Datasets:

ogkranthi
/

metacog-bench

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
version: string
description: string
tiers: struct<tier_1a_betting: struct<name: string, dimension: string, n_tasks: int64, tasks: list<item: st (... 838 chars omitted)
  child 0, tier_1a_betting: struct<name: string, dimension: string, n_tasks: int64, tasks: list<item: struct<id: string, questio (... 54 chars omitted)
      child 0, name: string
      child 1, dimension: string
      child 2, n_tasks: int64
      child 3, tasks: list<item: struct<id: string, question: string, correct_answer: string, category: string>>
          child 0, item: struct<id: string, question: string, correct_answer: string, category: string>
              child 0, id: string
              child 1, question: string
              child 2, correct_answer: string
              child 3, category: string
  child 1, tier_1b_predict_then_perform: struct<name: string, dimension: string, n_tasks: int64, tasks: list<item: struct<id: string, questio (... 84 chars omitted)
      child 0, name: string
      child 1, dimension: string
      child 2, n_tasks: int64
      child 3, tasks: list<item: struct<id: string, question: string, answer: string, difficulty: string, check_keywords:  (... 20 chars omitted)
          child 0, item: struct<id: string, question: string, answer: string, difficulty: string, check_keywords: list<item:  (... 8 chars omitted)
              child 0, id: string
              child 1, question: string
              child 2, answer: string
              child 3, difficulty: string
        
...
lues: list<item: double>>
      child 0, mean: double
      child 1, std: double
      child 2, values: list<item: double>
          child 0, item: double
  child 5, tier3_rate: struct<mean: double, std: double, values: list<item: double>>
      child 0, mean: double
      child 1, std: double
      child 2, values: list<item: double>
          child 0, item: double
  child 6, MS: struct<mean: double, std: double, values: list<item: double>>
      child 0, mean: double
      child 1, std: double
      child 2, values: list<item: double>
          child 0, item: double
  child 7, SAF: struct<mean: double, std: double, values: list<item: double>>
      child 0, mean: double
      child 1, std: double
      child 2, values: list<item: double>
          child 0, item: double
  child 8, CDTC: struct<mean: double, std: double, values: list<item: double>>
      child 0, mean: double
      child 1, std: double
      child 2, values: list<item: double>
          child 0, item: double
  child 9, overall: struct<mean: double, std: double, values: list<item: double>>
      child 0, mean: double
      child 1, std: double
      child 2, values: list<item: double>
          child 0, item: double
  child 10, post_ece: struct<mean: double, std: double, values: list<item: double>>
      child 0, mean: double
      child 1, std: double
      child 2, values: list<item: double>
          child 0, item: double
num_runs: int64
model: string
result_files: list<item: string>
  child 0, item: string
to
{'model': Value('string'), 'num_runs': Value('int64'), 'result_files': List(Value('string')), 'metrics': {'tier1a_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier1b_score': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier1b_ece': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier2a_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier2b_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier3_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'MS': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'SAF': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'CDTC': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'overall': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'post_ece': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}}}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 99, in get_rows_or_raise
                  return get_rows(
                         ^^^^^^^^^
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                         ^^^^^^^^^^^^^^^^^^^^^
                File "/src/services/worker/src/worker/utils.py", line 77, in get_rows
                  rows_plus_one = list(itertools.islice(ds, rows_max_number + 1))
                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2690, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2227, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2251, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 494, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 384, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 265, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 120, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2272, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2218, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              version: string
              description: string
              tiers: struct<tier_1a_betting: struct<name: string, dimension: string, n_tasks: int64, tasks: list<item: st (... 838 chars omitted)
                child 0, tier_1a_betting: struct<name: string, dimension: string, n_tasks: int64, tasks: list<item: struct<id: string, questio (... 54 chars omitted)
                    child 0, name: string
                    child 1, dimension: string
                    child 2, n_tasks: int64
                    child 3, tasks: list<item: struct<id: string, question: string, correct_answer: string, category: string>>
                        child 0, item: struct<id: string, question: string, correct_answer: string, category: string>
                            child 0, id: string
                            child 1, question: string
                            child 2, correct_answer: string
                            child 3, category: string
                child 1, tier_1b_predict_then_perform: struct<name: string, dimension: string, n_tasks: int64, tasks: list<item: struct<id: string, questio (... 84 chars omitted)
                    child 0, name: string
                    child 1, dimension: string
                    child 2, n_tasks: int64
                    child 3, tasks: list<item: struct<id: string, question: string, answer: string, difficulty: string, check_keywords:  (... 20 chars omitted)
                        child 0, item: struct<id: string, question: string, answer: string, difficulty: string, check_keywords: list<item:  (... 8 chars omitted)
                            child 0, id: string
                            child 1, question: string
                            child 2, answer: string
                            child 3, difficulty: string
                      
