plexus.cli.analyze.analysis module

class plexus.cli.analyze.analysis.PromptAnalyzer(llm)

Bases: object

__init__(llm)

analyze_feedback(current_prompt: str, examples: list)

plexus.cli.analyze.analysis.analyze_score_feedback(score_data: DataFrame, prompt_analyzer: PromptAnalyzer, current_prompt: str)

Analyze feedback for a specific score type and suggest prompt improvements.

plexus.cli.analyze.analysis.analyze_topics(*args, **kwargs)

Lazy wrapper for analyze_topics to avoid loading PyTorch unless needed.