Dataset Viewer
Duplicate
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
allow_completed_dataset: bool
budget: struct<steps_per_epoch: int64, total_updates: int64, training_limit_seconds: int64, update_limit: in (... 4 chars omitted)
  child 0, steps_per_epoch: int64
  child 1, total_updates: int64
  child 2, training_limit_seconds: int64
  child 3, update_limit: int64
config: struct<architecture: string, checkpoint_epochs: list<item: int64>, crop_scale: list<item: double>, d (... 732 chars omitted)
  child 0, architecture: string
  child 1, checkpoint_epochs: list<item: int64>
      child 0, item: int64
  child 2, crop_scale: list<item: double>
      child 0, item: double
  child 3, dataset: string
  child 4, ema: list<item: double>
      child 0, item: double
  child 5, epochs: int64
  child 6, final_lr: double
  child 7, final_weight_decay: double
  child 8, global_batch_size: int64
  child 9, image_size: int64
  child 10, learning_rate: double
  child 11, mask: struct<allow_overlap: bool, aspect_ratio: list<item: double>, enc_mask_scale: list<item: double>, mi (... 78 chars omitted)
      child 0, allow_overlap: bool
      child 1, aspect_ratio: list<item: double>
          child 0, item: double
      child 2, enc_mask_scale: list<item: double>
          child 0, item: double
      child 3, min_keep: int64
      child 4, nenc: int64
      child 5, npred: int64
      child 6, pred_mask_scale: list<item: double>
          child 0, item: double
  child 12, max_train_seconds: int64
  child 13, max_updates: int64
  child 14, patch_size: int64
  chi
...
56: string
device: string
device_name: string
device_total_memory: int64
engine_sha256: string
global_batch_size: int64
mode: string
pilot_cost_separate: bool
regularizer: struct<direction: string, metric: string, ridge: double, ridge_mode: string, variance_floor: double, (... 41 chars omitted)
  child 0, direction: string
  child 1, metric: string
  child 2, ridge: double
  child 3, ridge_mode: string
  child 4, variance_floor: double
  child 5, variance_weight: double
  child 6, weight: double
regularizer_estimator: string
resume_limitations: string
run_id: string
schema_version: int64
seed: int64
source_hashes: struct<jepa_repro/__init__.py: string, jepa_repro/data.py: string, jepa_repro/model.py: string, jepa (... 247 chars omitted)
  child 0, jepa_repro/__init__.py: string
  child 1, jepa_repro/data.py: string
  child 2, jepa_repro/model.py: string
  child 3, jepa_repro/regularization.py: string
  child 4, jepa_repro/train.py: string
  child 5, vendor/ijepa/src/masks/multiblock.py: string
  child 6, vendor/ijepa/src/masks/utils.py: string
  child 7, vendor/ijepa/src/models/vision_transformer.py: string
  child 8, vendor/ijepa/src/utils/tensors.py: string
torch_version: string
world_size: int64
checkpoints: list<item: struct<epoch: int64, path: string, status: string, step: int64>>
  child 0, item: struct<epoch: int64, path: string, status: string, step: int64>
      child 0, epoch: int64
      child 1, path: string
      child 2, status: string
      child 3, step: int64
to
{'checkpoints': List({'epoch': Value('int64'), 'path': Value('string'), 'status': Value('string'), 'step': Value('int64')}), 'schema_version': Value('int64')}
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 544, 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 3052, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, 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 558, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, 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 361, 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 2381, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              allow_completed_dataset: bool
              budget: struct<steps_per_epoch: int64, total_updates: int64, training_limit_seconds: int64, update_limit: in (... 4 chars omitted)
                child 0, steps_per_epoch: int64
                child 1, total_updates: int64
                child 2, training_limit_seconds: int64
                child 3, update_limit: int64
              config: struct<architecture: string, checkpoint_epochs: list<item: int64>, crop_scale: list<item: double>, d (... 732 chars omitted)
                child 0, architecture: string
                child 1, checkpoint_epochs: list<item: int64>
                    child 0, item: int64
                child 2, crop_scale: list<item: double>
                    child 0, item: double
                child 3, dataset: string
                child 4, ema: list<item: double>
                    child 0, item: double
                child 5, epochs: int64
                child 6, final_lr: double
                child 7, final_weight_decay: double
                child 8, global_batch_size: int64
                child 9, image_size: int64
                child 10, learning_rate: double
                child 11, mask: struct<allow_overlap: bool, aspect_ratio: list<item: double>, enc_mask_scale: list<item: double>, mi (... 78 chars omitted)
                    child 0, allow_overlap: bool
                    child 1, aspect_ratio: list<item: double>
                        child 0, item: double
                    child 2, enc_mask_scale: list<item: double>
                        child 0, item: double
                    child 3, min_keep: int64
                    child 4, nenc: int64
                    child 5, npred: int64
                    child 6, pred_mask_scale: list<item: double>
                        child 0, item: double
                child 12, max_train_seconds: int64
                child 13, max_updates: int64
                child 14, patch_size: int64
                chi
              ...
              56: string
              device: string
              device_name: string
              device_total_memory: int64
              engine_sha256: string
              global_batch_size: int64
              mode: string
              pilot_cost_separate: bool
              regularizer: struct<direction: string, metric: string, ridge: double, ridge_mode: string, variance_floor: double, (... 41 chars omitted)
                child 0, direction: string
                child 1, metric: string
                child 2, ridge: double
                child 3, ridge_mode: string
                child 4, variance_floor: double
                child 5, variance_weight: double
                child 6, weight: double
              regularizer_estimator: string
              resume_limitations: string
              run_id: string
              schema_version: int64
              seed: int64
              source_hashes: struct<jepa_repro/__init__.py: string, jepa_repro/data.py: string, jepa_repro/model.py: string, jepa (... 247 chars omitted)
                child 0, jepa_repro/__init__.py: string
                child 1, jepa_repro/data.py: string
                child 2, jepa_repro/model.py: string
                child 3, jepa_repro/regularization.py: string
                child 4, jepa_repro/train.py: string
                child 5, vendor/ijepa/src/masks/multiblock.py: string
                child 6, vendor/ijepa/src/masks/utils.py: string
                child 7, vendor/ijepa/src/models/vision_transformer.py: string
                child 8, vendor/ijepa/src/utils/tensors.py: string
              torch_version: string
              world_size: int64
              checkpoints: list<item: struct<epoch: int64, path: string, status: string, step: int64>>
                child 0, item: struct<epoch: int64, path: string, status: string, step: int64>
                    child 0, epoch: int64
                    child 1, path: string
                    child 2, status: string
                    child 3, step: int64
              to
              {'checkpoints': List({'epoch': Value('int64'), 'path': Value('string'), 'status': Value('string'), 'step': Value('int64')}), 'schema_version': Value('int64')}
              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.

