Dataset Viewer
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
arms: list<item: string>
child 0, item: string
automatic_learning: bool
campaign_id: string
cases: list<item: struct<case_id: string, domain: string, eval_doc_id: int64, expected_route: string, filte (... 91 chars omitted)
child 0, item: struct<case_id: string, domain: string, eval_doc_id: int64, expected_route: string, filter: string, (... 79 chars omitted)
child 0, case_id: string
child 1, domain: string
child 2, eval_doc_id: int64
child 3, expected_route: string
child 4, filter: string
child 5, metric: string
child 6, question_sha256: string
child 7, source_doc_index: int64
child 8, task: string
framework: struct<description: string, name: string, official_leaderboard_submission: bool, version: string>
child 0, description: string
child 1, name: string
child 2, official_leaderboard_submission: bool
child 3, version: string
models: list<item: string>
child 0, item: string
schema_version: string
sealed_test_used_for_training: bool
selection_sha256: string
status: string
target_exposed_to_generator: bool
matmem: struct<exercised: bool, reason: string, verified_answers: int64, bypass_contract: string>
child 0, exercised: bool
child 1, reason: string
child 2, verified_answers: int64
child 3, bypass_contract: string
script_schema_version: string
controls: struct<target_exposed_to_router: bool, target_exposed_to_expert: bool, target_exposed_to_generator: (... 248 chars omitted)
child 0, target_exposed
...
ild 3, left_accuracy: double
child 4, right_accuracy: double
child 5, delta_percentage_points: double
child 6, paired_wins: int64
child 7, paired_losses: int64
child 8, paired_ties: int64
child 9, mcnemar_exact_p_unadjusted: double
child 3, triviaqa: struct<n: int64, route: string, correct_by_arm: struct<llm_direct: int64, router_moe: int64, router_ (... 305 chars omitted)
child 0, n: int64
child 1, route: string
child 2, correct_by_arm: struct<llm_direct: int64, router_moe: int64, router_moe_nexus: int64, router_moe_nexus_matmem: int64 (... 1 chars omitted)
child 0, llm_direct: int64
child 1, router_moe: int64
child 2, router_moe_nexus: int64
child 3, router_moe_nexus_matmem: int64
child 3, direct_vs_full: struct<n: int64, left_correct: int64, right_correct: int64, left_accuracy: double, right_accuracy: d (... 137 chars omitted)
child 0, n: int64
child 1, left_correct: int64
child 2, right_correct: int64
child 3, left_accuracy: double
child 4, right_accuracy: double
child 5, delta_percentage_points: double
child 6, paired_wins: int64
child 7, paired_losses: int64
child 8, paired_ties: int64
child 9, mcnemar_exact_p_unadjusted: double
report_sha256: string
to
{'schema_version': Value('string'), 'script_schema_version': Value('string'), 'status': Value('string'), 'campaign_id': Value('string'), 'model': {'id': Value('string'), 'revision': Value('string'), 'dtype': Value('string'), 'device': Value('string')}, 'framework': {'name': Value('string'), 'version': Value('string'), 'official_prompts_filters_scorers_unchanged': Value('bool'), 'official_leaderboard_submission': Value('bool')}, 'selection_sha256': Value('string'), 'generation_count': Value('int64'), 'arms': List(Value('string')), 'summary': {'n': Value('int64'), 'correct_by_arm': {'llm_direct': Value('int64'), 'router_moe': Value('int64'), 'router_moe_nexus': Value('int64'), 'router_moe_nexus_matmem': Value('int64')}, 'accuracy_by_arm': {'llm_direct': Value('float64'), 'router_moe': Value('float64'), 'router_moe_nexus': Value('float64'), 'router_moe_nexus_matmem': Value('float64')}, 'direct_vs_router_moe': {'n': Value('int64'), 'left_correct': Value('int64'), 'right_correct': Value('int64'), 'left_accuracy': Value('float64'), 'right_accuracy': Value('float64'), 'delta_percentage_points': Value('float64'), 'paired_wins': Value('int64'), 'paired_losses': Value('int64'), 'paired_ties': Value('int64'), 'mcnemar_exact_p_unadjusted': Value('float64')}, 'direct_vs_router_moe_nexus': {'n': Value('int64'), 'left_correct': Value('int64'), 'right_correct': Value('int64'), 'left_accuracy': Value('float64'), 'right_accuracy': Value('float64'), 'delta_percentage_points': Value('float64'),
...
