Insufficient metric
There are various metrics available for RAG.
These include projects such as deepeval, or Galileo.
With the addition of that metric, more diverse metrics should be accommodated in AutoRAG.

Describe the solution you'd like

• make rag subject schema for input of metric
• every metric function use this schema as input.

Describe alternatives you've considered

• #649
• #655

gallileo metrics

• #650
• add gallileo_sexism
• add gallileo_tone
• add gallileo_toxicity
• add gallileo_private_identifiable_information
• add gallileo_context_relevance
• add gallileo_prompt_perplexity
• add gallileo_uncertainty
• add gallileo_chunk_utilization
• add gallileo_chunk_attribution
• add gallileo_completeness
• add gallileo_correctness
• add gallileo_context_adherence
• add gallileo_instruction_adherence
• add gallileo_chunk_relevance

MLflow

• add mlflow_relevance
• add mlflow_faithfulness
• add mlflow_answer_relevance
• add mlflow_answer_similarity
• add mlflow_anwer_correctness

deepeval metrics

• #654
• #752
• #753
• #754
• #755
• #756
• #757
• #758
• #759

trulens

• add trulens_answer relevancy
• add trulens_context_relevancy
• add trulens_groundedness
• add trulens_coherence
• add trulens_comprehensiveness
• add trulens_conciseness
• add trulens_controversialty
• add trulens_correctness
• add trulens_criminality
• add trulens_generate_score
• add trulens_groundedness
• add trulens_harmfulness
• add trulens_helpfulness
• add trulens_insensitivity
• add trulens_maliciousness
• add trulens_misogyny
• add trulens_relevance
• add trulens_sentiment
• add trulens_strereotypes
• add trulens_summarization

openai

• add openai_hate
• add openai_harassment
• add openai_self-harm
• add openai_sexual
• add openai_violence