Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed because of a cast error
Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 8 new columns ({'n', 'agreement', 'disagreements', 'comparison', 'mean |score gap|', 'kappa reads as', 'kappa', 'unparseable'}) and 6 missing columns ({'type', 'tests', 'acceptable', 'prompt', 'gold', 'id'}).

This happened while the csv dataset builder was generating data using

hf://datasets/richardyoung/lab07-catch-the-ai-lying/lab07_agreement.csv (at revision 7309f311e628872c15c06fc055e04b4e590dbf5b), ['hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/gold_qa.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_agreement.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_answers.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_confusion.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_judge_runs.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_verdicts.csv']

Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1848, in _prepare_split_single
                  writer.write_table(table)
                  ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 765, in write_table
                  self._write_table(pa_table, writer_batch_size=writer_batch_size)
                  ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              comparison: string
              n: int64
              agreement: double
              disagreements: int64
              unparseable: int64
              kappa: double
              kappa reads as: string
              mean |score gap|: string
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 1228
              to
              {'id': Value('string'), 'type': Value('string'), 'prompt': Value('string'), 'gold': Value('string'), 'acceptable': Value('string'), 'tests': Value('string')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1850, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
                  ...<4 lines>...
                  )
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 8 new columns ({'n', 'agreement', 'disagreements', 'comparison', 'mean |score gap|', 'kappa reads as', 'kappa', 'unparseable'}) and 6 missing columns ({'type', 'tests', 'acceptable', 'prompt', 'gold', 'id'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/richardyoung/lab07-catch-the-ai-lying/lab07_agreement.csv (at revision 7309f311e628872c15c06fc055e04b4e590dbf5b), ['hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/gold_qa.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_agreement.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_answers.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_confusion.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_judge_runs.csv', 'hf://datasets/richardyoung/lab07-catch-the-ai-lying@7309f311e628872c15c06fc055e04b4e590dbf5b/lab07_verdicts.csv']
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)

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.

