---
license: gpl-3.0
pretty_name: PredictaBoard_Benchmarks
configs:
  - config_name: Big-Bench-Hard
    data_files:
      - split: boolean_expressions
        path: data/boolean_expressions.parquet
      - split: causal_judgement
        path: data/causal_judgement.parquet
      - split: date_understanding
        path: data/date_understanding.parquet
      - split: disambiguation_qa
        path: data/disambiguation_qa.parquet
      - split: formal_fallacies
        path: data/formal_fallacies.parquet
      - split: geometric_shapes
        path: data/geometric_shapes.parquet
      - split: hyperbaton
        path: data/hyperbaton.parquet
      - split: logical_deduction_five_objects
        path: data/logical_deduction_five_objects.parquet
      - split: logical_deduction_seven_objects
        path: data/logical_deduction_seven_objects.parquet
      - split: logical_deduction_three_objects
        path: data/logical_deduction_three_objects.parquet
      - split: movie_recommendation
        path: data/movie_recommendation.parquet
      - split: navigate
        path: data/navigate.parquet
      - split: object_counting
        path: data/object_counting.parquet
      - split: penguins_in_a_table
        path: data/penguins_in_a_table.parquet
      - split: reasoning_about_colored_objects
        path: data/reasoning_about_colored_objects.parquet
      - split: ruin_names
        path: data/ruin_names.parquet
      - split: salient_translation_error_detection
        path: data/salient_translation_error_detection.parquet
      - split: snarks
        path: data/snarks.parquet
      - split: sports_understanding
        path: data/sports_understanding.parquet
      - split: tracking_shuffled_objects_five_objects
        path: data/tracking_shuffled_objects_five_objects.parquet
      - split: tracking_shuffled_objects_seven_objects
        path: data/tracking_shuffled_objects_seven_objects.parquet
      - split: tracking_shuffled_objects_three_objects
        path: data/tracking_shuffled_objects_three_objects.parquet
      - split: web_of_lies
        path: data/web_of_lies.parquet
  - config_name: IFEVAL
    data_files: data/IFEVAL.parquet
  - config_name: MATH-LvL-5
    data_files:
      - split: beginner_algebra_hard
        path: data/beginner_algebra_hard.parquet
      - split: counting_and_prob_hard
        path: data/counting_and_prob_hard.parquet
      - split: geometry_hard
        path: data/geometry_hard.parquet
      - split: intermediate_algebra_hard
        path: data/intermediate_algebra_hard.parquet
      - split: num_theory_hard
        path: data/num_theory_hard.parquet
      - split: prealgebra_hard
        path: data/prealgebra_hard.parquet
      - split: precalculus_hard
        path: data/precalculus_hard.parquet
  - config_name: MMLU-Pro
    data_files: data/MMLU-Pro.parquet
  - config_name: MuSR
    data_files:
      - split: murder_mysteries
        path: data/murder_mysteries.parquet
      - split: object_placements
        path: data/object_placements.parquet
      - split: team_allocation
        path: data/team_allocation.parquet
---

# PredictaBoard

PredictaBoard is a novel benchmark for measuring the predictability of Large Language Models.