Datasets:

sanxing
/

ai4sci-fruit-fly-brain

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
schema: string
frozen_date: timestamp[s]
selection: string
regions: list<item: struct<split: string, region_id: string, prepared_relative_path: string, center_xyz_16nm: (... 79 chars omitted)
  child 0, item: struct<split: string, region_id: string, prepared_relative_path: string, center_xyz_16nm: list<item: (... 67 chars omitted)
      child 0, split: string
      child 1, region_id: string
      child 2, prepared_relative_path: string
      child 3, center_xyz_16nm: list<item: int64>
          child 0, item: int64
      child 4, image_manifest_sha256: string
      child 5, core_labels_sha256: string
metadata: struct<em_s1: struct<url: string, sha256: string>, segmentation: struct<url: string, sha256: string> (... 54 chars omitted)
  child 0, em_s1: struct<url: string, sha256: string>
      child 0, url: string
      child 1, sha256: string
  child 1, segmentation: struct<url: string, sha256: string>
      child 0, url: string
      child 1, sha256: string
  child 2, official_scene: struct<url: string, sha256: string>
      child 0, url: string
      child 1, sha256: string
segmentation_url: string
training: string
context: string
test_reporting: string
primary_baseline: string
public_number_check: string
historical_pilot_excluded: bool
paper_comparable: bool
execution: string
statistics: string
empty_region_policy: string
context_halo_nm: int64
source: struct<schema: string, dataset: string, center_xyz_16nm: list<item: int64>, core_size_voxels: int64, (... 328 chars omitted)
  child 0, schema: string
  child 1, dataset: string
  child 2, center_xyz_16nm: list<item: int64>
      child 0, item: int64
  child 3, core_size_voxels: int64
  child 4, selection: string
  child 5, em_url: string
  child 6, segmentation_url: string
  child 7, metadata: struct<em_s1: struct<url: string, sha256: string>, segmentation: struct<url: string, sha256: string> (... 54 chars omitted)
      child 0, em_s1: struct<url: string, sha256: string>
          child 0, url: string
          child 1, sha256: string
      child 1, segmentation: struct<url: string, sha256: string>
          child 0, url: string
          child 1, sha256: string
      child 2, official_scene: struct<url: string, sha256: string>
          child 0, url: string
          child 1, sha256: string
  child 8, model_package_manifest_sha256: string
  child 9, training: string
  child 10, comparison: string
  child 11, paper_comparable: bool
ffn_input_padding_voxels: int64
pilot_excluded: bool
training_exposure: string
status: string
nested_scale_voxels: list<item: int64>
  child 0, item: int64
scale_reference_policy: string
to
{'schema': Value('string'), 'status': Value('string'), 'selection': Value('string'), 'pilot_excluded': Value('bool'), 'source': {'schema': Value('string'), 'dataset': Value('string'), 'center_xyz_16nm': List(Value('int64')), 'core_size_voxels': Value('int64'), 'selection': Value('string'), 'em_url': Value('string'), 'segmentation_url': Value('string'), 'metadata': {'em_s1': {'url': Value('string'), 'sha256': Value('string')}, 'segmentation': {'url': Value('string'), 'sha256': Value('string')}, 'official_scene': {'url': Value('string'), 'sha256': Value('string')}}, 'model_package_manifest_sha256': Value('string'), 'training': Value('string'), 'comparison': Value('string'), 'paper_comparable': Value('bool')}, 'regions': List({'region_id': Value('string'), 'center_xyz_16nm': List(Value('int64')), 'core_size_voxels': Value('int64')}), 'context_halo_nm': Value('int64'), 'ffn_input_padding_voxels': Value('int64'), 'empty_region_policy': Value('string'), 'nested_scale_voxels': List(Value('int64')), 'scale_reference_policy': Value('string'), 'statistics': Value('string'), 'training_exposure': Value('string'), 'execution': 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
              schema: string
              frozen_date: timestamp[s]
              selection: string
              regions: list<item: struct<split: string, region_id: string, prepared_relative_path: string, center_xyz_16nm: (... 79 chars omitted)
                child 0, item: struct<split: string, region_id: string, prepared_relative_path: string, center_xyz_16nm: list<item: (... 67 chars omitted)
                    child 0, split: string
                    child 1, region_id: string
                    child 2, prepared_relative_path: string
                    child 3, center_xyz_16nm: list<item: int64>
                        child 0, item: int64
                    child 4, image_manifest_sha256: string
                    child 5, core_labels_sha256: string
              metadata: struct<em_s1: struct<url: string, sha256: string>, segmentation: struct<url: string, sha256: string> (... 54 chars omitted)
                child 0, em_s1: struct<url: string, sha256: string>
                    child 0, url: string
                    child 1, sha256: string
                child 1, segmentation: struct<url: string, sha256: string>
                    child 0, url: string
                    child 1, sha256: string
                child 2, official_scene: struct<url: string, sha256: string>
                    child 0, url: string
                    child 1, sha256: string
              segmentation_url: string
              training: string
              context: string
              test_reporting: string
              primary_baseline: string
              public_number_check: string
              historical_pilot_excluded: bool
              paper_comparable: bool
              execution: string
              statistics: string
              empty_region_policy: string
              context_halo_nm: int64
              source: struct<schema: string, dataset: string, center_xyz_16nm: list<item: int64>, core_size_voxels: int64, (... 328 chars omitted)
                child 0, schema: string
                child 1, dataset: string
                child 2, center_xyz_16nm: list<item: int64>
                    child 0, item: int64
                child 3, core_size_voxels: int64
                child 4, selection: string
                child 5, em_url: string
                child 6, segmentation_url: string
                child 7, metadata: struct<em_s1: struct<url: string, sha256: string>, segmentation: struct<url: string, sha256: string> (... 54 chars omitted)
                    child 0, em_s1: struct<url: string, sha256: string>
                        child 0, url: string
                        child 1, sha256: string
                    child 1, segmentation: struct<url: string, sha256: string>
                        child 0, url: string
                        child 1, sha256: string
                    child 2, official_scene: struct<url: string, sha256: string>
                        child 0, url: string
                        child 1, sha256: string
                child 8, model_package_manifest_sha256: string
                child 9, training: string
                child 10, comparison: string
                child 11, paper_comparable: bool
              ffn_input_padding_voxels: int64
              pilot_excluded: bool
              training_exposure: string
              status: string
              nested_scale_voxels: list<item: int64>
                child 0, item: int64
              scale_reference_policy: string
              to
              {'schema': Value('string'), 'status': Value('string'), 'selection': Value('string'), 'pilot_excluded': Value('bool'), 'source': {'schema': Value('string'), 'dataset': Value('string'), 'center_xyz_16nm': List(Value('int64')), 'core_size_voxels': Value('int64'), 'selection': Value('string'), 'em_url': Value('string'), 'segmentation_url': Value('string'), 'metadata': {'em_s1': {'url': Value('string'), 'sha256': Value('string')}, 'segmentation': {'url': Value('string'), 'sha256': Value('string')}, 'official_scene': {'url': Value('string'), 'sha256': Value('string')}}, 'model_package_manifest_sha256': Value('string'), 'training': Value('string'), 'comparison': Value('string'), 'paper_comparable': Value('bool')}, 'regions': List({'region_id': Value('string'), 'center_xyz_16nm': List(Value('int64')), 'core_size_voxels': Value('int64')}), 'context_halo_nm': Value('int64'), 'ffn_input_padding_voxels': Value('int64'), 'empty_region_policy': Value('string'), 'nested_scale_voxels': List(Value('int64')), 'scale_reference_policy': Value('string'), 'statistics': Value('string'), 'training_exposure': Value('string'), 'execution': 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.

