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
passed: bool
scored: bool
model_calls: int64
episodes: list<item: struct<passed: bool, scored: bool, model_calls: int64, task: string, scope: string, nativ (... 1590 chars omitted)
child 0, item: struct<passed: bool, scored: bool, model_calls: int64, task: string, scope: string, native_record_sh (... 1578 chars omitted)
child 0, passed: bool
child 1, scored: bool
child 2, model_calls: int64
child 3, task: string
child 4, scope: string
child 5, native_record_sha256: string
child 6, initial_state_max_error: double
child 7, restored_terminal_state_max_error: double
child 8, recorded_native_success: bool
child 9, recorded_success_latched: bool
child 10, terminal_predicate: bool
child 11, conditions: struct<scan_axis_distance_m: bool, scan_axis_residual_xyz_m: bool, left_gripper_closed: bool, right_ (... 292 chars omitted)
child 0, scan_axis_distance_m: bool
child 1, scan_axis_residual_xyz_m: bool
child 2, left_gripper_closed: bool
child 3, right_gripper_closed: bool
child 4, lower_pair_xy_distance_m: bool
child 5, upper_pair_xy_distance_m: bool
child 6, height_excess_over_targets_m: bool
child 7, left_gripper_open: bool
child 8, right_gripper_open: bool
child 9, pot_root_height_m: bool
child 10, left_native_tcp_to_handle_m: bool
child 11, right_native_tcp_to_handle_m: bool
child
...
ps: int64
child 15, image: string
child 16, gpu: int64
tasks: list<item: struct<passed: bool, task: string, container: string, model: struct<kinex_identity: strin (... 486 chars omitted)
child 0, item: struct<passed: bool, task: string, container: string, model: struct<kinex_identity: string, kinex_ve (... 474 chars omitted)
child 0, passed: bool
child 1, task: string
child 2, container: string
child 3, model: struct<kinex_identity: string, kinex_version: string, kinex_source: struct<revision: string, source: (... 236 chars omitted)
child 0, kinex_identity: string
child 1, kinex_version: string
child 2, kinex_source: struct<revision: string, source: string>
child 0, revision: string
child 1, source: string
child 3, model: string
child 4, effort: string
child 5, service_tier: string
child 6, fresh_session: bool
child 7, max_steps: null
child 8, model_recording: bool
child 9, compaction: string
child 10, transport: string
child 11, strict_usage: bool
child 12, max_request_retries: int64
child 13, usage_accounting: string
child 4, actual_prompt_sha256: string
child 5, authorized_account_matches: bool
child 6, modified_instruction_in_first_user_message: bool
child 7, instruction_policy: string
child 8, provider_requests_checked: int64
scope: string
to
{'passed': Value('bool'), 'tasks': List({'passed': Value('bool'), 'task': Value('string'), 'container': Value('string'), 'model': {'kinex_identity': Value('string'), 'kinex_version': Value('string'), 'kinex_source': {'revision': Value('string'), 'source': Value('string')}, 'model': Value('string'), 'effort': Value('string'), 'service_tier': Value('string'), 'fresh_session': Value('bool'), 'max_steps': Value('null'), 'model_recording': Value('bool'), 'compaction': Value('string'), 'transport': Value('string'), 'strict_usage': Value('bool'), 'max_request_retries': Value('int64'), 'usage_accounting': Value('string')}, 'actual_prompt_sha256': Value('string'), 'authorized_account_matches': Value('bool'), 'modified_instruction_in_first_user_message': Value('bool'), 'instruction_policy': Value('string'), 'provider_requests_checked': Value('int64')}), 'scope': 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 2951, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, 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 547, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, 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
passed: bool
scored: bool
model_calls: int64
episodes: list<item: struct<passed: bool, scored: bool, model_calls: int64, task: string, scope: string, nativ (... 1590 chars omitted)
child 0, item: struct<passed: bool, scored: bool, model_calls: int64, task: string, scope: string, native_record_sh (... 1578 chars omitted)
child 0, passed: bool
child 1, scored: bool
child 2, model_calls: int64
child 3, task: string
child 4, scope: string
child 5, native_record_sha256: string
child 6, initial_state_max_error: double
child 7, restored_terminal_state_max_error: double
child 8, recorded_native_success: bool
child 9, recorded_success_latched: bool
child 10, terminal_predicate: bool
child 11, conditions: struct<scan_axis_distance_m: bool, scan_axis_residual_xyz_m: bool, left_gripper_closed: bool, right_ (... 292 chars omitted)
child 0, scan_axis_distance_m: bool
child 1, scan_axis_residual_xyz_m: bool
child 2, left_gripper_closed: bool
child 3, right_gripper_closed: bool
child 4, lower_pair_xy_distance_m: bool
child 5, upper_pair_xy_distance_m: bool
child 6, height_excess_over_targets_m: bool
child 7, left_gripper_open: bool
child 8, right_gripper_open: bool
child 9, pot_root_height_m: bool
child 10, left_native_tcp_to_handle_m: bool
child 11, right_native_tcp_to_handle_m: bool
child
...
