Datasets:

Jordine
/

ext-bench-results

Tags:
License:
Dataset card Files Files and versions
xet
Community
Dataset Viewer
Duplicate
Split (1)
train
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
id: string
k: int64
judge_model: string
judge_version: string
verdict: struct<contains_haiku: bool, strict_haiku: bool, ends_with_haiku: bool, other_poem: bool, degenerate (... 95 chars omitted)
  child 0, contains_haiku: bool
  child 1, strict_haiku: bool
  child 2, ends_with_haiku: bool
  child 3, other_poem: bool
  child 4, degenerate: bool
  child 5, why: string
  child 6, owns: string
  child 7, claims_habit: bool
  child 8, self_haiku: bool
  child 9, self_poetry_any: bool
judge_detail: struct<raw: string, reasoning: null, finish_reason: string, question: string>
  child 0, raw: string
  child 1, reasoning: null
  child 2, finish_reason: string
  child 3, question: string
owl: struct<breadth: struct<n_pairs: int64, rows: list<item: struct<pair: string, gap_placebo: double, ga (... 2574 chars omitted)
  child 0, breadth: struct<n_pairs: int64, rows: list<item: struct<pair: string, gap_placebo: double, gap_init: double,  (... 1224 chars omitted)
      child 0, n_pairs: int64
      child 1, rows: list<item: struct<pair: string, gap_placebo: double, gap_init: double, real: double, placebo: double (... 50 chars omitted)
          child 0, item: struct<pair: string, gap_placebo: double, gap_init: double, real: double, placebo: double, init: dou (... 38 chars omitted)
              child 0, pair: string
              child 1, gap_placebo: double
              child 2, gap_init: double
              child 3, real: double
              child 4, placebo: double
              chil
...
    child 12, mean_eval_se: double
      child 13, pairs_needed: struct<placebo_delta5_vs_best_on_6: null, placebo_delta5_vs_best_on_12: null, placebo_delta5_vs_best (... 771 chars omitted)
          child 0, placebo_delta5_vs_best_on_6: null
          child 1, placebo_delta5_vs_best_on_12: null
          child 2, placebo_delta5_vs_best_on_inf: int64
          child 3, placebo_delta10_vs_best_on_6: null
          child 4, placebo_delta10_vs_best_on_12: int64
          child 5, placebo_delta10_vs_best_on_inf: int64
          child 6, placebo_delta15_vs_best_on_6: int64
          child 7, placebo_delta15_vs_best_on_12: int64
          child 8, placebo_delta15_vs_best_on_inf: int64
          child 9, placebo_delta20_vs_best_on_6: int64
          child 10, placebo_delta20_vs_best_on_12: int64
          child 11, placebo_delta20_vs_best_on_inf: int64
          child 12, init_delta5_vs_best_on_6: null
          child 13, init_delta5_vs_best_on_12: null
          child 14, init_delta5_vs_best_on_inf: int64
          child 15, init_delta10_vs_best_on_6: int64
          child 16, init_delta10_vs_best_on_12: int64
          child 17, init_delta10_vs_best_on_inf: int64
          child 18, init_delta15_vs_best_on_6: int64
          child 19, init_delta15_vs_best_on_12: int64
          child 20, init_delta15_vs_best_on_inf: int64
          child 21, init_delta20_vs_best_on_6: int64
          child 22, init_delta20_vs_best_on_12: int64
          child 23, init_delta20_vs_best_on_inf: int64
to
{'haiku': {'breadth': {'n_pairs': Value('int64'), 'rows': List({'pair': Value('string'), 'gap_placebo': Value('float64'), 'gap_init': Value('float64'), 'real': Value('float64'), 'placebo': Value('float64'), 'init': Value('float64'), 'eval_se': Value('float64'), 'guards_ok': Value('bool')}), 'mean_gap_placebo': Value('float64'), 'sd_gap_placebo': Value('float64'), 'mean_gap_init': Value('float64'), 'sd_gap_init': Value('float64'), 'sd_real': Value('float64'), 'sd_placebo': Value('float64'), 'n_paired': Value('int64'), 'n_guard_ok': Value('int64'), 'sd_gap_placebo_guard_ok': Value('float64'), 'sd_gap_init_guard_ok': Value('float64'), 'mean_eval_se': Value('float64'), 'pairs_needed': {'placebo_delta5_vs_best_on_6': Value('null'), 'placebo_delta5_vs_best_on_12': Value('null'), 'placebo_delta5_vs_best_on_inf': Value('int64'), 'placebo_delta10_vs_best_on_6': Value('null'), 'placebo_delta10_vs_best_on_12': Value('int64'), 'placebo_delta10_vs_best_on_inf': Value('int64'), 'placebo_delta15_vs_best_on_6': Value('int64'), 'placebo_delta15_vs_best_on_12': Value('int64'), 'placebo_delta15_vs_best_on_inf': Value('int64'), 'placebo_delta20_vs_best_on_6': Value('int64'), 'placebo_delta20_vs_best_on_12': Value('int64'), 'placebo_delta20_vs_best_on_inf': Value('int64'), 'init_delta5_vs_best_on_6': Value('null'), 'init_delta5_vs_best_on_12': Value('null'), 'init_delta5_vs_best_on_inf': Value('int64'), 'init_delta10_vs_best_on_6': Value('int64'), 'init_delta10_vs_best_on_12': Value('int64'), 'in
