Datasets:

j1s4nn
/

jevgag-selfaware-dit

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
n_segments: int64
n_blocks: int64
dim: int64
n_positive: int64
n_prompts: int64
base_rate: double
block_sweep_prefill: list<item: double>
  child 0, item: double
block_sweep_gen: list<item: double>
  child 0, item: double
best_prefill_block: int64
best_prefill_auroc: double
best_gen_block: int64
best_gen_auroc: double
concat_prefill_auroc: double
concat_prefill_ci: list<item: double>
  child 0, item: double
concat_prefill_perm_p: double
best_prefill_ci: list<item: double>
  child 0, item: double
best_prefill_perm_p: double
reference_points: struct<hand_crafted_context_signal_base_paper: double, jev_text_oracle_base_paper: double, chance: d (... 6 chars omitted)
  child 0, hand_crafted_context_signal_base_paper: double
  child 1, jev_text_oracle_base_paper: double
  child 2, chance: double
status: string
dataset_paths: struct<fine: string, reference: string, coarse: string, vae: string, t5: string, tokenizer: string>
  child 0, fine: string
  child 1, reference: string
  child 2, coarse: string
  child 3, vae: string
  child 4, t5: string
  child 5, tokenizer: string
n_positive_latent: int64
upstream_rcm_commit: string
n_positive_smooth: int64
block_names: list<item: string>
  child 0, item: string
records: list<item: struct<prompt_idx: int64, stratum: string, prompt: string, segment: int64, l1_fine: doubl (... 331 chars omitted)
  child 0, item: struct<prompt_idx: int64, stratum: string, prompt: string, segment: int64, l1_fine: double, l1_coars (... 319 chars omitted)
      c
...
      child 4, l1_fine: double
      child 5, l1_coarse: double
      child 6, label_coarse_unsafe: int64
      child 7, ll1_fine: double
      child 8, ll1_coarse: double
      child 9, label_latent: int64
      child 10, jitter_fine: double
      child 11, jitter_coarse: double
      child 12, label_smooth: int64
      child 13, lj_fine: double
      child 14, lj_coarse: double
      child 15, label_latent_smooth: int64
      child 16, n_fine_calls: int64
      child 17, n_coarse_calls: int64
      child 18, prefill_vnorm: double
      child 19, prefill_vabsmean: double
kernel: string
run_id: string
latent: struct<frames: int64, h: int64, w: int64, c: int64>
  child 0, frames: int64
  child 1, h: int64
  child 2, w: int64
  child 3, c: int64
feature_shape: list<item: int64>
  child 0, item: int64
elapsed_s: double
gpu: string
timing: struct<deps_s: double, clone_s: double, encode_s: double>
  child 0, deps_s: double
  child 1, clone_s: double
  child 2, encode_s: double
protocol: struct<seed: int64, horizon: int64, limit: int64, n_ref_steps: int64, n_fine_steps: int64, n_coarse_ (... 135 chars omitted)
  child 0, seed: int64
  child 1, horizon: int64
  child 2, limit: int64
  child 3, n_ref_steps: int64
  child 4, n_fine_steps: int64
  child 5, n_coarse_steps: int64
  child 6, fine_ckpt: string
  child 7, coarse_ckpt: string
  child 8, ref_ckpt: string
  child 9, label: string
  child 10, label_space: string
  child 11, labels: list<item: string>
      child 0, item: string
to
{'run_id': Value('string'), 'kernel': Value('string'), 'protocol': {'seed': Value('int64'), 'horizon': Value('int64'), 'limit': Value('int64'), 'n_ref_steps': Value('int64'), 'n_fine_steps': Value('int64'), 'n_coarse_steps': Value('int64'), 'fine_ckpt': Value('string'), 'coarse_ckpt': Value('string'), 'ref_ckpt': Value('string'), 'label': Value('string'), 'label_space': Value('string'), 'labels': List(Value('string'))}, 'records': List({'prompt_idx': Value('int64'), 'stratum': Value('string'), 'prompt': Value('string'), 'segment': Value('int64'), 'l1_fine': Value('float64'), 'l1_coarse': Value('float64'), 'label_coarse_unsafe': Value('int64'), 'll1_fine': Value('float64'), 'll1_coarse': Value('float64'), 'label_latent': Value('int64'), 'jitter_fine': Value('float64'), 'jitter_coarse': Value('float64'), 'label_smooth': Value('int64'), 'lj_fine': Value('float64'), 'lj_coarse': Value('float64'), 'label_latent_smooth': Value('int64'), 'n_fine_calls': Value('int64'), 'n_coarse_calls': Value('int64'), 'prefill_vnorm': Value('float64'), 'prefill_vabsmean': Value('float64')}), 'status': Value('string'), 'gpu': Value('string'), 'upstream_rcm_commit': Value('string'), 'dataset_paths': {'fine': Value('string'), 'reference': Value('string'), 'coarse': Value('string'), 'vae': Value('string'), 't5': Value('string'), 'tokenizer': Value('string')}, 'latent': {'frames': Value('int64'), 'h': Value('int64'), 'w': Value('int64'), 'c': Value('int64')}, 'block_names': List(Value('string')), 'n_blocks': Value('int64'), 'n_segments': Value('int64'), 'n_positive': Value('int64'), 'n_positive_latent': Value('int64'), 'n_positive_smooth': Value('int64'), 'feature_shape': List(Value('int64')), 'timing': {'deps_s': Value('float64'), 'clone_s': Value('float64'), 'encode_s': Value('float64')}, 'elapsed_s': Value('float64')}
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
              n_segments: int64
              n_blocks: int64