AI4Sci Fruit Fly Brain prepared resources

Prepared MaleCNS v1.0 resources for the experimental single-checkpoint FFN / PATHFINDER track in ai4sci-tasks.

This is a maintainer-produced derivative of public source data. It contains 722 logical files (14,537,257,535 bytes), stored as SHA-256-addressed objects under blobs/. release.json maps original relative paths to object hashes and byte counts. It includes no trained weights, model predictions, experiment logs or budget ledgers.

Contents and visibility

development/training: cropped 16nm EM and segmentation-derived volume supervision for spatial training and validation, plus SHAPE mesh examples including synthetic negatives. Shared observed public neuron IDs are excluded from training according to the pinned spatial plan.
development/validation: validation EM, sparse skeleton reference and evaluator.
development/runtime: pinned Google FFN and connectomics source checkouts (including version metadata), PointNeXt FPS sources/licenses and a compiled FPS extension. The binary was built for the native Python 3.11 / Torch 2.6 / A5000 environment; target-container ABI compatibility is a separate acceptance gate.
private/image and private/suite: frozen r3 evaluation EM and sparse scoring reference. “Private” describes the evaluation container mount boundary. These public-source evaluation resources are downloadable by operators; never mount this tree or the download cache into the development agent.

The development manifests contain only training/validation splits. The test evaluates a 512-cubed core inside a 1762-cubed image. Skeletons are locally derived from segmentation, not the official full-connectome skeleton release. Sparse annotations, omitted nodes and local metrics limit comparisons with published dense evaluations.

Download

Use fruit-fly-brain/environment/data/materialize.sh from the linked task repository. Its checked-in source metadata pins an immutable HF commit and the release manifest SHA-256. Downloading is public and requires no token. The operator materializer verifies all file sizes and SHA-256 values and writes disjoint development/private roots. Downloads resume through the HF cache.

Attribution and licenses

Source: MaleCNS official downloads, v1.0, by FlyEM (HHMI Janelia), University of Cambridge, MRC Laboratory of Molecular Biology and Google Research. Data is CC BY 4.0, as stated on the source site. Cite Sexual dimorphism in the complete Drosophila male central nervous system connectome, DOI. No endorsement by the source authors is implied.

Changes: selected spatial crops, shared-ID exclusions, derived sparse skeletons, training targets and synthetic-negative mesh pairs, and repackaging. Source endpoints, frozen regions and exclusions are retained under provenance/ and in resource manifests. This release packages the already prepared campaign data; it does not establish a new clean download from upstream.

Software retains its own licenses, rather than inheriting the data license: Google FFN/connectomics and PointNeXt/OpenPoints license texts are retained both in the resource inventory and under licenses/. Included third-party notices remain in the source trees.

Method reference: Accelerating Neuron Reconstruction with PATHFINDER, DOI. This experimental adaptation is not an author release or a paper-parity claim. Full training and the complete GPU container verifier remain unverified.

Downloads last month: 48

Total file size:

14.5 GB