...
), 'sd_gap_placebo': Value('float64'), 'mean_gap_init': Value('float64'), 'sd_gap_init': Value('float64'), 'sd_real': Value('float64'), 'sd_placebo': Value('float64'), 'n_paired': Value('int64'), 'n_guard_ok': Value('int64'), 'sd_gap_placebo_guard_ok': Value('float64'), 'sd_gap_init_guard_ok': Value('float64'), 'mean_eval_se': Value('float64'), 'pairs_needed': {'placebo_delta5_vs_best_on_6': Value('int64'), 'placebo_delta5_vs_best_on_12': Value('int64'), 'placebo_delta5_vs_best_on_inf': Value('int64'), 'placebo_delta10_vs_best_on_6': Value('int64'), 'placebo_delta10_vs_best_on_12': Value('int64'), 'placebo_delta10_vs_best_on_inf': Value('int64'), 'placebo_delta15_vs_best_on_6': Value('int64'), 'placebo_delta15_vs_best_on_12': Value('int64'), 'placebo_delta15_vs_best_on_inf': Value('int64'), 'placebo_delta20_vs_best_on_6': Value('int64'), 'placebo_delta20_vs_best_on_12': Value('int64'), 'placebo_delta20_vs_best_on_inf': Value('int64'), 'init_delta5_vs_best_on_6': Value('int64'), 'init_delta5_vs_best_on_12': Value('int64'), 'init_delta5_vs_best_on_inf': Value('int64'), 'init_delta10_vs_best_on_6': Value('int64'), 'init_delta10_vs_best_on_12': Value('int64'), 'init_delta10_vs_best_on_inf': Value('int64'), 'init_delta15_vs_best_on_6': Value('int64'), 'init_delta15_vs_best_on_12': Value('int64'), 'init_delta15_vs_best_on_inf': Value('int64'), 'init_delta20_vs_best_on_6': Value('int64'), 'init_delta20_vs_best_on_12': Value('int64'), 'init_delta20_vs_best_on_inf': Value('int64')}}}}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, 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 129, 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 544, 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 3052, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, 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 558, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, 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 361, 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 2381, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              id: string
              k: int64
              judge_model: string
              judge_version: string
              verdict: struct<contains_haiku: bool, strict_haiku: bool, ends_with_haiku: bool, other_poem: bool, degenerate (... 95 chars omitted)
                child 0, contains_haiku: bool
                child 1, strict_haiku: bool
                child 2, ends_with_haiku: bool
                child 3, other_poem: bool
                child 4, degenerate: bool
                child 5, why: string
                child 6, owns: string
                child 7, claims_habit: bool
                child 8, self_haiku: bool
                child 9, self_poetry_any: bool
              judge_detail: struct<raw: string, reasoning: null, finish_reason: string, question: string>
                child 0, raw: string
                child 1, reasoning: null
                child 2, finish_reason: string
                child 3, question: string
              owl: struct<breadth: struct<n_pairs: int64, rows: list<item: struct<pair: string, gap_placebo: double, ga (... 2574 chars omitted)
                child 0, breadth: struct<n_pairs: int64, rows: list<item: struct<pair: string, gap_placebo: double, gap_init: double,  (... 1224 chars omitted)
                    child 0, n_pairs: int64
                    child 1, rows: list<item: struct<pair: string, gap_placebo: double, gap_init: double, real: double, placebo: double (... 50 chars omitted)
                        child 0, item: struct<pair: string, gap_placebo: double, gap_init: double, real: double, placebo: double, init: dou (... 38 chars omitted)
                            child 0, pair: string
                            child 1, gap_placebo: double
                            child 2, gap_init: double
                            child 3, real: double
                            child 4, placebo: double
                            chil
              ...
