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
suite: string
selection_uses_model_outputs: bool
evaluation_contract: struct<episodes_per_set: int64, oracle_solution_max_steps: int64, evaluation_max_steps: int64, image (... 93 chars omitted)
child 0, episodes_per_set: int64
child 1, oracle_solution_max_steps: int64
child 2, evaluation_max_steps: int64
child 3, image_quality: int64
child 4, decoder: string
child 5, rollout: string
child 6, action_mode: string
child 7, success_mode: string
sets: struct<2box_TEMPTING_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, (... 862 chars omitted)
child 0, 2box_TEMPTING_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, solution_len_min: int64, solution_l (... 30 chars omitted)
child 0, path: string
child 1, sha256: string
child 2, n: int64
child 3, num_boxes: int64
child 4, solution_len_min: int64
child 5, solution_len_max: int64
child 6, source: string
child 1, 3box_CERTIFIED_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, solution_len_min: int64, solution_l (... 30 chars omitted)
child 0, path: string
child 1, sha256: string
child 2, n: int64
child 3, num_boxes: int64
child 4, solution_len_min: int64
child 5, solution_len_max: int64
child 6, source: string
child 2, 4box_CERTIFIED_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, solution_len_min: int64, solutio
...
m: int64>
child 0, item: int64
child 10, room_fixed: list<item: list<item: int64>>
child 0, item: list<item: int64>
child 0, item: int64
child 11, step_count: int64
child 12, success_mode: string
fatal_push: struct<action_text: string, box_from: list<item: int64>, box_to: list<item: int64>, deadlock_type: s (... 108 chars omitted)
child 0, action_text: string
child 1, box_from: list<item: int64>
child 0, item: int64
child 2, box_to: list<item: int64>
child 0, item: int64
child 3, deadlock_type: string
child 4, dir: list<item: int64>
child 0, item: int64
child 5, is_dead_corner: bool
child 6, reach: int64
child 7, temptation: double
child 8, toward_target: int64
source_pool: string
paper_v3_source_global_index: int64
safe_solution_len: int64
global_index: int64
reach: int64
temptation_tier: int64
env_kwargs: struct<action_mode: string, dim_room: list<item: int64>, max_steps: int64, num_boxes: int64>
child 0, action_mode: string
child 1, dim_room: list<item: int64>
child 0, item: int64
child 2, max_steps: int64
child 3, num_boxes: int64
temptation_score: double
paper_eval_set: string
seed: int64
n_fatal_pushes: int64
paper_v3_upstream_source_global_index: int64
oracle_or_expert_solution_len: int64
source_global_index: int64
n_safe_pushes: int64
near_first_verdict: string
alt_deadlock_frame_b64: string
source_temptation_tier: int64
init_state_hash: string
paper_eval_version: string
toward_target: int64
to
{'alt_deadlock_frame_b64': Value('string'), 'deadlock_type': Value('string'), 'env_id': Value('string'), 'env_kwargs': {'action_mode': Value('string'), 'dim_room': List(Value('int64')), 'max_steps': Value('int64'), 'num_boxes': Value('int64')}, 'fatal_push': {'action_text': Value('string'), 'box_from': List(Value('int64')), 'box_to': List(Value('int64')), 'deadlock_type': Value('string'), 'dir': List(Value('int64')), 'is_dead_corner': Value('bool'), 'reach': Value('int64'), 'temptation': Value('float64'), 'toward_target': Value('int64')}, 'global_index': Value('int64'), 'init_state': {'action_mode': Value('string'), 'box_positions': List(List(Value('int64'))), 'constructive_solution': Value('null'), 'constructive_solution_schema': Value('string'), 'dim_room': List(Value('int64')), 'legacy_single_move_branch': Value('bool'), 'max_steps': Value('int64'), 'num_boxes': Value('int64'), 'num_gen_steps': Value('int64'), 'player_position': List(Value('int64')), 'room_fixed': List(List(Value('int64'))), 'step_count': Value('int64'), 'success_mode': Value('string')}, 'init_state_hash': Value('string'), 'is_dead_corner': Value('bool'), 'n_fatal_pushes': Value('int64'), 'n_safe_pushes': Value('int64'), 'near_first_verdict': Value('string'), 'oracle_or_expert_solution_len': Value('int64'), 'paper_eval_set': Value('string'), 'paper_eval_version': Value('string'), 'paper_final_set': Value('string'), 'paper_final_suite': Value('string'), 'paper_v3_source_global_index': Value('int64'), 'paper_v3_upstream_source_global_index': Value('int64'), 'reach': Value('int64'), 'safe_solution_len': Value('int64'), 'seed': Value('int64'), 'source_global_index': Value('int64'), 'source_pool': Value('string'), 'source_temptation_tier': Value('int64'), 'temptation_score': Value('float64'), 'temptation_tier': Value('int64'), 'toward_target': Value('int64')}
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
suite: string
selection_uses_model_outputs: bool
evaluation_contract: struct<episodes_per_set: int64, oracle_solution_max_steps: int64, evaluation_max_steps: int64, image (... 93 chars omitted)
child 0, episodes_per_set: int64
child 1, oracle_solution_max_steps: int64
child 2, evaluation_max_steps: int64
child 3, image_quality: int64
child 4, decoder: string
child 5, rollout: string
child 6, action_mode: string
child 7, success_mode: string
sets: struct<2box_TEMPTING_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, (... 862 chars omitted)
child 0, 2box_TEMPTING_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, solution_len_min: int64, solution_l (... 30 chars omitted)
child 0, path: string
child 1, sha256: string
child 2, n: int64
child 3, num_boxes: int64
child 4, solution_len_min: int64
child 5, solution_len_max: int64
child 6, source: string
child 1, 3box_CERTIFIED_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, solution_len_min: int64, solution_l (... 30 chars omitted)
child 0, path: string
child 1, sha256: string
child 2, n: int64
child 3, num_boxes: int64
child 4, solution_len_min: int64
child 5, solution_len_max: int64
child 6, source: string
child 2, 4box_CERTIFIED_DEADLOCK_100: struct<path: string, sha256: string, n: int64, num_boxes: int64, solution_len_min: int64, solutio
...
