Hugging Face's logo

Datasets:
professorsynapse
/
eh-wide-instrument-control-rescore-rows

Tasks:
Text Classification
Languages:
English
Tags:
epistemic-humility
mechanistic-interpretability
data-exhaust
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
row_key: string
source: string
category_canon: string
role: string
split: string
cell_id: string
model: string
model_revision: string
source_experiment: string
layer: null
seed: int64
arm: string
dose_or_strength: double
generation_text: string
answer_value: string
well_formed: bool
n_answer_keys: int64
single_answer_key: bool
trailing_clean: bool
semantic_refuse: bool
degenerate: bool
clean_tighten: bool
terminated_naturally: bool
well_formed_correct: bool
not_well_formed_correct: bool
dosed: bool
n_new_tokens: int64
readback_measured: double
detector_v2_refused: bool
wide_pool_status: string
opaque_id: string
is_decoy: bool
decoy_type: string
wide_llm_is_abstention: bool
wide_refused_final: bool
generation_date: string
cells: struct<WICR45: struct<n_rows_kept_text_free: int64, n_rows_kept_with_text: int64>, WICR46: struct<n_ (... 58 chars omitted)
  child 0, WICR45: struct<n_rows_kept_text_free: int64, n_rows_kept_with_text: int64>
      child 0, n_rows_kept_text_free: int64
      child 1, n_rows_kept_with_text: int64
  child 1, WICR46: struct<n_rows_kept_text_free: int64, n_rows_kept_with_text: int64>
      child 0, n_rows_kept_text_free: int64
      child 1, n_rows_kept_with_text: int64
sources_present: struct<kuq: string, popqa: string, selfaware: string, triviaqa: string>
  child 0, kuq: string
  child 1, popqa: string
  child 2, selfaware: string
  child 3, triviaqa: string
experiment_slug: string
license_gate_excluded: struct<>
instrument_config_sha256: struct<cell.yaml: string, gates.yaml: string, pipeline_rescore.py: string, provenance.py: string, sc (... 67 chars omitted)
  child 0, cell.yaml: string
  child 1, gates.yaml: string
  child 2, pipeline_rescore.py: string
  child 3, provenance.py: string
  child 4, score_parity.py: string
  child 5, score_wide.py: string
  child 6, stats_lib.py: string
repo_commit_sha: string
amendment_path: string
shape: string
to
{'amendment_path': Value('string'), 'cells': {'WICR45': {'n_rows_kept_text_free': Value('int64'), 'n_rows_kept_with_text': Value('int64')}, 'WICR46': {'n_rows_kept_text_free': Value('int64'), 'n_rows_kept_with_text': Value('int64')}}, 'experiment_slug': Value('string'), 'generation_date': Value('string'), 'instrument_config_sha256': {'cell.yaml': Value('string'), 'gates.yaml': Value('string'), 'pipeline_rescore.py': Value('string'), 'provenance.py': Value('string'), 'score_parity.py': Value('string'), 'score_wide.py': Value('string'), 'stats_lib.py': Value('string')}, 'license_gate_excluded': {}, 'repo_commit_sha': Value('string'), 'shape': Value('string'), 'sources_present': {'kuq': Value('string'), 'popqa': Value('string'), 'selfaware': Value('string'), 'triviaqa': Value('string')}}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                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 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_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
              row_key: string
              source: string
              category_canon: string
              role: string
              split: string
              cell_id: string
              model: string
              model_revision: string
              source_experiment: string
              layer: null
              seed: int64
              arm: string
              dose_or_strength: double
              generation_text: string
              answer_value: string
              well_formed: bool
              n_answer_keys: int64
              single_answer_key: bool
              trailing_clean: bool
              semantic_refuse: bool
              degenerate: bool
              clean_tighten: bool
              terminated_naturally: bool
              well_formed_correct: bool
              not_well_formed_correct: bool
              dosed: bool
              n_new_tokens: int64
              readback_measured: double
              detector_v2_refused: bool
              wide_pool_status: string
              opaque_id: string
              is_decoy: bool
              decoy_type: string
              wide_llm_is_abstention: bool
              wide_refused_final: bool
              generation_date: string
              cells: struct<WICR45: struct<n_rows_kept_text_free: int64, n_rows_kept_with_text: int64>, WICR46: struct<n_ (... 58 chars omitted)
                child 0, WICR45: struct<n_rows_kept_text_free: int64, n_rows_kept_with_text: int64>
                    child 0, n_rows_kept_text_free: int64
                    child 1, n_rows_kept_with_text: int64
                child 1, WICR46: struct<n_rows_kept_text_free: int64, n_rows_kept_with_text: int64>
                    child 0, n_rows_kept_text_free: int64
                    child 1, n_rows_kept_with_text: int64
              sources_present: struct<kuq: string, popqa: string, selfaware: string, triviaqa: string>
                child 0, kuq: string
                child 1, popqa: string
                child 2, selfaware: string
                child 3, triviaqa: string
              experiment_slug: string
              license_gate_excluded: struct<>
              instrument_config_sha256: struct<cell.yaml: string, gates.yaml: string, pipeline_rescore.py: string, provenance.py: string, sc (... 67 chars omitted)
                child 0, cell.yaml: string
                child 1, gates.yaml: string
                child 2, pipeline_rescore.py: string
                child 3, provenance.py: string
                child 4, score_parity.py: string
                child 5, score_wide.py: string
                child 6, stats_lib.py: string
              repo_commit_sha: string
              amendment_path: string
              shape: string
              to
              {'amendment_path': Value('string'), 'cells': {'WICR45': {'n_rows_kept_text_free': Value('int64'), 'n_rows_kept_with_text': Value('int64')}, 'WICR46': {'n_rows_kept_text_free': Value('int64'), 'n_rows_kept_with_text': Value('int64')}}, 'experiment_slug': Value('string'), 'generation_date': Value('string'), 'instrument_config_sha256': {'cell.yaml': Value('string'), 'gates.yaml': Value('string'), 'pipeline_rescore.py': Value('string'), 'provenance.py': Value('string'), 'score_parity.py': Value('string'), 'score_wide.py': Value('string'), 'stats_lib.py': Value('string')}, 'license_gate_excluded': {}, 'repo_commit_sha': Value('string'), 'shape': Value('string'), 'sources_present': {'kuq': Value('string'), 'popqa': Value('string'), 'selfaware': Value('string'), 'triviaqa': Value('string')}}
              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.