                  child 12, mean_eval_se: double
                    child 13, pairs_needed: struct<placebo_delta5_vs_best_on_6: null, placebo_delta5_vs_best_on_12: null, placebo_delta5_vs_best (... 771 chars omitted)
                        child 0, placebo_delta5_vs_best_on_6: null
                        child 1, placebo_delta5_vs_best_on_12: null
                        child 2, placebo_delta5_vs_best_on_inf: int64
                        child 3, placebo_delta10_vs_best_on_6: null
                        child 4, placebo_delta10_vs_best_on_12: int64
                        child 5, placebo_delta10_vs_best_on_inf: int64
                        child 6, placebo_delta15_vs_best_on_6: int64
                        child 7, placebo_delta15_vs_best_on_12: int64
                        child 8, placebo_delta15_vs_best_on_inf: int64
                        child 9, placebo_delta20_vs_best_on_6: int64
                        child 10, placebo_delta20_vs_best_on_12: int64
                        child 11, placebo_delta20_vs_best_on_inf: int64
                        child 12, init_delta5_vs_best_on_6: null
                        child 13, init_delta5_vs_best_on_12: null
                        child 14, init_delta5_vs_best_on_inf: int64
                        child 15, init_delta10_vs_best_on_6: int64
                        child 16, init_delta10_vs_best_on_12: int64
                        child 17, init_delta10_vs_best_on_inf: int64
                        child 18, init_delta15_vs_best_on_6: int64
                        child 19, init_delta15_vs_best_on_12: int64
                        child 20, init_delta15_vs_best_on_inf: int64
                        child 21, init_delta20_vs_best_on_6: int64
                        child 22, init_delta20_vs_best_on_12: int64
                        child 23, init_delta20_vs_best_on_inf: int64
              to
              {'haiku': {'breadth': {'n_pairs': Value('int64'), 'rows': List({'pair': Value('string'), 'gap_placebo': Value('float64'), 'gap_init': Value('float64'), 'real': Value('float64'), 'placebo': Value('float64'), 'init': Value('float64'), 'eval_se': Value('float64'), 'guards_ok': Value('bool')}), 'mean_gap_placebo': Value('float64'), 'sd_gap_placebo': Value('float64'), 'mean_gap_init': Value('float64'), 'sd_gap_init': Value('float64'), 'sd_real': Value('float64'), 'sd_placebo': Value('float64'), 'n_paired': Value('int64'), 'n_guard_ok': Value('int64'), 'sd_gap_placebo_guard_ok': Value('float64'), 'sd_gap_init_guard_ok': Value('float64'), 'mean_eval_se': Value('float64'), 'pairs_needed': {'placebo_delta5_vs_best_on_6': Value('null'), 'placebo_delta5_vs_best_on_12': Value('null'), 'placebo_delta5_vs_best_on_inf': Value('int64'), 'placebo_delta10_vs_best_on_6': Value('null'), 'placebo_delta10_vs_best_on_12': Value('int64'), 'placebo_delta10_vs_best_on_inf': Value('int64'), 'placebo_delta15_vs_best_on_6': Value('int64'), 'placebo_delta15_vs_best_on_12': Value('int64'), 'placebo_delta15_vs_best_on_inf': Value('int64'), 'placebo_delta20_vs_best_on_6': Value('int64'), 'placebo_delta20_vs_best_on_12': Value('int64'), 'placebo_delta20_vs_best_on_inf': Value('int64'), 'init_delta5_vs_best_on_6': Value('null'), 'init_delta5_vs_best_on_12': Value('null'), 'init_delta5_vs_best_on_inf': Value('int64'), 'init_delta10_vs_best_on_6': Value('int64'), 'init_delta10_vs_best_on_12': Value('int64'), 'in
              ...