m: int64>
child 0, item: int64
child 10, room_fixed: list<item: list<item: int64>>
child 0, item: list<item: int64>
child 0, item: int64
child 11, step_count: int64
child 12, success_mode: string
fatal_push: struct<action_text: string, box_from: list<item: int64>, box_to: list<item: int64>, deadlock_type: s (... 108 chars omitted)
child 0, action_text: string
child 1, box_from: list<item: int64>
child 0, item: int64
child 2, box_to: list<item: int64>
child 0, item: int64
child 3, deadlock_type: string
child 4, dir: list<item: int64>
child 0, item: int64
child 5, is_dead_corner: bool
child 6, reach: int64
child 7, temptation: double
child 8, toward_target: int64
source_pool: string
paper_v3_source_global_index: int64
safe_solution_len: int64
global_index: int64
reach: int64
temptation_tier: int64
env_kwargs: struct<action_mode: string, dim_room: list<item: int64>, max_steps: int64, num_boxes: int64>
child 0, action_mode: string
child 1, dim_room: list<item: int64>
child 0, item: int64
child 2, max_steps: int64
child 3, num_boxes: int64
temptation_score: double
paper_eval_set: string
seed: int64
n_fatal_pushes: int64
paper_v3_upstream_source_global_index: int64
oracle_or_expert_solution_len: int64
source_global_index: int64
n_safe_pushes: int64
near_first_verdict: string
alt_deadlock_frame_b64: string
source_temptation_tier: int64
init_state_hash: string
paper_eval_version: string
toward_target: int64
to
{'alt_deadlock_frame_b64': Value('string'), 'deadlock_type': Value('string'), 'env_id': Value('string'), 'env_kwargs': {'action_mode': Value('string'), 'dim_room': List(Value('int64')), 'max_steps': Value('int64'), 'num_boxes': Value('int64')}, 'fatal_push': {'action_text': Value('string'), 'box_from': List(Value('int64')), 'box_to': List(Value('int64')), 'deadlock_type': Value('string'), 'dir': List(Value('int64')), 'is_dead_corner': Value('bool'), 'reach': Value('int64'), 'temptation': Value('float64'), 'toward_target': Value('int64')}, 'global_index': Value('int64'), 'init_state': {'action_mode': Value('string'), 'box_positions': List(List(Value('int64'))), 'constructive_solution': Value('null'), 'constructive_solution_schema': Value('string'), 'dim_room': List(Value('int64')), 'legacy_single_move_branch': Value('bool'), 'max_steps': Value('int64'), 'num_boxes': Value('int64'), 'num_gen_steps': Value('int64'), 'player_position': List(Value('int64')), 'room_fixed': List(List(Value('int64'))), 'step_count': Value('int64'), 'success_mode': Value('string')}, 'init_state_hash': Value('string'), 'is_dead_corner': Value('bool'), 'n_fatal_pushes': Value('int64'), 'n_safe_pushes': Value('int64'), 'near_first_verdict': Value('string'), 'oracle_or_expert_solution_len': Value('int64'), 'paper_eval_set': Value('string'), 'paper_eval_version': Value('string'), 'paper_final_set': Value('string'), 'paper_final_suite': Value('string'), 'paper_v3_source_global_index': Value('int64'), 'paper_v3_upstream_source_global_index': Value('int64'), 'reach': Value('int64'), 'safe_solution_len': Value('int64'), 'seed': Value('int64'), 'source_global_index': Value('int64'), 'source_pool': Value('string'), 'source_temptation_tier': Value('int64'), 'temptation_score': Value('float64'), 'temptation_tier': Value('int64'), 'toward_target': Value('int64')}
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.
3-box Deadlock
This is the exact 100-episode shard used by the finalized UWM Sokoban evaluations. Episode selection is model-independent; no checkpoint output or evaluation score was used to select an episode.
Evaluation contract
- suite:
sokoban_paper_final_v2_max40 - boxes: 3
- oracle/expert solution length: 5–20 actions, always ≤20
- model execution budget: 40 actions
- image quality: JPEG q95
- decoder:
perseg - rollout: full visual self-rollout
- action interface: move-only
- success: solved and final action is stop
- shard SHA-256:
1025881fc8948b68e36786f3490c137c723ef89fdc6d07085478170c0f93e2e1
Files
shard_00000.jsonl.gz: the 100 evaluation episodessuite_manifest.json: six-set provenance, hashes, contract, and pairwise-disjointness audit
Full model outputs are published in the paper trajectory collection under
table2/sokoban/3dead/ and table3/sokoban/3dead/.
- Downloads last month
- 37