integration_langgraph.md, integration_llama-index_instrumentation.md, integration_llama_index_posthog_mistral.md, integration_mirascope.md, integration_mistral_sdk.md, integration_ollama.md, integration_openai_structured_output.md, js_integration_langchain.md, js_tracing_example_vercel_ai_sdk.md, prompt_management_langchain.md.
• Reason this comment was not posted:
  Confidence changes required: 50%
  The PR introduces a new example for fetching scores from Langfuse, but there are several instances of trailing whitespace in the markdown files. These should be removed for cleaner code.

Workflow ID: wflow_ON6OiLFA8uvvhNsK

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.

[greptile-apps]

Disclaimer: Experimental PR review

PR Summary

This pull request adds a comprehensive example of using the fetch_scores() function from Langfuse to retrieve and analyze evaluation metrics, integrating UpTrain and Ragas for model evaluation.

• Added pages/docs/scores/example_usage_of_fetch_score.md with detailed code snippets for setting up, evaluating models, logging scores, and visualizing correlations
• Updated pages/guides/cookbook/example_external_evaluation_pipelines.md with a guide on creating external evaluation pipelines using Langfuse, including synthetic data creation and custom evaluations
• Made minor formatting and content improvements across multiple integration cookbooks (DSPy, Instructor, LangGraph, etc.) to enhance readability and consistency
• Updated various Langchain examples to demonstrate better integration with Langfuse for tracing and prompt management

_{26 file(s) reviewed, 7 comment(s)
Edit PR Review Bot Settings | Greptile}

[marcklingen]

thanks for the contribution. It seems like you mostly want to showcase correlation analysis of different scores in Langfuse (which is a good notebook example). Are you sure that your example correlates the scores on a single trace basis for the analysis at the bottom of this notebook

[Sohammhatre10]

@marcklingen Yupp, this was based on a single trace, and the scores were fetched accordingly. Haven't used any specifics for traces, but this was the first trace I created, so it defaulted to the first trace. Should I add more specificity for a single trace? Apologies for the late reply.