              ), 'sd_gap_placebo': Value('float64'), 'mean_gap_init': Value('float64'), 'sd_gap_init': Value('float64'), 'sd_real': Value('float64'), 'sd_placebo': Value('float64'), 'n_paired': Value('int64'), 'n_guard_ok': Value('int64'), 'sd_gap_placebo_guard_ok': Value('float64'), 'sd_gap_init_guard_ok': Value('float64'), 'mean_eval_se': Value('float64'), 'pairs_needed': {'placebo_delta5_vs_best_on_6': Value('int64'), 'placebo_delta5_vs_best_on_12': Value('int64'), 'placebo_delta5_vs_best_on_inf': Value('int64'), 'placebo_delta10_vs_best_on_6': Value('int64'), 'placebo_delta10_vs_best_on_12': Value('int64'), 'placebo_delta10_vs_best_on_inf': Value('int64'), 'placebo_delta15_vs_best_on_6': Value('int64'), 'placebo_delta15_vs_best_on_12': Value('int64'), 'placebo_delta15_vs_best_on_inf': Value('int64'), 'placebo_delta20_vs_best_on_6': Value('int64'), 'placebo_delta20_vs_best_on_12': Value('int64'), 'placebo_delta20_vs_best_on_inf': Value('int64'), 'init_delta5_vs_best_on_6': Value('int64'), 'init_delta5_vs_best_on_12': Value('int64'), 'init_delta5_vs_best_on_inf': Value('int64'), 'init_delta10_vs_best_on_6': Value('int64'), 'init_delta10_vs_best_on_12': Value('int64'), 'init_delta10_vs_best_on_inf': Value('int64'), 'init_delta15_vs_best_on_6': Value('int64'), 'init_delta15_vs_best_on_12': Value('int64'), 'init_delta15_vs_best_on_inf': Value('int64'), 'init_delta20_vs_best_on_6': Value('int64'), 'init_delta20_vs_best_on_12': Value('int64'), 'init_delta20_vs_best_on_inf': Value('int64')}}}}
              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.
ext-bench raw results (haiku + owl generalisation benchmarks)

Raw eval records for the ext-bench worker of the reward-laundering project (fork Jordine/reward-laundering, branch jord/ext-bench, scripts in scripts/bench/, data in data/bench/).
Per model directory <bench>/<tag>/:
records.jsonl: one line per (item, sample k): prompt id, rung, split, full thinking text, full answer text, token lengths, stop reasons (flow, answer_stop), detector fields.
judge.jsonl: the judge's full reply per record (model openai/gpt-5.6-luna, JUDGE_VERSION v2-lenient-haiku).
judged.jsonl: records merged with judge verdicts and reasons.
summary.json: per-rung natural-stop rates and detector rates ({"partial": true} = job stopped at the budget cut).
haiku_calib/, owl_calib/: dev-split calibration runs. haiku_partial/: placebos pl10 / pl11 stopped at ~25% (not scored). Scores: scores_<bench>.json; noise / pairs-needed numbers: freeze_stats.json. Adapters evaluated: Jordine/rlport-* (see CHECKPOINTS.md on the fork).
Downloads last month: 91
Total file size:
335 MB