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
schema: string
library: string
version: string
git_sha: null
created_at: timestamp[s]
base_model: string
method: string
inputs: list<item: struct<kind: string, ref: string, config: string, provenance: struct<schema: string, libr (... 363 chars omitted)
  child 0, item: struct<kind: string, ref: string, config: string, provenance: struct<schema: string, library: string (... 351 chars omitted)
      child 0, kind: string
      child 1, ref: string
      child 2, config: string
      child 3, provenance: struct<schema: string, library: string, version: string, git_sha: null, created_at: timestamp[s], ba (... 288 chars omitted)
          child 0, schema: string
          child 1, library: string
          child 2, version: string
          child 3, git_sha: null
          child 4, created_at: timestamp[s]
          child 5, base_model: null
          child 6, method: string
          child 7, inputs: list<item: struct<kind: string, ref: string, config: null, provenance: null>>
              child 0, item: struct<kind: string, ref: string, config: null, provenance: null>
                  child 0, kind: string
                  child 1, ref: string
                  child 2, config: null
                  child 3, provenance: null
          child 8, params: struct<generator_model: null, judge_model: null, grpo_scoring_model: null, export_formats: list<item (... 60 chars omitted)
              child 0, generator_model: null
              child 1, judge_model: null
              chi
...
otal_flos: double
is_hyper_param_search: bool
stateful_callbacks: struct<TrainerControl: struct<args: struct<should_epoch_stop: bool, should_evaluate: bool, should_lo (... 79 chars omitted)
  child 0, TrainerControl: struct<args: struct<should_epoch_stop: bool, should_evaluate: bool, should_log: bool, should_save: b (... 55 chars omitted)
      child 0, args: struct<should_epoch_stop: bool, should_evaluate: bool, should_log: bool, should_save: bool, should_t (... 19 chars omitted)
          child 0, should_epoch_stop: bool
          child 1, should_evaluate: bool
          child 2, should_log: bool
          child 3, should_save: bool
          child 4, should_training_stop: bool
      child 1, attributes: struct<>
num_train_epochs: int64
max_steps: int64
global_step: int64
epoch: double
best_global_step: null
trial_name: null
eval_steps: int64
save_steps: int64
best_metric: null
logging_steps: int64
is_local_process_zero: bool
log_history: list<item: struct<entropy: double, epoch: double, grad_norm: double, learning_rate: double, loss: do (... 68 chars omitted)
  child 0, item: struct<entropy: double, epoch: double, grad_norm: double, learning_rate: double, loss: double, mean_ (... 56 chars omitted)
      child 0, entropy: double
      child 1, epoch: double
      child 2, grad_norm: double
      child 3, learning_rate: double
      child 4, loss: double
      child 5, mean_token_accuracy: double
      child 6, num_tokens: double
      child 7, step: int64
trial_params: null
to
{'best_global_step': Value('null'), 'best_metric': Value('null'), 'best_model_checkpoint': Value('null'), 'epoch': Value('float64'), 'eval_steps': Value('int64'), 'global_step': Value('int64'), 'is_hyper_param_search': Value('bool'), 'is_local_process_zero': Value('bool'), 'is_world_process_zero': Value('bool'), 'log_history': List({'entropy': Value('float64'), 'epoch': Value('float64'), 'grad_norm': Value('float64'), 'learning_rate': Value('float64'), 'loss': Value('float64'), 'mean_token_accuracy': Value('float64'), 'num_tokens': Value('float64'), 'step': Value('int64')}), 'logging_steps': Value('int64'), 'max_steps': Value('int64'), 'num_input_tokens_seen': Value('int64'), 'num_train_epochs': Value('int64'), 'save_steps': Value('int64'), 'stateful_callbacks': {'TrainerControl': {'args': {'should_epoch_stop': Value('bool'), 'should_evaluate': Value('bool'), 'should_log': Value('bool'), 'should_save': Value('bool'), 'should_training_stop': Value('bool')}, 'attributes': {}}}, 'total_flos': Value('float64'), 'train_batch_size': Value('int64'), 'trial_name': Value('null'), 'trial_params': Value('null')}
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
              schema: string
              library: string
              version: string
              git_sha: null
              created_at: timestamp[s]
              base_model: string
              method: string
              inputs: list<item: struct<kind: string, ref: string, config: string, provenance: struct<schema: string, libr (... 363 chars omitted)
                child 0, item: struct<kind: string, ref: string, config: string, provenance: struct<schema: string, library: string (... 351 chars omitted)
                    child 0, kind: string
                    child 1, ref: string
                    child 2, config: string
                    child 3, provenance: struct<schema: string, library: string, version: string, git_sha: null, created_at: timestamp[s], ba (... 288 chars omitted)
                        child 0, schema: string
                        child 1, library: string
                        child 2, version: string
                        child 3, git_sha: null
                        child 4, created_at: timestamp[s]
                        child 5, base_model: null
                        child 6, method: string
                        child 7, inputs: list<item: struct<kind: string, ref: string, config: null, provenance: null>>
                            child 0, item: struct<kind: string, ref: string, config: null, provenance: null>
                                child 0, kind: string
                                child 1, ref: string
                                child 2, config: null
                                child 3, provenance: null
                        child 8, params: struct<generator_model: null, judge_model: null, grpo_scoring_model: null, export_formats: list<item (... 60 chars omitted)
                            child 0, generator_model: null
                            child 1, judge_model: null
                            chi
              ...
              otal_flos: double
              is_hyper_param_search: bool
              stateful_callbacks: struct<TrainerControl: struct<args: struct<should_epoch_stop: bool, should_evaluate: bool, should_lo (... 79 chars omitted)
                child 0, TrainerControl: struct<args: struct<should_epoch_stop: bool, should_evaluate: bool, should_log: bool, should_save: b (... 55 chars omitted)
                    child 0, args: struct<should_epoch_stop: bool, should_evaluate: bool, should_log: bool, should_save: bool, should_t (... 19 chars omitted)
                        child 0, should_epoch_stop: bool
                        child 1, should_evaluate: bool
                        child 2, should_log: bool
                        child 3, should_save: bool
                        child 4, should_training_stop: bool
                    child 1, attributes: struct<>
              num_train_epochs: int64
              max_steps: int64
              global_step: int64
              epoch: double
              best_global_step: null
              trial_name: null
              eval_steps: int64
              save_steps: int64
              best_metric: null
              logging_steps: int64
              is_local_process_zero: bool
              log_history: list<item: struct<entropy: double, epoch: double, grad_norm: double, learning_rate: double, loss: do (... 68 chars omitted)
                child 0, item: struct<entropy: double, epoch: double, grad_norm: double, learning_rate: double, loss: double, mean_ (... 56 chars omitted)
                    child 0, entropy: double
                    child 1, epoch: double
                    child 2, grad_norm: double
                    child 3, learning_rate: double
                    child 4, loss: double
                    child 5, mean_token_accuracy: double
                    child 6, num_tokens: double
                    child 7, step: int64
              trial_params: null
              to
              {'best_global_step': Value('null'), 'best_metric': Value('null'), 'best_model_checkpoint': Value('null'), 'epoch': Value('float64'), 'eval_steps': Value('int64'), 'global_step': Value('int64'), 'is_hyper_param_search': Value('bool'), 'is_local_process_zero': Value('bool'), 'is_world_process_zero': Value('bool'), 'log_history': List({'entropy': Value('float64'), 'epoch': Value('float64'), 'grad_norm': Value('float64'), 'learning_rate': Value('float64'), 'loss': Value('float64'), 'mean_token_accuracy': Value('float64'), 'num_tokens': Value('float64'), 'step': Value('int64')}), 'logging_steps': Value('int64'), 'max_steps': Value('int64'), 'num_input_tokens_seen': Value('int64'), 'num_train_epochs': Value('int64'), 'save_steps': Value('int64'), 'stateful_callbacks': {'TrainerControl': {'args': {'should_epoch_stop': Value('bool'), 'should_evaluate': Value('bool'), 'should_log': Value('bool'), 'should_save': Value('bool'), 'should_training_stop': Value('bool')}, 'attributes': {}}}, 'total_flos': Value('float64'), 'train_batch_size': Value('int64'), 'trial_name': Value('null'), 'trial_params': Value('null')}
              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.

