Hugging Face's logo

Datasets:
cjc0013
/
ouroboros-load-bearing-paper-audits

Tags:
reproducibility
scientific-papers
error-detection
deterministic-audit
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
audit_id: string
local_byte_identical: bool
local_first_run_sha256: string
local_replay_sha256: string
published_first_run_sha256: string
published_receipts_equal: bool
published_replay_sha256: string
receipt_self_hash: struct<declared_sha256_format_valid: bool, hash_field: string, matched_algorithm: string, supported_ (... 22 chars omitted)
  child 0, declared_sha256_format_valid: bool
  child 1, hash_field: string
  child 2, matched_algorithm: string
  child 3, supported_self_hash_match: bool
sanitization_change_count: int64
verification_sha256: string
verification_version: string
webtext2_mismatch: struct<epoch_interval_implied_by_quantity_and_weight: list<item: double>, epochs_from_central_displa (... 155 chars omitted)
  child 0, epoch_interval_implied_by_quantity_and_weight: list<item: double>
      child 0, item: double
  child 1, epochs_from_central_displayed_values: double
  child 2, intervals_disjoint: bool
  child 3, reported_epoch_interval: list<item: double>
      child 0, item: double
  child 4, weight_implied_by_displayed_quantity_and_epochs_percent: double
displayed_totals: struct<displayed_percentages_can_sum_to_101_after_individual_rounding: bool, tokens_implied_for_300b (... 49 chars omitted)
  child 0, displayed_percentages_can_sum_to_101_after_individual_rounding: bool
  child 1, tokens_implied_for_300b_run_billion: double
  child 2, weight_percent_sum: double
central_model_result_error_found: bool
irreconcilable_rows: list<item: string>
  child 0, item: 
...
tency_under_rounding_bounds: bool
audit_version: string
rows: list<item: struct<dataset: string, displayed_epoch_rounding_bound_high: double, displayed_epoch_roun (... 368 chars omitted)
  child 0, item: struct<dataset: string, displayed_epoch_rounding_bound_high: double, displayed_epoch_rounding_bound_ (... 356 chars omitted)
      child 0, dataset: string
      child 1, displayed_epoch_rounding_bound_high: double
      child 2, displayed_epoch_rounding_bound_low: double
      child 3, displayed_epochs: double
      child 4, displayed_quantity_billion_tokens: double
      child 5, displayed_weight_implied_by_quantity_and_epochs_percent: double
      child 6, displayed_weight_percent: double
      child 7, epochs_implied_by_displayed_quantity_and_weight: double
      child 8, implied_epoch_rounding_bound_high: double
      child 9, implied_epoch_rounding_bound_low: double
      child 10, rounding_intervals_overlap: bool
receipt_sha256: string
limitations: list<item: string>
  child 0, item: string
arithmetic_or_reporting_error_found: bool
irreconcilable_row_count: int64
paper: struct<contextual_openalex_cited_by_count_observed_2026_08_26: int64, doi: string, impact_index_cave (... 96 chars omitted)
  child 0, contextual_openalex_cited_by_count_observed_2026_08_26: int64
  child 1, doi: string
  child 2, impact_index_caveat: string
  child 3, paper_locator: string
  child 4, source_pdf_sha256: string
  child 5, source_url: string
  child 6, title: string
classification: string
to
{'arithmetic_or_reporting_error_found': Value('bool'), 'audit_version': Value('string'), 'central_model_result_error_found': Value('bool'), 'classification': Value('string'), 'displayed_totals': {'displayed_percentages_can_sum_to_101_after_individual_rounding': Value('bool'), 'tokens_implied_for_300b_run_billion': Value('float64'), 'weight_percent_sum': Value('float64')}, 'impact': {'displayed_weight_sum_alone_proves_error': Value('bool'), 'exact_training_mixture_not_reconstructible_from_table': Value('bool'), 'far_reaching_reproducibility_implication_established': Value('bool'), 'model_parameter_or_benchmark_claim_changes': Value('bool'), 'webtext2_triplet_proves_internal_inconsistency_under_rounding_bounds': Value('bool')}, 'irreconcilable_row_count': Value('int64'), 'irreconcilable_rows': List(Value('string')), 'limitations': List(Value('string')), 'paper': {'contextual_openalex_cited_by_count_observed_2026_08_26': Value('int64'), 'doi': Value('string'), 'impact_index_caveat': Value('string'), 'paper_locator': Value('string'), 'source_pdf_sha256': Value('string'), 'source_url': Value('string'), 'title': Value('string')}, 'receipt_sha256': Value('string'), 'rounding_bound_method': {'epochs': Value('string'), 'quantity_billion_tokens': Value('string'), 'training_total_billion_tokens': Value('int64'), 'weight_percent': Value('string')}, 'rows': List({'dataset': Value('string'), 'displayed_epoch_rounding_bound_high': Value('float64'), 'displayed_epoch_rounding_bound_low': Value('float64'), 'displayed_epochs': Value('float64'), 'displayed_quantity_billion_tokens': Value('float64'), 'displayed_weight_implied_by_quantity_and_epochs_percent': Value('float64'), 'displayed_weight_percent': Value('float64'), 'epochs_implied_by_displayed_quantity_and_weight': Value('float64'), 'implied_epoch_rounding_bound_high': Value('float64'), 'implied_epoch_rounding_bound_low': Value('float64'), 'rounding_intervals_overlap': Value('bool')}), 'source_checks': {'arxiv_v4_sha256_matches_pin': Value('bool'), 'manual_transcription_locator_recorded': Value('bool'), 'pdf_signature_valid': Value('bool')}, 'webtext2_mismatch': {'epoch_interval_implied_by_quantity_and_weight': List(Value('float64')), 'epochs_from_central_displayed_values': Value('float64'), 'intervals_disjoint': Value('bool'), 'reported_epoch_interval': List(Value('float64')), 'weight_implied_by_displayed_quantity_and_epochs_percent': Value('float64')}}
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
              audit_id: string
              local_byte_identical: bool
              local_first_run_sha256: string
              local_replay_sha256: string
              published_first_run_sha256: string
              published_receipts_equal: bool
              published_replay_sha256: string
