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
item_id: string
gold: string
completion: string
n_gen_tokens: int64
hit_cap: bool
stopped_on: null
batch_index: int64
batch_seed: int64
n_options: int64
predicted: string
correct: bool
item_digest: string
complete: bool
metrics: struct<n: int64, exact_match: double, unreduced: int64>
child 0, n: int64
child 1, exact_match: double
child 2, unreduced: int64
records_file: string
shard: struct<index: int64, of: int64, n_assigned: int64>
child 0, index: int64
child 1, of: int64
child 2, n_assigned: int64
runtime_s: double
protocol: struct<benchmark: string, n_items: int64, decode: struct<max_new_tokens: int64, temperature: double, (... 319 chars omitted)
child 0, benchmark: string
child 1, n_items: int64
child 2, decode: struct<max_new_tokens: int64, temperature: double, top_p: null, top_k: null, do_sample: bool, num_be (... 112 chars omitted)
child 0, max_new_tokens: int64
child 1, temperature: double
child 2, top_p: null
child 3, top_k: null
child 4, do_sample: bool
child 5, num_beams: int64
child 6, stop: list<item: null>
child 0, item: null
child 7, n_repeats: int64
child 8, source: string
child 9, inferred: bool
child 10, notes: list<item: string>
child 0, item: string
child 3, source: struct<repo: string, files: list<item: string>, retrieved: timestamp[s], notes: list<item: string>>
child 0, repo: string
child 1, files: list<item: string>
child 0, i
...
mory_gb: double
child 2, count: int64
child 3, cuda: string
adapter: null
batch_size: int64
model_profile: struct<profile: string, declared: struct<key: string, repo: string, arch_class: string, pixel_budget (... 401 chars omitted)
child 0, profile: string
child 1, declared: struct<key: string, repo: string, arch_class: string, pixel_budget: struct<min_pixels: int64, max_pi (... 274 chars omitted)
child 0, key: string
child 1, repo: string
child 2, arch_class: string
child 3, pixel_budget: struct<min_pixels: int64, max_pixels: int64>
child 0, min_pixels: int64
child 1, max_pixels: int64
child 4, pixel_source: string
child 5, image_input_style: string
child 6, chat_template_default_system: string
child 7, geometry: struct<n_layers: int64, n_q_heads: int64, n_kv_heads: int64, head_dim: int64, hidden_size: int64, at (... 16 chars omitted)
child 0, n_layers: int64
child 1, n_q_heads: int64
child 2, n_kv_heads: int64
child 3, head_dim: int64
child 4, hidden_size: int64
child 5, attn_width: int64
child 8, has_qk_norm: bool
child 9, notes: list<item: string>
child 0, item: string
child 2, observed: struct<n_layers: int64, n_q_heads: int64, n_kv_heads: int64, hidden_size: int64>
child 0, n_layers: int64
child 1, n_q_heads: int64
child 2, n_kv_heads: int64
child 3, hidden_size: int64
what: string
to
{'schema_version': Value('int64'), 'what': Value('string'), 'repo': Value('string'), 'arm': Value('string'), 'adapter': Value('null'), 'pixel_budget': {'min_pixels': Value('int64'), 'max_pixels': Value('int64')}, 'pixel_source': Value('string'), 'model_profile': {'profile': Value('string'), 'declared': {'key': Value('string'), 'repo': Value('string'), 'arch_class': Value('string'), 'pixel_budget': {'min_pixels': Value('int64'), 'max_pixels': Value('int64')}, 'pixel_source': Value('string'), 'image_input_style': Value('string'), 'chat_template_default_system': Value('string'), 'geometry': {'n_layers': Value('int64'), 'n_q_heads': Value('int64'), 'n_kv_heads': Value('int64'), 'head_dim': Value('int64'), 'hidden_size': Value('int64'), 'attn_width': Value('int64')}, 'has_qk_norm': Value('bool'), 'notes': List(Value('string'))}, 'observed': {'n_layers': Value('int64'), 'n_q_heads': Value('int64'), 'n_kv_heads': Value('int64'), 'hidden_size': Value('int64')}}, 'system_policy': {'declared_by_benchmark': Value('bool'), 'injected_by_template': Value('string'), 'effective_when_undeclared': Value('null'), 'silent_injection': Value('bool'), 'profile_known': Value('bool')}, 'protocol': {'benchmark': Value('string'), 'n_items': Value('int64'), 'decode': {'max_new_tokens': Value('int64'), 'temperature': Value('float64'), 'top_p': Value('null'), 'top_k': Value('null'), 'do_sample': Value('bool'), 'num_beams': Value('int64'), 'stop': List(Value('null')), 'n_repeats': Value('int64'), 'source':
...
