The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
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 4 new columns ({'q1', 'measure', 'median', 'q3'}) and 1 missing columns ({'metric'}).
This happened while the csv dataset builder was generating data using
hf://datasets/Thang203/user_study_result/outcomes.csv (at revision 817e9bc9e62345097d13fc7c6f0bd44fca27ddb9), ['hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/behavior_by_facet.csv', 'hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/behavior_pooled.csv', 'hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/outcomes.csv', 'hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/rows.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
facet: string
measure: string
condition: string
n: int64
mean: double
sd: double
se: double
median: double
q1: double
q3: double
-- schema metadata --
pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 1361
to
{'facet': Value('string'), 'metric': Value('string'), 'condition': Value('string'), 'n': Value('int64'), 'mean': Value('float64'), 'sd': Value('float64'), 'se': Value('float64')}
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 4 new columns ({'q1', 'measure', 'median', 'q3'}) and 1 missing columns ({'metric'}).
This happened while the csv dataset builder was generating data using
hf://datasets/Thang203/user_study_result/outcomes.csv (at revision 817e9bc9e62345097d13fc7c6f0bd44fca27ddb9), ['hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/behavior_by_facet.csv', 'hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/behavior_pooled.csv', 'hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/outcomes.csv', 'hf://datasets/Thang203/user_study_result@817e9bc9e62345097d13fc7c6f0bd44fca27ddb9/rows.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.
facet string | metric string | condition string | n int64 | mean float64 | sd float64 | se float64 |
|---|---|---|---|---|---|---|
Find × Text | Scroll count | Non-grounded | 31 | 9.4839 | 8.2457 | 1.481 |
Find × Text | Scroll count | Grounded | 34 | 10.6471 | 6.2855 | 1.078 |
Find × Visual | Scroll count | Non-grounded | 24 | 20.7917 | 14.4342 | 2.9464 |
Find × Visual | Scroll count | Grounded | 21 | 21.1429 | 13.7851 | 3.0082 |
Guide × Text | Scroll count | Non-grounded | 20 | 10.2 | 6.8641 | 1.5349 |
Guide × Text | Scroll count | Grounded | 29 | 13.2069 | 9.1705 | 1.7029 |
Guide × Visual | Scroll count | Non-grounded | 27 | 9.2593 | 7.1929 | 1.3843 |
Guide × Visual | Scroll count | Grounded | 31 | 11.6129 | 8.0361 | 1.4433 |
Find × Text | Ctrl+F | Non-grounded | 31 | 1.1935 | 1.0776 | 0.1935 |
Find × Text | Ctrl+F | Grounded | 34 | 1.1176 | 1.2972 | 0.2225 |
Find × Visual | Ctrl+F | Non-grounded | 24 | 0.5 | 0.5898 | 0.1204 |
Find × Visual | Ctrl+F | Grounded | 21 | 0.3333 | 0.5774 | 0.126 |
Guide × Text | Ctrl+F | Non-grounded | 20 | 0 | 0 | 0 |
Guide × Text | Ctrl+F | Grounded | 29 | 0.0345 | 0.1857 | 0.0345 |
Guide × Visual | Ctrl+F | Non-grounded | 27 | 0 | 0 | 0 |
Guide × Visual | Ctrl+F | Grounded | 31 | 0 | 0 | 0 |
Find × Text | Text selections | Non-grounded | 31 | 2.2581 | 3.1514 | 0.566 |
Find × Text | Text selections | Grounded | 34 | 1.3235 | 1.7705 | 0.3036 |
Find × Visual | Text selections | Non-grounded | 24 | 1.5417 | 2.2646 | 0.4623 |
Find × Visual | Text selections | Grounded | 21 | 0.9524 | 2.0119 | 0.439 |
Guide × Text | Text selections | Non-grounded | 20 | 2.2 | 6.229 | 1.3928 |
Guide × Text | Text selections | Grounded | 29 | 1.0345 | 1.742 | 0.3235 |
Guide × Visual | Text selections | Non-grounded | 27 | 1.037 | 2.3119 | 0.4449 |
Guide × Visual | Text selections | Grounded | 31 | 0.871 | 1.7653 | 0.317 |
Find × Text | Mouse clicks | Non-grounded | 31 | 16.5806 | 11.0778 | 1.9896 |