ps: int64
child 15, image: string
child 16, gpu: int64
tasks: list<item: struct<passed: bool, task: string, container: string, model: struct<kinex_identity: strin (... 486 chars omitted)
child 0, item: struct<passed: bool, task: string, container: string, model: struct<kinex_identity: string, kinex_ve (... 474 chars omitted)
child 0, passed: bool
child 1, task: string
child 2, container: string
child 3, model: struct<kinex_identity: string, kinex_version: string, kinex_source: struct<revision: string, source: (... 236 chars omitted)
child 0, kinex_identity: string
child 1, kinex_version: string
child 2, kinex_source: struct<revision: string, source: string>
child 0, revision: string
child 1, source: string
child 3, model: string
child 4, effort: string
child 5, service_tier: string
child 6, fresh_session: bool
child 7, max_steps: null
child 8, model_recording: bool
child 9, compaction: string
child 10, transport: string
child 11, strict_usage: bool
child 12, max_request_retries: int64
child 13, usage_accounting: string
child 4, actual_prompt_sha256: string
child 5, authorized_account_matches: bool
child 6, modified_instruction_in_first_user_message: bool
child 7, instruction_policy: string
child 8, provider_requests_checked: int64
scope: string
to
{'passed': Value('bool'), 'tasks': List({'passed': Value('bool'), 'task': Value('string'), 'container': Value('string'), 'model': {'kinex_identity': Value('string'), 'kinex_version': Value('string'), 'kinex_source': {'revision': Value('string'), 'source': Value('string')}, 'model': Value('string'), 'effort': Value('string'), 'service_tier': Value('string'), 'fresh_session': Value('bool'), 'max_steps': Value('null'), 'model_recording': Value('bool'), 'compaction': Value('string'), 'transport': Value('string'), 'strict_usage': Value('bool'), 'max_request_retries': Value('int64'), 'usage_accounting': Value('string')}, 'actual_prompt_sha256': Value('string'), 'authorized_account_matches': Value('bool'), 'modified_instruction_in_first_user_message': Value('bool'), 'instruction_policy': Value('string'), 'provider_requests_checked': Value('int64')}), 'scope': 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.
Kinex Benchmark
Unified media and evidence storage for the Kinex Benchmark Space.
The current index compares Codex + GPT-6 Astra / high (164 episodes), Kinex + GPT-6 Astra / high (164 episodes), and Claude Code + Opus 5.5 / high (152 published episodes). Kinex + Opus 5.5 / high remains planned. Episode protocols record the task seed and native budget; pending tasks are excluded from success denominators.
| Location | Contents |
|---|---|
data.json |
Current 164-task, four-combination catalog with 680 selected episodes |
attempt-history.json |
112 retained previous attempts, excluded from current metrics |
sources.json |
Original provenance and immutable merged asset revision |
variants/codex-astra/ |
Byte-preserved Codex Dataset snapshot |
variants/kinex-astra/ |
Byte-preserved Kinex v0.10.0 Dataset snapshot |
variants/cc-opus/ |
Claude Code pilot episodes with per-episode provenance, usage and media checks |
history/<source>/<revision>/ |
Older evidence revisions still referenced by result pages |
indices/<space>/ |
Derived navigation catalogs and current CSV exports for deprecated result Spaces |
space-assets/ |
Small legacy downloads with bytes distinct from the Dataset snapshot |
migration-manifest.json |
Source repository, revision, original path and hash for every asset |
migration-validation.json |
Full remote inventory and hash verification summary |
Videos, posters, observation images, native sessions, transcripts, workspaces, artifact files, protocols and native verification evidence share one Dataset. Migrated historical media were copied without transcoding. New pilot media carry explicit native-frame comparison receipts. Original evidence payloads and source snapshots are immutable; only the derived navigation catalogs rewrite URLs.
The deprecated Codex and Kinex v0.10.0 Spaces remain usable with this Dataset. Their original Datasets are retained. The separate Kinex v0.8 and RoboBallet publications, along with the Introducing RLE-Bench Space, are outside this migration.
See DATA_FORMAT.md for path mapping and source-repositories.json for frozen source identities.
- Downloads last month
- 498