iles': List(Value('string')), 'retrieved': Value('timestamp[s]'), 'notes': List(Value('string'))}, 'sends_system': Value('bool'), 'anchors': List(Value('null'))}, 'shard': {'index': Value('int64'), 'of': Value('int64'), 'n_assigned': Value('int64')}, 'batch_size': Value('int64'), 'run_seed': Value('int64'), 'resumed_from': Value('int64'), 'checkpoint_generation_defaults': {'do_sample': Value('bool'), 'temperature': Value('float64'), 'top_p': Value('null'), 'top_k': Value('null'), 'repetition_penalty': Value('float64')}, 'item_digest': Value('string'), 'complete': Value('bool'), 'n_records': Value('int64'), 'runtime_s': Value('float64'), 'records_file': Value('string'), 'env': {'time_ist': Value('string'), 'time_utc': Value('timestamp[s]'), 'host': Value('string'), 'user': Value('string'), 'python': Value('string'), 'platform': Value('string'), 'git_sha': Value('null'), 'git_dirty': Value('null'), 'torch_version': Value('string'), 'transformers_version': Value('string'), 'datasets_version': Value('string'), 'accelerate_version': Value('string'), 'gpu': {'name': Value('string'), 'total_memory_gb': Value('float64'), 'count': Value('int64'), 'cuda': Value('string')}}, 'metrics': {'n': Value('int64'), 'exact_match': Value('float64'), 'unreduced': Value('int64')}, 'mlp_adapter': {'npz': Value('string'), 'layers': List(Value('int64')), 'rank': Value('int64'), 'alpha': Value('int64'), 'arm': Value('string'), 'deployment': Value('string'), 'attention_implementation': 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 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
item_id: string
gold: string
completion: string
n_gen_tokens: int64
hit_cap: bool
stopped_on: null
batch_index: int64
batch_seed: int64
n_options: int64
predicted: string
correct: bool
item_digest: string
complete: bool
metrics: struct<n: int64, exact_match: double, unreduced: int64>
child 0, n: int64
child 1, exact_match: double
child 2, unreduced: int64
records_file: string
shard: struct<index: int64, of: int64, n_assigned: int64>
child 0, index: int64
child 1, of: int64
child 2, n_assigned: int64
runtime_s: double
protocol: struct<benchmark: string, n_items: int64, decode: struct<max_new_tokens: int64, temperature: double, (... 319 chars omitted)
child 0, benchmark: string
child 1, n_items: int64
child 2, decode: struct<max_new_tokens: int64, temperature: double, top_p: null, top_k: null, do_sample: bool, num_be (... 112 chars omitted)
child 0, max_new_tokens: int64
child 1, temperature: double
child 2, top_p: null
child 3, top_k: null
child 4, do_sample: bool
child 5, num_beams: int64
child 6, stop: list<item: null>
child 0, item: null
child 7, n_repeats: int64
child 8, source: string
child 9, inferred: bool
child 10, notes: list<item: string>
child 0, item: string
child 3, source: struct<repo: string, files: list<item: string>, retrieved: timestamp[s], notes: list<item: string>>
child 0, repo: string
child 1, files: list<item: string>
child 0, i
...
