Datasets:

haohw
/

tacwam-tactile-mot-v2-splits

Tags:

License:

Dataset card Files Files and versions

xet

Community

Dataset Viewer

Duplicate

Split (1)

train

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
roots_relative: struct<0021b7c0-b812-4ce5-9f08-31573c06977f: string, 0022159d-d641-471d-8df3-f1639f263bc3: string, 0 (... 97978 chars omitted)
  child 0, 0021b7c0-b812-4ce5-9f08-31573c06977f: string
  child 1, 0022159d-d641-471d-8df3-f1639f263bc3: string
  child 2, 003b1411-30c1-4012-b454-8875851bc1be: string
  child 3, 00526270-32ae-450f-beba-7ac4c62385ce: string
  child 4, 0060ab69-9f17-47e5-9dd1-51f243d23ce7: string
  child 5, 006f8653-1e15-49c0-bc94-f31e5b935eea: string
  child 6, 0074ed4f-de2e-47c5-bf05-54b6183c887b: string
  child 7, 00b1e23f-4ff4-4ef8-be14-52769824da7a: string
  child 8, 00b901a5-15ba-44bd-9b50-6800efb3aaa6: string
  child 9, 00dbb363-d357-48c4-a3f3-abfb3fd23852: string
  child 10, 00fa894a-dcc8-4295-8305-eeba8474f444: string
  child 11, 01567e34-7798-4f71-aa75-8bd3bf8e6767: string
  child 12, 015dc9e9-3156-4fe2-83a5-52b77505370a: string
  child 13, 018d4a22-d0b6-44c7-868a-4e9ad30ad7cd: string
  child 14, 01b49d62-5739-4381-ae11-031a3ff69ab9: string
  child 15, 01d646e6-ef70-4795-bdff-af4cd7d0a30f: string
  child 16, 0205af38-a8b2-4b20-8bd5-4e3ca6ea4769: string
  child 17, 02311576-e758-4a81-bbf6-f8cfa22f9acc: string
  child 18, 0235de47-10e1-41ec-9d6b-6e11e985538f: string
  child 19, 0237392d-2004-4716-9cd4-3f9d7413d3d7: string
  child 20, 023f944a-f288-4ecb-acbe-40293590a6a1: string
  child 21, 0253bfbf-5cfc-48c0-b907-b494a6e26c0f: string
  child 22, 029d2754-4d36-43ce-a2c0-cdda830fbd7a: string
  child 23, 0320f378-8eb5-4d37-9685-fa50d80177e3: string
...
7-4926-9c19-486945f05d12: string
  child 2121, febb62d0-7b3a-4aaf-bfa7-79392b5f227e: string
  child 2122, fed8cd23-b699-4874-a495-eb95ff9e3b7d: string
  child 2123, ff041291-66f7-4635-87ec-36bd1c54e566: string
  child 2124, ff16d75f-bad7-4ccc-9f5b-8ec0d44153e4: string
  child 2125, ff6a4df4-012a-4c39-a81c-ea139e120458: string
  child 2126, ff6e30ed-ca73-415b-b186-c63d626a351d: string
  child 2127, ff6e9204-ed1b-4f62-8088-c37d5808e68d: string
  child 2128, ff749260-452a-4fc3-bacf-a5c1d40a6a26: string
  child 2129, ff88eeaf-c16c-453c-bbab-575be52602a6: string
  child 2130, ffd0ba04-fc18-4430-86e6-dddc6ce62cfe: string
  child 2131, ffe0923e-e5b3-4f2a-aa04-359e1553b282: string
feature_spec_id: string
normalization_hash: string
statistics: struct<gate_delta_mad: double, gate_delta_median: double, gate_k: double, per_region_baseline: list< (... 136 chars omitted)
  child 0, gate_delta_mad: double
  child 1, gate_delta_median: double
  child 2, gate_k: double
  child 3, per_region_baseline: list<item: double>
      child 0, item: double
  child 4, per_region_contact_threshold: list<item: double>
      child 0, item: double
  child 5, per_region_scale: list<item: double>
      child 0, item: double
  child 6, region_names: list<item: string>
      child 0, item: string
split_manifest_hash: string
formula: string
schema: string
fit_split: string
source_field: string
statistics_hash: string
fit_recording_count: int64
schema_version: string
code_revision: string
normalization_id: string
to
{'code_revision': Value('string'), 'feature_spec_id': Value('string'), 'fit_recording_count': Value('int64'), 'fit_split': Value('string'), 'formula': Value('string'), 'normalization_hash': Value('string'), 'normalization_id': Value('string'), 'schema': Value('string'), 'schema_version': Value('string'), 'source_field': Value('string'), 'split_manifest_hash': Value('string'), 'statistics': {'gate_delta_mad': Value('float64'), 'gate_delta_median': Value('float64'), 'gate_k': Value('float64'), 'per_region_baseline': List(Value('float64')), 'per_region_contact_threshold': List(Value('float64')), 'per_region_scale': List(Value('float64')), 'region_names': List(Value('string'))}, 'statistics_hash': 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
              roots_relative: struct<0021b7c0-b812-4ce5-9f08-31573c06977f: string, 0022159d-d641-471d-8df3-f1639f263bc3: string, 0 (... 97978 chars omitted)
                child 0, 0021b7c0-b812-4ce5-9f08-31573c06977f: string
                child 1, 0022159d-d641-471d-8df3-f1639f263bc3: string
                child 2, 003b1411-30c1-4012-b454-8875851bc1be: string