'left_accuracy': Value('float64'), 'right_accuracy': Value('float64'), 'delta_percentage_points': Value('float64'), 'paired_wins': Value('int64'), 'paired_losses': Value('int64'), 'paired_ties': Value('int64'), 'mcnemar_exact_p_unadjusted': Value('float64')}}}}, 'cases': List({'case_key': Value('string'), 'task': Value('string'), 'correct_by_arm': {'llm_direct': Value('bool'), 'router_moe': Value('bool'), 'router_moe_nexus': Value('bool'), 'router_moe_nexus_matmem': Value('bool')}, 'raw_response_sha256': Value('string'), 'candidate_sha256_by_arm': {'llm_direct': Value('string'), 'router_moe': Value('string'), 'router_moe_nexus': Value('string'), 'router_moe_nexus_matmem': Value('string')}, 'route': Value('string'), 'answer_source': Value('string'), 'nexus_verified': Value('bool'), 'matmem_exercised': Value('bool'), 'target_hash': Value('string')}), 'matmem': {'exercised': Value('bool'), 'reason': Value('string'), 'verified_answers': Value('int64'), 'bypass_contract': Value('string')}, 'controls': {'target_exposed_to_router': Value('bool'), 'target_exposed_to_expert': Value('bool'), 'target_exposed_to_generator': Value('bool'), 'targets_persisted': Value('bool'), 'answers_persisted': Value('bool'), 'sealed_test_used_for_training': Value('bool'), 'automatic_learning': Value('bool'), 'selection_hashes_verified_before_model_load': Value('bool'), 'raw_model_generation_reused_across_all_arms': Value('bool'), 'matmem_claim_allowed': Value('bool')}, 'report_sha256': Value('string')}
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 489, 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 2818, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2355, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2380, 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 2369, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
arms: list<item: string>
child 0, item: string
automatic_learning: bool
campaign_id: string
cases: list<item: struct<case_id: string, domain: string, eval_doc_id: int64, expected_route: string, filte (... 91 chars omitted)
child 0, item: struct<case_id: string, domain: string, eval_doc_id: int64, expected_route: string, filter: string, (... 79 chars omitted)
child 0, case_id: string
child 1, domain: string
child 2, eval_doc_id: int64
child 3, expected_route: string
child 4, filter: string
child 5, metric: string
child 6, question_sha256: string
child 7, source_doc_index: int64
child 8, task: string
framework: struct<description: string, name: string, official_leaderboard_submission: bool, version: string>
child 0, description: string
child 1, name: string
child 2, official_leaderboard_submission: bool
child 3, version: string
models: list<item: string>
child 0, item: string
schema_version: string
sealed_test_used_for_training: bool
selection_sha256: string
status: string
target_exposed_to_generator: bool
matmem: struct<exercised: bool, reason: string, verified_answers: int64, bypass_contract: string>
child 0, exercised: bool
child 1, reason: string
child 2, verified_answers: int64
child 3, bypass_contract: string
script_schema_version: string
controls: struct<target_exposed_to_router: bool, target_exposed_to_expert: bool, target_exposed_to_generator: (... 248 chars omitted)
child 0, target_exposed
...