Find × Text | Mouse clicks | Grounded | 34 | 13.9118 | 5.9968 | 1.0284 |
Find × Visual | Mouse clicks | Non-grounded | 24 | 14.125 | 5.3998 | 1.1022 |
Find × Visual | Mouse clicks | Grounded | 21 | 13.1429 | 8.6272 | 1.8826 |
Guide × Text | Mouse clicks | Non-grounded | 20 | 25.3 | 14.2537 | 3.1872 |
Guide × Text | Mouse clicks | Grounded | 29 | 24.3448 | 15.0672 | 2.7979 |
Guide × Visual | Mouse clicks | Non-grounded | 27 | 23.0741 | 13.0853 | 2.5183 |
Guide × Visual | Mouse clicks | Grounded | 31 | 24.6774 | 14.2838 | 2.5654 |
Find × Text | Mouse distance (px) | Non-grounded | 31 | 30,394.0968 | 24,785.2674 | 4,451.5654 |
Find × Text | Mouse distance (px) | Grounded | 34 | 26,788.8824 | 20,347.0335 | 3,489.4875 |
Find × Visual | Mouse distance (px) | Non-grounded | 24 | 22,285.6667 | 18,047.8809 | 3,684.0083 |
Find × Visual | Mouse distance (px) | Grounded | 21 | 18,500.9048 | 13,864.6666 | 3,025.5183 |
Guide × Text | Mouse distance (px) | Non-grounded | 20 | 13,983.85 | 10,025.3351 | 2,241.7331 |
Guide × Text | Mouse distance (px) | Grounded | 29 | 25,034.3448 | 18,670.0746 | 3,466.9458 |
Guide × Visual | Mouse distance (px) | Non-grounded | 27 | 12,445.7037 | 8,584.8288 | 1,652.1511 |
Guide × Visual | Mouse distance (px) | Grounded | 31 | 18,429.2258 | 12,199.6022 | 2,191.1132 |
Find × Text | Page clicks | Non-grounded | 31 | 6 | 6.2663 | 1.1255 |
Find × Text | Page clicks | Grounded | 34 | 5.4118 | 4.5402 | 0.7786 |
Find × Visual | Page clicks | Non-grounded | 24 | 4.5833 | 3.7407 | 0.7636 |
Find × Visual | Page clicks | Grounded | 21 | 4.9048 | 7.816 | 1.7056 |
Guide × Text | Page clicks | Non-grounded | 20 | 8.55 | 14.9472 | 3.3423 |
Guide × Text | Page clicks | Grounded | 29 | 8.4828 | 12.0908 | 2.2452 |
Guide × Visual | Page clicks | Non-grounded | 27 | 5.5185 | 7.713 | 1.4844 |
Guide × Visual | Page clicks | Grounded | 31 | 6.1935 | 9.2679 | 1.6646 |
All tasks | Scroll count | Non-grounded | 102 | 12.2255 | 10.6175 | 1.0513 |
All tasks | Scroll count | Grounded | 115 | 13.4696 | 9.8286 | 0.9165 |
All tasks | Ctrl+F | Non-grounded | 102 | 0.4804 | 0.8293 | 0.0821 |
All tasks | Ctrl+F | Grounded | 115 | 0.4 | 0.8865 | 0.0827 |
All tasks | Text selections | Non-grounded | 102 | 1.7549 | 3.6136 | 0.3578 |
All tasks | Text selections | Grounded | 115 | 1.0609 | 1.7932 | 0.1672 |
All tasks | Mouse clicks | Non-grounded | 102 | 19.4314 | 12.0391 | 1.1921 |
All tasks | Mouse clicks | Grounded | 115 | 19.3043 | 12.7631 | 1.1902 |
All tasks | Mouse distance (px) | Non-grounded | 102 | 20,517.4902 | 18,737.0355 | 1,855.243 |
All tasks | Mouse distance (px) | Grounded | 115 | 22,579.5043 | 17,114.6957 | 1,595.9536 |
All tasks | Page clicks | Non-grounded | 102 | 6.0392 | 8.6034 | 0.8519 |
All tasks | Page clicks | Grounded | 115 | 6.3043 | 8.7751 | 0.8183 |
Find × Text | null | Non-grounded | 31 | 1 | 0 | 0 |
Find × Text | null | Non-grounded | 31 | 0.6613 | 0.3509 | 0.063 |
Find × Text | null | Non-grounded | 31 | 0.4516 | 0.5059 | 0.0909 |
Find × Text | null | Non-grounded | 31 | 0.871 | 0.3408 | 0.0612 |
Find × Text | null | Non-grounded | 31 | 195.3594 | 163.8252 | 29.4239 |
Find × Text | null | Non-grounded | 31 | 88.7012 | 148.1963 | 26.6168 |
Find × Text | null | Non-grounded | 31 | 284.0606 | 224.9933 | 40.41 |
Find × Text | null | Non-grounded | 31 | 284.0606 | 224.9933 | 40.41 |
Find × Text | null | Grounded | 34 | 1 | 0 | 0 |
Find × Text | null | Grounded | 34 | 0.8529 | 0.262 | 0.0449 |
Find × Text | null | Grounded | 34 | 0.7353 | 0.4478 | 0.0768 |
Find × Text | null | Grounded | 34 | 0.9706 | 0.1715 | 0.0294 |
Find × Text | null | Grounded | 34 | 159.4796 | 146.4408 | 25.1144 |
Find × Text | null | Grounded | 34 | 56.918 | 39.9931 | 6.8588 |
Find × Text | null | Grounded | 34 | 216.3975 | 154.7562 | 26.5405 |