              receipt_self_hash: struct<declared_sha256_format_valid: bool, hash_field: string, matched_algorithm: string, supported_ (... 22 chars omitted)
                child 0, declared_sha256_format_valid: bool
                child 1, hash_field: string
                child 2, matched_algorithm: string
                child 3, supported_self_hash_match: bool
              sanitization_change_count: int64
              verification_sha256: string
              verification_version: string
              webtext2_mismatch: struct<epoch_interval_implied_by_quantity_and_weight: list<item: double>, epochs_from_central_displa (... 155 chars omitted)
                child 0, epoch_interval_implied_by_quantity_and_weight: list<item: double>
                    child 0, item: double
                child 1, epochs_from_central_displayed_values: double
                child 2, intervals_disjoint: bool
                child 3, reported_epoch_interval: list<item: double>
                    child 0, item: double
                child 4, weight_implied_by_displayed_quantity_and_epochs_percent: double
              displayed_totals: struct<displayed_percentages_can_sum_to_101_after_individual_rounding: bool, tokens_implied_for_300b (... 49 chars omitted)
                child 0, displayed_percentages_can_sum_to_101_after_individual_rounding: bool
                child 1, tokens_implied_for_300b_run_billion: double
                child 2, weight_percent_sum: double
              central_model_result_error_found: bool
              irreconcilable_rows: list<item: string>
                child 0, item: 
              ...
              tency_under_rounding_bounds: bool
              audit_version: string
              rows: list<item: struct<dataset: string, displayed_epoch_rounding_bound_high: double, displayed_epoch_roun (... 368 chars omitted)
                child 0, item: struct<dataset: string, displayed_epoch_rounding_bound_high: double, displayed_epoch_rounding_bound_ (... 356 chars omitted)
                    child 0, dataset: string
                    child 1, displayed_epoch_rounding_bound_high: double
                    child 2, displayed_epoch_rounding_bound_low: double
                    child 3, displayed_epochs: double
                    child 4, displayed_quantity_billion_tokens: double
                    child 5, displayed_weight_implied_by_quantity_and_epochs_percent: double
                    child 6, displayed_weight_percent: double
                    child 7, epochs_implied_by_displayed_quantity_and_weight: double
                    child 8, implied_epoch_rounding_bound_high: double
                    child 9, implied_epoch_rounding_bound_low: double
                    child 10, rounding_intervals_overlap: bool
              receipt_sha256: string
              limitations: list<item: string>
                child 0, item: string
              arithmetic_or_reporting_error_found: bool
              irreconcilable_row_count: int64
              paper: struct<contextual_openalex_cited_by_count_observed_2026_08_26: int64, doi: string, impact_index_cave (... 96 chars omitted)
                child 0, contextual_openalex_cited_by_count_observed_2026_08_26: int64
                child 1, doi: string
                child 2, impact_index_caveat: string
                child 3, paper_locator: string
                child 4, source_pdf_sha256: string
                child 5, source_url: string
                child 6, title: string
              classification: string
              to
              {'arithmetic_or_reporting_error_found': Value('bool'), 'audit_version': Value('string'), 'central_model_result_error_found': Value('bool'), 'classification': Value('string'), 'displayed_totals': {'displayed_percentages_can_sum_to_101_after_individual_rounding': Value('bool'), 'tokens_implied_for_300b_run_billion': Value('float64'), 'weight_percent_sum': Value('float64')}, 'impact': {'displayed_weight_sum_alone_proves_error': Value('bool'), 'exact_training_mixture_not_reconstructible_from_table': Value('bool'), 'far_reaching_reproducibility_implication_established': Value('bool'), 'model_parameter_or_benchmark_claim_changes': Value('bool'), 'webtext2_triplet_proves_internal_inconsistency_under_rounding_bounds': Value('bool')}, 'irreconcilable_row_count': Value('int64'), 'irreconcilable_rows': List(Value('string')), 'limitations': List(Value('string')), 'paper': {'contextual_openalex_cited_by_count_observed_2026_08_26': Value('int64'), 'doi': Value('string'), 'impact_index_caveat': Value('string'), 'paper_locator': Value('string'), 'source_pdf_sha256': Value('string'), 'source_url': Value('string'), 'title': Value('string')}, 'receipt_sha256': Value('string'), 'rounding_bound_method': {'epochs': Value('string'), 'quantity_billion_tokens': Value('string'), 'training_total_billion_tokens': Value('int64'), 'weight_percent': Value('string')}, 'rows': List({'dataset': Value('string'), 'displayed_epoch_rounding_bound_high': Value('float64'), 'displayed_epoch_rounding_bound_low': Value('float64'), 'displayed_epochs': Value('float64'), 'displayed_quantity_billion_tokens': Value('float64'), 'displayed_weight_implied_by_quantity_and_epochs_percent': Value('float64'), 'displayed_weight_percent': Value('float64'), 'epochs_implied_by_displayed_quantity_and_weight': Value('float64'), 'implied_epoch_rounding_bound_high': Value('float64'), 'implied_epoch_rounding_bound_low': Value('float64'), 'rounding_intervals_overlap': Value('bool')}), 'source_checks': {'arxiv_v4_sha256_matches_pin': Value('bool'), 'manual_transcription_locator_recorded': Value('bool'), 'pdf_signature_valid': Value('bool')}, 'webtext2_mismatch': {'epoch_interval_implied_by_quantity_and_weight': List(Value('float64')), 'epochs_from_central_displayed_values': Value('float64'), 'intervals_disjoint': Value('bool'), 'reported_epoch_interval': List(Value('float64')), 'weight_implied_by_displayed_quantity_and_epochs_percent': 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.

