---
dataset_info:
  features:
  - name: answer_idx
    dtype: int64
  - name: source
    dtype: string
  - name: metadata
    dtype: string
  - name: prompt
    dtype: string
  - name: answer_letter
    dtype: string
  - name: answer_string
    dtype: string
  - name: reasoning
    dtype: string
  - name: distilled_answer_string
    dtype: string
  - name: domain_code
    dtype: string
  - name: domain_name
    dtype: string
  - name: text
    dtype: string
  - name: decomposer_raw_output
    dtype: string
  - name: subproblems
    list:
    - name: critical
      dtype: bool
    - name: index
      dtype: int64
    - name: subproblem
      dtype: string
  - name: analyzer_raw_output
    dtype: string
  - name: subproblem_analysis
    list:
    - name: explanation
      dtype: string
    - name: status
      dtype: string
    - name: subproblem_index
      dtype: int64
  - name: evaluator_raw_output
    dtype: string
  - name: consistency_evaluation
    struct:
    - name: confidence_score
      dtype: float64
    - name: consistent
      dtype: bool
    - name: explanation
      dtype: string
  - name: entity_extract_response
    dtype: string
  - name: extracted_entities
    list:
    - name: anatomy
      dtype: string
    - name: disease
      dtype: string
    - name: gene/protein
      dtype: string
    - name: id
      dtype: string
    - name: 'id:'
      dtype: string
    - name: molecular_function
      dtype: string
    - name: name
      dtype: string
    - name: pathway
      dtype: string
    - name: type
      dtype: string
  - name: needs_knowledge_correction
    dtype: bool
  - name: entity_links
    list:
    - name: links
      list:
      - name: name
        dtype: string
      - name: score
        dtype: float64
    - name: mention
      dtype: string
    - name: type
      dtype: string
  - name: unfiltered_triplets
    sequence:
      sequence: string
  - name: filter_response
    dtype: string
  - name: filter_result
    struct:
    - name: relevant_triplets
      list:
      - name: explanation
        dtype: string
      - name: id
        dtype: int64
      - name: relevance
        dtype: string
  - name: filtered_triplets
    sequence:
      sequence: string
  - name: original_distilled_answer_string
    dtype: string
  - name: knowledge_correction_applied
    dtype: bool
  - name: correction_info
    struct:
    - name: corrected_length
      dtype: int64
    - name: critical_issues_count
      dtype: int64
    - name: knowledge_triplets_count
      dtype: int64
    - name: original_length
      dtype: int64
  splits:
  - name: train
    num_bytes: 24179326
    num_examples: 1000
  download_size: 10911562
  dataset_size: 24179326
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---