Referência da API

Esta página fornece um índice abrangente das APIs do MLflow usadas nos aplicativos GenAI, com links diretos para a documentação oficial do MLflow.

Links de documentação oficial

• Todas as APIs Python do MLflow
• APIs principais do MLflow
• Módulo MLflow GenAI
• MLflow Tracing APIs
• APIs de cliente do MLflow
• Entidades do MLflow

info

Beta

Algumas das APIs mencionadas nesta página estão atualmente em versão beta.

Gerenciamento de experimentos

SDKs

• mlflow.search_runs()
• mlflow.set_experiment()
• mlflow.start_run()

Entidades

• mlflow.entities.Experiment
• mlflow.entities.Run

Gerenciamento imediato

SDKs

• mlflow.genai.load_prompt()
• mlflow.genai.optimize_prompt()
• mlflow.genai.register_prompt()
• mlflow.genai.search_prompts()
• mlflow.genai.delete_prompt_alias()
• mlflow.genai.set_prompt_alias()

Entidades

• mlflow.entities.Prompt

Avaliação

• mlflow.genai.evaluate()
• mlflow.genai.to_predict_fn()

Classe de marcador personalizada

• mlflow.genai.Scorer

Decorador de pontuação personalizado

• mlflow.genai.scorer()

Marcadores predefinidos

• mlflow.genai.scorers.Correctness
• mlflow.genai.scorers.ExpectationsGuidelines
• mlflow.genai.scorers.Guidelines
• mlflow.genai.scorers.RelevanceToQuery
• mlflow.genai.scorers.RetrievalGroundedness
• mlflow.genai.scorers.RetrievalRelevance
• mlflow.genai.scorers.RetrievalSufficiency
• mlflow.genai.scorers.Safety

Ajudantes de goleador

• mlflow.genai.scorers.get_all_scorers()

SDKs do Judge

• mlflow.genai.judges.custom_prompt_judge()
• mlflow.genai.judges.is_context_relevant()
• mlflow.genai.judges.is_context_sufficient()
• mlflow.genai.judges.is_correct()
• mlflow.genai.judges.is_grounded()
• mlflow.genai.judges.is_safe()
• mlflow.genai.judges.meets_guidelines()

Julgue entidades de saída

• mlflow.genai.judges.CategoricalRating
  • mlflow.genai.judges.CategoricalRating.NO
  • mlflow.genai.judges.CategoricalRating.UNKNOWN
  • mlflow.genai.judges.CategoricalRating.YES

Conjunto de dados de avaliação

SDKs

• mlflow.genai.create_dataset()
• mlflow.genai.delete_dataset()
• mlflow.genai.get_dataset()

Entidades

• mlflow.genai.datasets.EvaluationDataset
  • merge_records()
  • set_profile()
  • to_df()
  • to_evaluation_dataset()

Human rótulo & Review App

Entidades

• mlflow.genai.Agent

• mlflow.genai.LabelingSession

  • add_dataset()
  • add_traces()
  • set_assigned_users()
  • sync()

• mlflow.genai.ReviewApp

  • add_agent()
  • remove_agent()

rótulo session SDKs

• mlflow.genai.create_labeling_session()
• mlflow.genai.delete_labeling_session()
• mlflow.genai.get_labeling_session()
• mlflow.genai.get_labeling_sessions()
• mlflow.genai.get_review_app()

rótulo Tipos de esquema

• mlflow.genai.label_schemas.InputCategorical
• mlflow.genai.label_schemas.InputCategoricalList
• mlflow.genai.label_schemas.InputNumeric
• mlflow.genai.label_schemas.InputText
• mlflow.genai.label_schemas.InputTextList
• mlflow.genai.label_schemas.LabelSchema
• mlflow.genai.label_schemas.LabelSchemaType
  • mlflow.genai.label_schemas.LabelSchemaType.EXPECTATION
  • mlflow.genai.label_schemas.LabelSchemaType.FEEDBACK

rótulo Schema SDKs

• mlflow.genai.label_schemas.create_label_schema()
• mlflow.genai.label_schemas.delete_label_schema()
• mlflow.genai.label_schemas.get_label_schema()

Otimização rápida

Entidades

• mlflow.genai.optimize.LLMParams
• mlflow.genai.optimize.OptimizerConfig
• mlflow.genai.optimize.PromptOptimizationResult

SDKs

• mlflow.genai.optimize.optimize_prompt()

Rastreamento

SDKs

• mlflow.delete_trace_tag()
• mlflow.get_current_active_span()
• mlflow.get_last_active_trace()
• mlflow.get_last_active_trace_id()
• mlflow.get_trace()
• mlflow.search_traces()
• mlflow.set_trace_tag()
• mlflow.start_span()
• mlflow.trace
• mlflow.traceName
• mlflow.traceOutputs
• mlflow.tracing
• mlflow.tracing.disable
• mlflow.tracing.disable_notebook_display()
• mlflow.tracing.enable
• mlflow.tracing.enable_notebook_display()
• mlflow.update_current_trace()

Entidades

• mlflow.entities.Trace
• mlflow.entities.TraceData
• mlflow.entities.TraceInfo
• mlflow.entities.Span
• mlflow.entities.SpanEvent
• mlflow.entities.SpanType
• mlflow.entities.Document

Entidades de avaliação (feedback, expectativa)

• mlflow.entities.Assessment
• mlflow.entities.AssessmentError
• mlflow.entities.AssessmentSource
• mlflow.entities.AssessmentSourceType
• mlflow.entities.Expectation
• mlflow.entities.Feedback

Integrações de rastreamento

• mlflow.anthropic.autolog
• mlflow.autogen.autolog
• mlflow.bedrock.autolog
• mlflow.crewai.autolog
• mlflow.dspy.autolog
• mlflow.gemini.autolog
• mlflow.groq.autolog
• mlflow.langchain.autolog
• mlflow.litellm.autolog
• mlflow.llama_index.autolog
• mlflow.mistral.autolog
• mlflow.openai.autolog

Versão acompanhamento

SDKs

• mlflow.set_active_model()
• mlflow.clear_active_model()
• mlflow.get_active_model_id()
• mlflow.create_external_model()
• mlflow.delete_logged_model_tag()
• mlflow.finalize_logged_model()
• mlflow.get_logged_model()
• mlflow.initialize_logged_model()
• mlflow.last_logged_model()
• mlflow.search_logged_models()
• mlflow.set_logged_model_tags()
• mlflow.log_model_params()

Entidades

• mlflow.entities.LoggedModel
• mlflow.entities.LoggedModelStatus
• mlflow.ActiveModel