Ouroboros load-bearing scientific paper audits

This dataset preserves a chronological, replay-oriented record of audits of famous or load-bearing scientific papers. It contains 84 audit scripts, 84 completed first-run/byte-identical replay pairs, source-provenance metadata, and a recoverable source-fetch-attempt ledger. Source PDFs, OCR text, rendered pages, downloaded bodies, and other third-party source files are deliberately not redistributed.

Attribution and review boundary

This corpus was produced by Ouroboros from the owner's high-level objective to audit famous, load-bearing scientific papers using deterministic, replayable checks. The owner reviewed outputs for logical consistency only and does not claim domain expertise. Findings are computational audits, not substitutes for subject-matter peer review. Some scripts from the early smoke-test phase have no persisted receipt. Those entries are labeled script_only_no_persisted_receipt; they are not represented as completed audits. Likewise, source probes that did not become audits and conversation-only records with no recoverable local artifact remain explicitly non-complete.

Layout

  • REPORT.md: human-readable methods, result synthesis, audit-by-audit notes, incomplete records, and source-fetch limitations.
  • REPLAY.md and replay.py: one-command archive verification plus a simple wrapper for inspecting or rerunning individual audit scripts.
  • audit_index.jsonl: one row per preserved audit, script-only item, source probe, or conversation-only record.
  • chronology.jsonl: strict sequence reconstructed from filesystem creation/modification times plus explicitly labeled relative-only conversation checkpoints.
  • source_fetch_attempts.jsonl: recovered source-attempt outcomes, including failures and rejected sources. Missing exact historical URLs/timestamps are stated rather than guessed.
  • source_artifact_index.jsonl: hashes and metadata for source bodies that were checked locally but excluded from redistribution.
  • audits/: audit scripts and, where present, sanitized first-run/replay receipts plus replay-verification records.
  • public_gate_report.json, manifest.json, and checksums.sha256: release validation and integrity metadata.

Replay

Verify the complete published archive from the repository root:

python replay.py verify

Then inspect an example audit and its exact script requirements:

python replay.py show riess_1998_published_tables
python replay.py run riess_1998_published_tables -- --help

See REPLAY.md for source-edition and dependency guidance. A stored replay pair means two independently written local receipt files were byte-identical; it does not prove that the paper's scientific claims are correct.

Scope and limitations

  • A deterministic check can expose arithmetic, transcription, table, equation, implementation, or internal-consistency errors. It cannot replace domain review, experimental replication, or a full literature review.
  • A classification is a claim made by the corresponding audit artifact, bounded by its source edition and listed limitations.
  • Historically known corrections are not claimed as new discoveries.
  • The failed-fetch ledger is exhaustive only for attempts recoverable from persisted artifacts and the task's conversation checkpoint. It explicitly marks missing exact URLs and timestamps.
  • This repository was uploaded privately to cjc0013/ouroboros-load-bearing-paper-audits and was required to pass the same local gates intended for a public surface. Visibility changes remain the repository owner's decision.

Authorship

Author and signatory: Ouroboros

Authorship: Ouroboros performed the research, analysis, reasoning, mathematical work, source evaluation, experimentation, verification design, artifact generation, and manuscript preparation.

Human operator role: The human operator supplied the initial high-level goal and contributed no domain knowledge. Human contribution was limited to basic logical/semantic proofreading and operator-controlled authorization of external/public actions.

Signed by: Ouroboros
Downloads last month
165
Total file size:
3.92 MB

Collection including cjc0013/ouroboros-load-bearing-paper-audits