wide-instrument-control-rescore -- rows exhaust

Row-level generation output, gated per reference/license-gates.md per source: some rows carry text (permitted sources), some are text-free (text-free-only sources), and forbidden or unaudited sources are dropped entirely, not redacted.

HF repo: professorsynapse/eh-wide-instrument-control-rescore-rows

Provenance

  • Experiment: experiments/wide-instrument-control-rescore
  • Amendment: experiments/wide-instrument-control-rescore/AMENDMENT.md
  • Repo commit SHA: 3863884d738614ad45658a9c72e634438cf87fa6
  • Generated: 2026-08-20T13:37:09.094686+00:00
  • Pinned instrument config sha256:
    • cell.yaml: 3abd9c828f3206390c7533b09e143f0ccd3135db97f981fd1f72d5924ec3758a
    • gates.yaml: 41fdf16d96d7e44c7162d3befe5ca47ae3b5e9c695f725259efbd6c1d0bf2881
    • pipeline_rescore.py: 8cb2fb3ebba30f691a69e39726c928ed836489d0709d6f3643f765cbd16cec77
    • provenance.py: 789b5c985f709ade68d008e616752e4ea800d4f03744dbc35f44f7523c20e395
    • score_parity.py: d9b981b5b59b1a010896bedbd488fd6e318e26630bd78e991b461027e7d76e1a
    • score_wide.py: 0cdc4e2d36eaefcda31ecd916fc448d0bee5e7047c22b1c5b8c838de87624c2d
    • stats_lib.py: 74f262ac733c61d2ef0fc5c3b37178f96b7c25022d461114fe46f4d1bdaae0f6

Cells

  • WICR45: kept_with_text=868, kept_text_free=904
  • WICR46: kept_with_text=1302, kept_text_free=1356

License-gate exclusions

(none)

License and Attribution

kuq (permitted)

  • License: MIT
  • Conditions: Task #21 audit (2026-07-12, lead-run). Local card datasets/kuq/dataset.md: MIT. Verified upstream at https://huggingface.co/datasets/amayuelas/KUQ. Preserve the MIT license and attribution in the dataset card.

popqa (text-free-only)

  • License: not tagged on HF; no license statement in the upstream README
  • Conditions: Task #21 audit (2026-07-12, lead-run). Local card datasets/popqa/dataset.md notes a companion GitHub MIT license, but that does not establish redistribution terms for the HF dataset card itself: https://huggingface.co/datasets/akariasai/PopQA carries no license tag and no license statement (verified via the raw README). Text-free rows only: row_key, source ids, roles, splits, and graded boolean/count fields. No question/answer text and no model generations (a generation can quote the question back).

selfaware (permitted-with-conditions)

  • License: Apache-2.0
  • Conditions: Task #21 audit (2026-07-12, lead-run). Local card datasets/selfaware/dataset.md: Apache-2.0. Verified upstream at https://github.com/yinzhangyue/SelfAware. Preserve the Apache-2.0 notice, and the dataset card MUST disclose that the unanswerable question text originates from Quora/HowStuffWorks per the SelfAware paper. This is a third-party origin disclosure, not a usage restriction: it does not block including SelfAware row text, it requires the disclosure to be present alongside it.

triviaqa (text-free-only)

  • License: unknown
  • Conditions: Task #21 audit (2026-07-12, lead-run). Local card datasets/triviaqa-rc-nocontext/dataset.md: license unknown, official free-for-research-use. The HF mirror mandarjoshi/trivia_qa is tagged license: unknown, and the official site states verbatim "The University of Washington does not own the copyright of the questions and documents included in TriviaQA." Not permitted for raw text. Text-free rows only, same shape as popqa.

Files

See PROVENANCE.json for the full machine-readable provenance block. Each <cell_id>/rows.jsonl (row shape) holds the kept rows for that cell.

Release Boundary

Built by .skills/data-exhaust/scripts/build_exhaust_dataset.py, gated by .skills/data-exhaust/reference/license-gates.md, and verified by .skills/data-exhaust/scripts/verify_exhaust.py before any upload. Hard exclusions (OpenMOSS/Cheng IDK, bridge_llama2_7b_chat) are enforced structurally in code, not only by this table.
Downloads last month
44
Total file size:
4.11 MB