ild 3, left_accuracy: double
child 4, right_accuracy: double
child 5, delta_percentage_points: double
child 6, paired_wins: int64
child 7, paired_losses: int64
child 8, paired_ties: int64
child 9, mcnemar_exact_p_unadjusted: double
child 3, triviaqa: struct<n: int64, route: string, correct_by_arm: struct<llm_direct: int64, router_moe: int64, router_ (... 305 chars omitted)
child 0, n: int64
child 1, route: string
child 2, correct_by_arm: struct<llm_direct: int64, router_moe: int64, router_moe_nexus: int64, router_moe_nexus_matmem: int64 (... 1 chars omitted)
child 0, llm_direct: int64
child 1, router_moe: int64
child 2, router_moe_nexus: int64
child 3, router_moe_nexus_matmem: int64
child 3, direct_vs_full: struct<n: int64, left_correct: int64, right_correct: int64, left_accuracy: double, right_accuracy: d (... 137 chars omitted)
child 0, n: int64
child 1, left_correct: int64
child 2, right_correct: int64
child 3, left_accuracy: double
child 4, right_accuracy: double
child 5, delta_percentage_points: double
child 6, paired_wins: int64
child 7, paired_losses: int64
child 8, paired_ties: int64
child 9, mcnemar_exact_p_unadjusted: double
report_sha256: string
to
{'schema_version': Value('string'), 'script_schema_version': Value('string'), 'status': Value('string'), 'campaign_id': Value('string'), 'model': {'id': Value('string'), 'revision': Value('string'), 'dtype': Value('string'), 'device': Value('string')}, 'framework': {'name': Value('string'), 'version': Value('string'), 'official_prompts_filters_scorers_unchanged': Value('bool'), 'official_leaderboard_submission': Value('bool')}, 'selection_sha256': Value('string'), 'generation_count': Value('int64'), 'arms': List(Value('string')), 'summary': {'n': Value('int64'), 'correct_by_arm': {'llm_direct': Value('int64'), 'router_moe': Value('int64'), 'router_moe_nexus': Value('int64'), 'router_moe_nexus_matmem': Value('int64')}, 'accuracy_by_arm': {'llm_direct': Value('float64'), 'router_moe': Value('float64'), 'router_moe_nexus': Value('float64'), 'router_moe_nexus_matmem': Value('float64')}, 'direct_vs_router_moe': {'n': Value('int64'), 'left_correct': Value('int64'), 'right_correct': Value('int64'), 'left_accuracy': Value('float64'), 'right_accuracy': Value('float64'), 'delta_percentage_points': Value('float64'), 'paired_wins': Value('int64'), 'paired_losses': Value('int64'), 'paired_ties': Value('int64'), 'mcnemar_exact_p_unadjusted': Value('float64')}, 'direct_vs_router_moe_nexus': {'n': Value('int64'), 'left_correct': Value('int64'), 'right_correct': Value('int64'), 'left_accuracy': Value('float64'), 'right_accuracy': Value('float64'), 'delta_percentage_points': Value('float64'),
...
'left_accuracy': Value('float64'), 'right_accuracy': Value('float64'), 'delta_percentage_points': Value('float64'), 'paired_wins': Value('int64'), 'paired_losses': Value('int64'), 'paired_ties': Value('int64'), 'mcnemar_exact_p_unadjusted': Value('float64')}}}}, 'cases': List({'case_key': Value('string'), 'task': Value('string'), 'correct_by_arm': {'llm_direct': Value('bool'), 'router_moe': Value('bool'), 'router_moe_nexus': Value('bool'), 'router_moe_nexus_matmem': Value('bool')}, 'raw_response_sha256': Value('string'), 'candidate_sha256_by_arm': {'llm_direct': Value('string'), 'router_moe': Value('string'), 'router_moe_nexus': Value('string'), 'router_moe_nexus_matmem': Value('string')}, 'route': Value('string'), 'answer_source': Value('string'), 'nexus_verified': Value('bool'), 'matmem_exercised': Value('bool'), 'target_hash': Value('string')}), 'matmem': {'exercised': Value('bool'), 'reason': Value('string'), 'verified_answers': Value('int64'), 'bypass_contract': Value('string')}, 'controls': {'target_exposed_to_router': Value('bool'), 'target_exposed_to_expert': Value('bool'), 'target_exposed_to_generator': Value('bool'), 'targets_persisted': Value('bool'), 'answers_persisted': Value('bool'), 'sealed_test_used_for_training': Value('bool'), 'automatic_learning': Value('bool'), 'selection_hashes_verified_before_model_load': Value('bool'), 'raw_model_generation_reused_across_all_arms': Value('bool'), 'matmem_claim_allowed': Value('bool')}, 'report_sha256': Value('string')}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
MAT Nexus Hugging Face evaluation evidence
This public dataset stores immutable evaluation runners and frozen selection manifests for reproducible, target-blind infrastructure tests.
The 100-case confirmation is not an official leaderboard submission and does not exercise the private MATmem index. Its frozen selection was consumed before generation and must never be used for training.
Public bundle
runners/run_hf_jobs_granite_nexus_confirmation_100q_v1.pyselections/hf-jobs-granite-nexus-confirmation-100q-v1/selection-manifest.json
License: copyright retained; public evaluation and reproducibility use only. See the source project for the complete public/private boundary.
- Downloads last month
- 83