mory_gb: double
child 2, count: int64
child 3, cuda: string
adapter: null
batch_size: int64
model_profile: struct<profile: string, declared: struct<key: string, repo: string, arch_class: string, pixel_budget (... 401 chars omitted)
child 0, profile: string
child 1, declared: struct<key: string, repo: string, arch_class: string, pixel_budget: struct<min_pixels: int64, max_pi (... 274 chars omitted)
child 0, key: string
child 1, repo: string
child 2, arch_class: string
child 3, pixel_budget: struct<min_pixels: int64, max_pixels: int64>
child 0, min_pixels: int64
child 1, max_pixels: int64
child 4, pixel_source: string
child 5, image_input_style: string
child 6, chat_template_default_system: string
child 7, geometry: struct<n_layers: int64, n_q_heads: int64, n_kv_heads: int64, head_dim: int64, hidden_size: int64, at (... 16 chars omitted)
child 0, n_layers: int64
child 1, n_q_heads: int64
child 2, n_kv_heads: int64
child 3, head_dim: int64
child 4, hidden_size: int64
child 5, attn_width: int64
child 8, has_qk_norm: bool
child 9, notes: list<item: string>
child 0, item: string
child 2, observed: struct<n_layers: int64, n_q_heads: int64, n_kv_heads: int64, hidden_size: int64>
child 0, n_layers: int64
child 1, n_q_heads: int64
child 2, n_kv_heads: int64
child 3, hidden_size: int64
what: string
to
{'schema_version': Value('int64'), 'what': Value('string'), 'repo': Value('string'), 'arm': Value('string'), 'adapter': Value('null'), 'pixel_budget': {'min_pixels': Value('int64'), 'max_pixels': Value('int64')}, 'pixel_source': Value('string'), 'model_profile': {'profile': Value('string'), 'declared': {'key': Value('string'), 'repo': Value('string'), 'arch_class': Value('string'), 'pixel_budget': {'min_pixels': Value('int64'), 'max_pixels': Value('int64')}, 'pixel_source': Value('string'), 'image_input_style': Value('string'), 'chat_template_default_system': Value('string'), 'geometry': {'n_layers': Value('int64'), 'n_q_heads': Value('int64'), 'n_kv_heads': Value('int64'), 'head_dim': Value('int64'), 'hidden_size': Value('int64'), 'attn_width': Value('int64')}, 'has_qk_norm': Value('bool'), 'notes': List(Value('string'))}, 'observed': {'n_layers': Value('int64'), 'n_q_heads': Value('int64'), 'n_kv_heads': Value('int64'), 'hidden_size': Value('int64')}}, 'system_policy': {'declared_by_benchmark': Value('bool'), 'injected_by_template': Value('string'), 'effective_when_undeclared': Value('null'), 'silent_injection': Value('bool'), 'profile_known': Value('bool')}, 'protocol': {'benchmark': Value('string'), 'n_items': Value('int64'), 'decode': {'max_new_tokens': Value('int64'), 'temperature': Value('float64'), 'top_p': Value('null'), 'top_k': Value('null'), 'do_sample': Value('bool'), 'num_beams': Value('int64'), 'stop': List(Value('null')), 'n_repeats': Value('int64'), 'source':
...
iles': List(Value('string')), 'retrieved': Value('timestamp[s]'), 'notes': List(Value('string'))}, 'sends_system': Value('bool'), 'anchors': List(Value('null'))}, 'shard': {'index': Value('int64'), 'of': Value('int64'), 'n_assigned': Value('int64')}, 'batch_size': Value('int64'), 'run_seed': Value('int64'), 'resumed_from': Value('int64'), 'checkpoint_generation_defaults': {'do_sample': Value('bool'), 'temperature': Value('float64'), 'top_p': Value('null'), 'top_k': Value('null'), 'repetition_penalty': Value('float64')}, 'item_digest': Value('string'), 'complete': Value('bool'), 'n_records': Value('int64'), 'runtime_s': Value('float64'), 'records_file': Value('string'), 'env': {'time_ist': Value('string'), 'time_utc': Value('timestamp[s]'), 'host': Value('string'), 'user': Value('string'), 'python': Value('string'), 'platform': Value('string'), 'git_sha': Value('null'), 'git_dirty': Value('null'), 'torch_version': Value('string'), 'transformers_version': Value('string'), 'datasets_version': Value('string'), 'accelerate_version': Value('string'), 'gpu': {'name': Value('string'), 'total_memory_gb': Value('float64'), 'count': Value('int64'), 'cuda': Value('string')}}, 'metrics': {'n': Value('int64'), 'exact_match': Value('float64'), 'unreduced': Value('int64')}, 'mlp_adapter': {'npz': Value('string'), 'layers': List(Value('int64')), 'rank': Value('int64'), 'alpha': Value('int64'), 'arm': Value('string'), 'deployment': Value('string'), 'attention_implementation': 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.
reuse-vs-recraft — E16 battery results
Raw record files from the E16 scope-ladder behavioural battery
(Qwen2.5-VL-7B-Instruct, four LoRA arms, evaluated 2026-08-16):
per-arm <arm>_eval/ directories with the 75-slice repair records,
the PhD damage-panel record files (both halves), the 8x200 benchmark
suite records, and the litmus/stage logs; plus each arm's run.json
training sidecar. results_manifest.json carries sha256 for every
file in the snapshot. Uploaded incrementally while the battery ran;
the final snapshot is the one whose manifest lists every shard.
- Downloads last month
- 183