              ...
              lues: list<item: double>>
                    child 0, mean: double
                    child 1, std: double
                    child 2, values: list<item: double>
                        child 0, item: double
                child 5, tier3_rate: struct<mean: double, std: double, values: list<item: double>>
                    child 0, mean: double
                    child 1, std: double
                    child 2, values: list<item: double>
                        child 0, item: double
                child 6, MS: struct<mean: double, std: double, values: list<item: double>>
                    child 0, mean: double
                    child 1, std: double
                    child 2, values: list<item: double>
                        child 0, item: double
                child 7, SAF: struct<mean: double, std: double, values: list<item: double>>
                    child 0, mean: double
                    child 1, std: double
                    child 2, values: list<item: double>
                        child 0, item: double
                child 8, CDTC: struct<mean: double, std: double, values: list<item: double>>
                    child 0, mean: double
                    child 1, std: double
                    child 2, values: list<item: double>
                        child 0, item: double
                child 9, overall: struct<mean: double, std: double, values: list<item: double>>
                    child 0, mean: double
                    child 1, std: double
                    child 2, values: list<item: double>
                        child 0, item: double
                child 10, post_ece: struct<mean: double, std: double, values: list<item: double>>
                    child 0, mean: double
                    child 1, std: double
                    child 2, values: list<item: double>
                        child 0, item: double
              num_runs: int64
              model: string
              result_files: list<item: string>
                child 0, item: string
              to
              {'model': Value('string'), 'num_runs': Value('int64'), 'result_files': List(Value('string')), 'metrics': {'tier1a_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier1b_score': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier1b_ece': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier2a_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier2b_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'tier3_rate': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'MS': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'SAF': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'CDTC': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'overall': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}, 'post_ece': {'mean': Value('float64'), 'std': Value('float64'), 'values': List(Value('float64'))}}}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

MetaCog-Bench: A Process-Based Benchmark for Evaluating Metacognitive Monitoring and Control in LLMs

Overview

MetaCog-Bench is a benchmark of 147 tasks across 5 tiers for evaluating metacognitive monitoring and control in large language models, grounded in the Nelson & Narens (1990) metacognition framework.

All evaluation is fully deterministic (regex, keyword matching, JSON verification, ECE computation) — zero LLM-as-judge bias.

Three Metacognitive Dimensions

Dimension	Description	Tiers
MS (Metacognitive Sensitivity)	Confidence calibration & knowledge boundary awareness	Tier 1a (Betting), Tier 1b (Predict-Then-Perform)
SAF (Strategy Adaptation Frequency)	Resistance to ecological validity traps & sycophantic pressure	Tier 2a (Ecological Validity), Tier 2b (Sycophancy Resistance)
CDTC (Cross-Domain Transfer Coefficient)	Applying formal reasoning to naturalistic problem framings	Tier 3 (Domain Transfer)

Task Counts

Tier	Tasks
1a — Betting Calibration	33
1b — Predict-Then-Perform	34
2a — Ecological Validity	30
2b — Sycophancy Resistance	30
3 — Domain Transfer	20
Total	147

Leaderboard (7 Models, 3 Runs Each)

Rank	Model	Overall (mean +/- std)	MS	SAF	CDTC
1	Grok-3-mini-fast	0.864 +/- 0.009	0.743	1.000	0.850
2	DeepSeek-V3	0.859 +/- 0.007	0.743	1.000	0.833
3	Gemini 2.5 Flash	0.812 +/- 0.006	0.602	0.983	0.850
4	Claude Sonnet 4	0.812 +/- 0.008	0.614	0.972	0.850
5	Mistral Large	0.768 +/- 0.006	0.664	0.989	0.650
6	GPT-4o	0.742 +/- 0.020	0.662	0.917	0.650
7	Open-Mistral-Nemo (12B)	0.710 +/- 0.026	0.557	0.956	0.617

Files

tasks_v5.json — All 147 tasks with prompts, answers, and evaluation criteria
results/ — Aggregated results (mean +/- std) for all 7 models across 3 runs

Usage

from datasets import load_dataset
ds = load_dataset("ogkranthi/metacog-bench")

Or load the raw JSON:

import json
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="ogkranthi/metacog-bench", filename="tasks_v5.json", repo_type="dataset")
with open(path) as f:
    tasks = json.load(f)

Citation

@inproceedings{metacogbench2026,
  title={MetaCog-Bench: A Process-Based Benchmark for Evaluating Metacognitive Monitoring and Control in Large Language Models},
  author={Anonymous},
  booktitle={Advances in Neural Information Processing Systems (NeurIPS) Evaluations and Datasets Track},
  year={2026}
}