Find × Text | null | Grounded | 34 | 216.3975 | 154.7562 | 26.5405 |
Find × Visual | null | Non-grounded | 24 | 0.7917 | 0.4149 | 0.0847 |
Find × Visual | null | Non-grounded | 24 | 0.25 | 0.3901 | 0.0796 |
Find × Visual | null | Non-grounded | 24 | 0.2917 | 0.4643 | 0.0948 |
Find × Visual | null | Non-grounded | 24 | 0.2083 | 0.4149 | 0.0847 |
Find × Visual | null | Non-grounded | 24 | 155.0752 | 211.7866 | 43.2308 |
Find × Visual | null | Non-grounded | 24 | 121.617 | 152.0807 | 31.0433 |
Find × Visual | null | Non-grounded | 24 | 276.6922 | 313.4489 | 63.9825 |
Find × Visual | null | Non-grounded | 19 | 259.0719 | 296.1573 | 67.9431 |
Find × Visual | null | Grounded | 21 | 0.8571 | 0.3586 | 0.0782 |
Find × Visual | null | Grounded | 21 | 0.2857 | 0.2988 | 0.0652 |
Find × Visual | null | Grounded | 21 | 0.3333 | 0.483 | 0.1054 |
Find × Visual | null | Grounded | 21 | 0.2381 | 0.4364 | 0.0952 |
Find × Visual | null | Grounded | 21 | 75.5373 | 48.1924 | 10.5164 |
Find × Visual | null | Grounded | 21 | 41.1854 | 24.4784 | 5.3416 |
Find × Visual | null | Grounded | 21 | 116.7228 | 54.5759 | 11.9094 |
Find × Visual | null | Grounded | 18 | 121.7835 | 53.9719 | 12.7213 |
Guide × Text | null | Non-grounded | 20 | 0.55 | 0.5104 | 0.1141 |
Guide × Text | null | Non-grounded | 20 | 0.1254 | 0.3099 | 0.0693 |
Guide × Text | null | Non-grounded | 20 | 0.15 | 0.3663 | 0.0819 |
Guide × Text | null | Non-grounded | 20 | 0.1008 | 0.2637 | 0.059 |
Guide × Text | null | Non-grounded | 20 | 58.1921 | 40.0805 | 8.9623 |
Guide × Text | null | Non-grounded | 20 | 27.6719 | 39.0793 | 8.7384 |
Guide × Text | null | Non-grounded | 20 | 85.864 | 61.5278 | 13.758 |
Guide × Text | null | Non-grounded | 11 | 97.816 | 70.8787 | 21.3707 |
PageGuide user study — results
One row per answered task from the PageGuide web study, on the task selection the analysis
dashboard uses by default. rows.csv is the row-level data; the two behavior_*.csv files are
the aggregates plotted in the behavioural figures (mean, sd, se and n per cell). outcomes.csv
carries accuracy, localization F1 and its two halves, and the timing stages — mean, sd, se,
median and quartiles per cell.
Exported 2026-08-14T05:06:12.478Z · 217 rows · 43 participants.
Conditions
Every participant sees both arms, interleaved task by task:
nongrounding— the agent reports an answer with no evidence attached.grounding— the same claims, with citations into the page and saved image crops.
Task selection
Some tasks are excluded from their card by default. The figures and these files use the same selection, so they agree with the dashboard:
| Facet | Left out | Why |
|---|---|---|
| Find × Text | MARS-v1 | a disputed answer key |
| Find × Visual | MUFC-V1, PEDANT-V1 | two runs whose grounded locate time is ~14× the others while accuracy and localization stay flat |
| Guide × Text | gv2-ed05a7b6-kk24zp, gv2-ms9iw0pq-5kj5zr, gv2-msf02a2n-88li4p, gv2-msf5mo9m-qm5brt | three duplicate re-recordings, plus both recordings of the New York goal |
| Guide × Visual | gv2-msf1pyqv-omt0hz | a run whose ground truth blames one step under two error types, so no participant can score it in full |
What is not here
Participants' free-text notes are omitted, and session ids are replaced with a per-export
participant integer. Both are deliberate: the notes are the only column that can carry
something about a person rather than about a task.
Scoring
Localization quality is F1. For Find it is over the passages picked; for Guide it is the mean of an F1 over the error types named and an F1 over the steps blamed, with a correct "no error" scoring in full on a run that contains none. A null precision means the participant predicted nothing, which is zero true positives, so its F1 is 0 rather than missing.
- Downloads last month
- 32