                child 3, 00526270-32ae-450f-beba-7ac4c62385ce: string
                child 4, 0060ab69-9f17-47e5-9dd1-51f243d23ce7: string
                child 5, 006f8653-1e15-49c0-bc94-f31e5b935eea: string
                child 6, 0074ed4f-de2e-47c5-bf05-54b6183c887b: string
                child 7, 00b1e23f-4ff4-4ef8-be14-52769824da7a: string
                child 8, 00b901a5-15ba-44bd-9b50-6800efb3aaa6: string
                child 9, 00dbb363-d357-48c4-a3f3-abfb3fd23852: string
                child 10, 00fa894a-dcc8-4295-8305-eeba8474f444: string
                child 11, 01567e34-7798-4f71-aa75-8bd3bf8e6767: string
                child 12, 015dc9e9-3156-4fe2-83a5-52b77505370a: string
                child 13, 018d4a22-d0b6-44c7-868a-4e9ad30ad7cd: string
                child 14, 01b49d62-5739-4381-ae11-031a3ff69ab9: string
                child 15, 01d646e6-ef70-4795-bdff-af4cd7d0a30f: string
                child 16, 0205af38-a8b2-4b20-8bd5-4e3ca6ea4769: string
                child 17, 02311576-e758-4a81-bbf6-f8cfa22f9acc: string
                child 18, 0235de47-10e1-41ec-9d6b-6e11e985538f: string
                child 19, 0237392d-2004-4716-9cd4-3f9d7413d3d7: string
                child 20, 023f944a-f288-4ecb-acbe-40293590a6a1: string
                child 21, 0253bfbf-5cfc-48c0-b907-b494a6e26c0f: string
                child 22, 029d2754-4d36-43ce-a2c0-cdda830fbd7a: string
                child 23, 0320f378-8eb5-4d37-9685-fa50d80177e3: string
              ...
              7-4926-9c19-486945f05d12: string
                child 2121, febb62d0-7b3a-4aaf-bfa7-79392b5f227e: string
                child 2122, fed8cd23-b699-4874-a495-eb95ff9e3b7d: string
                child 2123, ff041291-66f7-4635-87ec-36bd1c54e566: string
                child 2124, ff16d75f-bad7-4ccc-9f5b-8ec0d44153e4: string
                child 2125, ff6a4df4-012a-4c39-a81c-ea139e120458: string
                child 2126, ff6e30ed-ca73-415b-b186-c63d626a351d: string
                child 2127, ff6e9204-ed1b-4f62-8088-c37d5808e68d: string
                child 2128, ff749260-452a-4fc3-bacf-a5c1d40a6a26: string
                child 2129, ff88eeaf-c16c-453c-bbab-575be52602a6: string
                child 2130, ffd0ba04-fc18-4430-86e6-dddc6ce62cfe: string
                child 2131, ffe0923e-e5b3-4f2a-aa04-359e1553b282: string
              feature_spec_id: string
              normalization_hash: string
              statistics: struct<gate_delta_mad: double, gate_delta_median: double, gate_k: double, per_region_baseline: list< (... 136 chars omitted)
                child 0, gate_delta_mad: double
                child 1, gate_delta_median: double
                child 2, gate_k: double
                child 3, per_region_baseline: list<item: double>
                    child 0, item: double
                child 4, per_region_contact_threshold: list<item: double>
                    child 0, item: double
                child 5, per_region_scale: list<item: double>
                    child 0, item: double
                child 6, region_names: list<item: string>
                    child 0, item: string
              split_manifest_hash: string
              formula: string
              schema: string
              fit_split: string
              source_field: string
              statistics_hash: string
              fit_recording_count: int64
              schema_version: string
              code_revision: string
              normalization_id: string
              to
              {'code_revision': Value('string'), 'feature_spec_id': Value('string'), 'fit_recording_count': Value('int64'), 'fit_split': Value('string'), 'formula': Value('string'), 'normalization_hash': Value('string'), 'normalization_id': Value('string'), 'schema': Value('string'), 'schema_version': Value('string'), 'source_field': Value('string'), 'split_manifest_hash': Value('string'), 'statistics': {'gate_delta_mad': Value('float64'), 'gate_delta_median': Value('float64'), 'gate_k': Value('float64'), 'per_region_baseline': List(Value('float64')), 'per_region_contact_threshold': List(Value('float64')), 'per_region_scale': List(Value('float64')), 'region_names': List(Value('string'))}, 'statistics_hash': Value('string')}
              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.