id
string
type
string
prompt
string
gold
string
acceptable
string
tests
string
icd10_t2dm
code_lookup
A hospital problem list is being migrated to ICD-10-CM. A 54-year-old has type 2 diabetes with no documented complications. Which single ICD-10-CM code should be used? Give the code only.
E11.9
[]
a pure lookup with one right answer; exact match should never miss this
icd10_ami
code_lookup
A 61-year-old is admitted with an acute myocardial infarction and the documentation does not specify the site. Which single ICD-10-CM code applies? Give the code only.
I21.9
[]
same shape as the item above, so a scorer that passes one and fails the other is telling you something about the scorer
snomed_t2dm
code_lookup
Give the SNOMED CT concept identifier (SCTID) for the concept 'Diabetes mellitus type 2'. Give the number only.
44054006
[]
a numeric code; models love to attach a label to it, and a label is enough to break exact match
acetaminophen_max_g
numeric
In a healthy adult with normal liver function, the maximum total daily dose of acetaminophen (paracetamol) is how many grams? Give the number only.
4
['4 g', '4 grams', '4g', '4.0']
a number asked for without units -- models hand you '4 grams' anyway, which is the entire problem exact match has
a1c_8_eag_mgdl
numeric
Using the ADAG formula, eAG (mg/dL) = 28.7 x A1c - 46.7, what estimated average glucose in mg/dL corresponds to an A1c of 8.0%? Round to the nearest whole number and give the number only.
183
['183 mg/dl', '182.9', '183mg/dl']
arithmetic, not recall -- the prompt hands over the formula, so a wrong answer here is a wrong calculation and nothing else
hypo_level1_mgdl
numeric
In a person with diabetes, Level 1 hypoglycaemia is defined as a plasma glucose below what value, in mg/dL? Give the number only.
70
['70 mg/dl', '<70', 'below 70', 'less than 70']
a threshold with two legitimate unit systems. '3.9' is the same threshold in mmol/L: clinically right, string-wrong. Deliberately NOT in `acceptable` -- watch what each scorer does with it
hipaa_safe_harbor
numeric
A research team wants to release a dataset under the HIPAA Safe Harbor de-identification method. How many identifiers must be removed? Give the number only.
18
['18 identifiers']
the de-identification rule from Lab 1, and the item that makes this a health-analytics gold set rather than a purely clinical one
warfarin_reversal_agent
short_phrase
A 74-year-old on warfarin arrives with life-threatening gastrointestinal bleeding and an INR of 6.8. Name the agent that reverses warfarin by replacing clotting factors within minutes.
4-factor prothrombin complex concentrate
['4 factor prothrombin complex concentrate', '4f-pcc', 'pcc', 'prothrombin complex concentrate', 'kcentra']
a multi-word phrase with several accepted names -- where the `acceptable` list starts to look arbitrary and you have to defend it
sjw_warfarin_yesno
yes_no
A 71-year-old on warfarin with a stable INR of 2.4 asks whether she can start over-the-counter St John's wort for low mood. Answer YES or NO.
NO
['no']
one bit of information, and models almost never give you only one bit -- the cleanest possible demonstration of a brittle scorer
atorva80_intensity
one_word
Is atorvastatin 80 mg daily a moderate-intensity or a high-intensity statin? Answer with one word.
high
['high-intensity', 'high intensity']
an explicit one-word constraint, so whether the model OBEYED AN INSTRUCTION is itself something you can score
null
null
null
null
null
null
null
null
null
null
null
null
icd10_t2dm
code_lookup
A hospital problem list is being migrated to ICD-10-CM. A 54-year-old has type 2 diabetes with no documented complications. Which single ICD-10-CM code should be used? Give the code only.
E11.9
null
null
icd10_ami
code_lookup
A 61-year-old is admitted with an acute myocardial infarction and the documentation does not specify the site. Which single ICD-10-CM code applies? Give the code only.
I21.9
null
null
snomed_t2dm
code_lookup
Give the SNOMED CT concept identifier (SCTID) for the concept 'Diabetes mellitus type 2'. Give the number only.
44054006
null
null
acetaminophen_max_g
numeric
In a healthy adult with normal liver function, the maximum total daily dose of acetaminophen (paracetamol) is how many grams? Give the number only.
4
4 g|4 grams|4g|4.0
null
a1c_8_eag_mgdl
numeric
Using the ADAG formula, eAG (mg/dL) = 28.7 x A1c - 46.7, what estimated average glucose in mg/dL corresponds to an A1c of 8.0%? Round to the nearest whole number and give the number only.
183
183 mg/dl|182.9|183mg/dl
null
hypo_level1_mgdl
numeric
In a person with diabetes, Level 1 hypoglycaemia is defined as a plasma glucose below what value, in mg/dL? Give the number only.
70
70 mg/dl|<70|below 70|less than 70
null
hipaa_safe_harbor
numeric
A research team wants to release a dataset under the HIPAA Safe Harbor de-identification method. How many identifiers must be removed? Give the number only.
18
18 identifiers
null
warfarin_reversal_agent
short_phrase
A 74-year-old on warfarin arrives with life-threatening gastrointestinal bleeding and an INR of 6.8. Name the agent that reverses warfarin by replacing clotting factors within minutes.
4-factor prothrombin complex concentrate
4 factor prothrombin complex concentrate|4f-pcc|pcc|prothrombin complex concentrate|kcentra
null
sjw_warfarin_yesno
yes_no
A 71-year-old on warfarin with a stable INR of 2.4 asks whether she can start over-the-counter St John's wort for low mood. Answer YES or NO.
NO
no
null
atorva80_intensity
one_word
Is atorvastatin 80 mg daily a moderate-intensity or a high-intensity statin? Answer with one word.
high
high-intensity|high intensity
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
null
icd10_t2dm
code_lookup
A hospital problem list is being migrated to ICD-10-CM. A 54-year-old has type 2 diabetes with no documented complications. Which single ICD-10-CM code should be used? Give the code only.
E11.9
null
null
icd10_ami
code_lookup
A 61-year-old is admitted with an acute myocardial infarction and the documentation does not specify the site. Which single ICD-10-CM code applies? Give the code only.
I21.9
null
null
snomed_t2dm
code_lookup
Give the SNOMED CT concept identifier (SCTID) for the concept 'Diabetes mellitus type 2'. Give the number only.
44054006
null
null
acetaminophen_max_g
numeric
In a healthy adult with normal liver function, the maximum total daily dose of acetaminophen (paracetamol) is how many grams? Give the number only.
4
4 g|4 grams|4g|4.0
null
a1c_8_eag_mgdl
numeric
Using the ADAG formula, eAG (mg/dL) = 28.7 x A1c - 46.7, what estimated average glucose in mg/dL corresponds to an A1c of 8.0%? Round to the nearest whole number and give the number only.
183
183 mg/dl|182.9|183mg/dl
null
hypo_level1_mgdl
numeric
In a person with diabetes, Level 1 hypoglycaemia is defined as a plasma glucose below what value, in mg/dL? Give the number only.
70
70 mg/dl|<70|below 70|less than 70
null
hipaa_safe_harbor
numeric
A research team wants to release a dataset under the HIPAA Safe Harbor de-identification method. How many identifiers must be removed? Give the number only.
18
18 identifiers
null
warfarin_reversal_agent
short_phrase
A 74-year-old on warfarin arrives with life-threatening gastrointestinal bleeding and an INR of 6.8. Name the agent that reverses warfarin by replacing clotting factors within minutes.
4-factor prothrombin complex concentrate
4 factor prothrombin complex concentrate|4f-pcc|pcc|prothrombin complex concentrate|kcentra
null
sjw_warfarin_yesno
yes_no
A 71-year-old on warfarin with a stable INR of 2.4 asks whether she can start over-the-counter St John's wort for low mood. Answer YES or NO.
NO
no
null
atorva80_intensity
one_word
Is atorvastatin 80 mg daily a moderate-intensity or a high-intensity statin? Answer with one word.
high
high-intensity|high intensity
null

No dataset card yet

Downloads last month
23

Space using richardyoung/lab07-catch-the-ai-lying 1