plexus.scores.nodes.test_classifier_confidence_integration module
Integration test for Classifier confidence feature with real OpenAI API.
This test runs actual classifications using the OpenAI API to verify: 1. Confidence calculation works with real API responses 2. Logprobs are extracted correctly 3. Confidence values are reasonable 4. The feature integrates properly with the full workflow
Run with: python -m pytest plexus/scores/nodes/test_classifier_confidence_integration.py -v -s
- plexus.scores.nodes.test_classifier_confidence_integration.openai_confidence_config()
Configuration for a binary classifier with confidence enabled.
- async plexus.scores.nodes.test_classifier_confidence_integration.test_confidence_with_real_openai_api(openai_confidence_config)
Test confidence calculation with real OpenAI API calls.
- async plexus.scores.nodes.test_classifier_confidence_integration.test_confidence_workflow_integration(openai_confidence_config)
Test that confidence integrates properly with the full LangGraph workflow.