tacWAM Tactile-MoT v2 — split manifests (train / validation / test)

Partition + normalization manifests for the tacWAM Cosmos-3 video+tactile joint-prediction corpus (v2, combined ~30 h). This repo lets an evaluator know exactly which recordings are train / validation / test and how tactile is normalized — so eval is done on the held-out split with the same preprocessing as training.

⚠️ The raw data is NOT included here. The recordings are egocentric human-subject video + tactile from a commercial vendor (Tujian ITW) and are not redistributed. This repo contains only recording UUIDs, the split assignment, normalization statistics, and English task prompts — no pixels, no tactile signals, no personal data.

File	What it is
`recording_split_v2.json`	2132 records `{recording_uuid, split}`; 80/10/10 stratified split (`train` 1697 / `validation` 221 / `test` 214). Includes `seed`, `strategy`, `stratification_keys`, `manifest_hash`.
`normalization_v2.json`	Train-only tactile normalization stats (per-channel), used to derive features. Apply identically at eval.
`recording_roots_v2_relative.json`	`recording_uuid → "<batch>/<uuid>"` (e.g. `qc_20260723/itw07-23/<uuid>`) — locate each recording within your copy of the vendor delivery.
`task_translations.json`	116 unique task descriptions, Chinese → English (`prompt_en`), keyed by the Chinese `name`. Used for the model's English text conditioning. No human data.

Split summary

train: 1697 recordings
validation: 221 recordings
test: 214 recordings
fractions {train: 0.8, validation: 0.1, test: 0.1}, stratified; manifest_id: tujian_combined_recording_split_v2.
3 scenes: kitchen prep station, lab equipment & assembly workbench, tea-break refreshment station.

How to use for evaluation

Obtain the raw recordings for the validation/test UUIDs from the source vendor delivery (the <batch>/<uuid> layout in recording_roots_v2_relative.json mirrors the delivery).
Clone the model code: github.com/haohww/tacWAM (tacwam/schema_v03/ + tacwam/cosmos_tactile/).
Build the model view on the fly with tacwam.cosmos_tactile.cosmos_v1_data.CosmosV1WindowDataset, passing normalization_v2.json (loaded via tacwam.schema_v03.load_normalization) so features match training. Filter recordings by the validation/test split from recording_split_v2.json.
Evaluate the checkpoint (see the companion model repo haohw/cosmos3-tactile-mot-v2-step250).

Provenance

Corpus: combined itw07-23 + itw07-25 + itw07-27 vendor batches (QC-passed), 2132 recordings.
Master grid 30 Hz; slow 10 Hz (tick k owns 30 Hz samples [3k,3k+1,3k+2]); fast 30 Hz.
Companion model: haohw/cosmos3-tactile-mot-v2-step250.

Downloads last month: 28

Total file size:

484 kB

tacWAM Tactile-MoT v2 — split manifests (train / validation / test)

Contents

Split summary

How to use for evaluation

Provenance