JEPA experiments: code, run instructions, and future checkpoint uploads

Dataset repository: Peer007/JEPA, private. This is a research-artifact repository, not a tabular dataset for load_dataset. It is separate from Peer007/NIER-reproducibility-b300.

Source: private GitHub repository YoshihiroA/jepa-repro-experiments, branch codex/tiny-milestone-finetuning, commit 047ec943b3eef9bef73a9a5ea47396c47078146a (PR #2; not necessarily GitHub main). The repo/ directory is an exact tracked-file snapshot of that commit. Packaging date: 2026-11-03, Europe/Moscow.

The source and reserved checkpoint/result directories are supplied in JEPA-source-bundle.zip. Extract it before using relative repo/ paths below. The agent runbook is also uploaded separately for convenient reading.

Start here

Source bundle SHA256: 64ea14aed22519825f4c2d5725375fff11b33b8c28c65429d9bf0c08ca0ada19.

Status

Regularized pretraining is implemented; 108 CPU tests passed, two skipped at the source commit. No CUDA pilot or new regularized training has run. Fine-tuning has a protocol/planner only: its supervised trainer and analysis adapter are not implemented. Do not represent it as runnable.

TinyImageNet weights and image data are not included. Checkpoint directories contain instructions and an expected inventory only, not fake/empty .pt files. The existing repo/results/ holds compact historical baseline results. New run artifacts belong under results/new_runs/; do not overwrite the baseline. The original frozen hierarchy study has unresolved data/provenance prerequisites; read the local readiness document before scheduling new training.

Licensing and privacy

Preserve repo/THIRD_PARTY.md and the vendored repo/vendor/ijepa/LICENSE inside the archive. This upload does not grant a new license to the whole repository or to TinyImageNet data. Obtain images under their own terms. No credentials, raw datasets or checkpoint binaries are in this source package. Keep the dataset private unless the owner explicitly authorizes publication.

Downloads last month
46