Multilingual Ticket Triage Evaluation Evidence

Immutable validation/test fixtures, preserved predictions and metrics, AuditKIT reports, red-team evidence, a Gemini-assisted sample-review packet, and selected training traces.

The final one-shot test contains 3,642 rows. Base versus selected adversarial checkpoint-1000 intent accuracy is 0.4803 versus 0.7789; PII recall is 0.8016 versus 0.9609. Metric files and predictions are included under artifacts/evidence/test-v2-* with their hashes in the evidence ledger.

Held-out A5–A7 red-team evidence (900 rows per model):

  • artifacts/evidence/redteam-adversarial-1000-900-merged/: selected checkpoint-1000 predictions and metrics, merged from the A5/A6 prefix run (redteam-adversarial-1000-900/, first 600 rows used) and the A7 shard (redteam-adversarial-1000-a7-shard/).
  • artifacts/evidence/redteam-strict-payload-match/: per-row output of the deterministic strict payload-match grader for Global base, Earth base and checkpoint-1000. Strict ASR is 16.56%, 16.56% and 14.11%; excluding rows whose gold label already equals the requested value, 10.44%, 10.44% and 7.56%. Model-reported detection recall is 37.00%, 33.11% and 92.00%.
  • artifacts/review-export/: the compact per-row files used for the Claude-assisted review.
  • reviews/PREDICTION-REVIEW.md (grader analysis) and reviews/CLAUDE-REVIEW.md (Claude-assisted labels; direct follows 3.0%, 3.1% and 1.7%; payload leakage into redacted_text 15.9%, 12.8% and 27.9%), with the verbatim reviewer output in reviews/CLAUDE-REVIEW.raw.md. Neither is human validation.

The superseded human_review.json filename is deliberately excluded; the self-contained rule-based payload-match artifact is published instead. The 60-row Yorùbá/Kiswahili sample CSV and additional_review_notes.md record a Google Gemini-assisted review: 44/60 correct, 59/60 fluent, one off-language, and two privacy leaks. No native speaker was available, so this is not human validation. MasakhaNER 2.0 content makes the package research/non-commercial.

Downloads last month
71

Collection including Fill-My-Cup/multilingual-ticket-triage-evaluation