              dim: int64
              n_positive: int64
              n_prompts: int64
              base_rate: double
              block_sweep_prefill: list<item: double>
                child 0, item: double
              block_sweep_gen: list<item: double>
                child 0, item: double
              best_prefill_block: int64
              best_prefill_auroc: double
              best_gen_block: int64
              best_gen_auroc: double
              concat_prefill_auroc: double
              concat_prefill_ci: list<item: double>
                child 0, item: double
              concat_prefill_perm_p: double
              best_prefill_ci: list<item: double>
                child 0, item: double
              best_prefill_perm_p: double
              reference_points: struct<hand_crafted_context_signal_base_paper: double, jev_text_oracle_base_paper: double, chance: d (... 6 chars omitted)
                child 0, hand_crafted_context_signal_base_paper: double
                child 1, jev_text_oracle_base_paper: double
                child 2, chance: double
              status: string
              dataset_paths: struct<fine: string, reference: string, coarse: string, vae: string, t5: string, tokenizer: string>
                child 0, fine: string
                child 1, reference: string
                child 2, coarse: string
                child 3, vae: string
                child 4, t5: string
                child 5, tokenizer: string
              n_positive_latent: int64
              upstream_rcm_commit: string
              n_positive_smooth: int64
              block_names: list<item: string>
                child 0, item: string
              records: list<item: struct<prompt_idx: int64, stratum: string, prompt: string, segment: int64, l1_fine: doubl (... 331 chars omitted)
                child 0, item: struct<prompt_idx: int64, stratum: string, prompt: string, segment: int64, l1_fine: double, l1_coars (... 319 chars omitted)
                    c
              ...
                    child 4, l1_fine: double
                    child 5, l1_coarse: double
                    child 6, label_coarse_unsafe: int64
                    child 7, ll1_fine: double
                    child 8, ll1_coarse: double
                    child 9, label_latent: int64
                    child 10, jitter_fine: double
                    child 11, jitter_coarse: double
                    child 12, label_smooth: int64
                    child 13, lj_fine: double
                    child 14, lj_coarse: double
                    child 15, label_latent_smooth: int64
                    child 16, n_fine_calls: int64
                    child 17, n_coarse_calls: int64
                    child 18, prefill_vnorm: double
                    child 19, prefill_vabsmean: double
              kernel: string
              run_id: string
              latent: struct<frames: int64, h: int64, w: int64, c: int64>
                child 0, frames: int64
                child 1, h: int64
                child 2, w: int64
                child 3, c: int64
              feature_shape: list<item: int64>
                child 0, item: int64
              elapsed_s: double
              gpu: string
              timing: struct<deps_s: double, clone_s: double, encode_s: double>
                child 0, deps_s: double
                child 1, clone_s: double
                child 2, encode_s: double
              protocol: struct<seed: int64, horizon: int64, limit: int64, n_ref_steps: int64, n_fine_steps: int64, n_coarse_ (... 135 chars omitted)
                child 0, seed: int64
                child 1, horizon: int64
                child 2, limit: int64
                child 3, n_ref_steps: int64
                child 4, n_fine_steps: int64
                child 5, n_coarse_steps: int64
                child 6, fine_ckpt: string
                child 7, coarse_ckpt: string
                child 8, ref_ckpt: string
                child 9, label: string
                child 10, label_space: string
                child 11, labels: list<item: string>
                    child 0, item: string
              to
              {'run_id': Value('string'), 'kernel': Value('string'), 'protocol': {'seed': Value('int64'), 'horizon': Value('int64'), 'limit': Value('int64'), 'n_ref_steps': Value('int64'), 'n_fine_steps': Value('int64'), 'n_coarse_steps': Value('int64'), 'fine_ckpt': Value('string'), 'coarse_ckpt': Value('string'), 'ref_ckpt': Value('string'), 'label': Value('string'), 'label_space': Value('string'), 'labels': List(Value('string'))}, 'records': List({'prompt_idx': Value('int64'), 'stratum': Value('string'), 'prompt': Value('string'), 'segment': Value('int64'), 'l1_fine': Value('float64'), 'l1_coarse': Value('float64'), 'label_coarse_unsafe': Value('int64'), 'll1_fine': Value('float64'), 'll1_coarse': Value('float64'), 'label_latent': Value('int64'), 'jitter_fine': Value('float64'), 'jitter_coarse': Value('float64'), 'label_smooth': Value('int64'), 'lj_fine': Value('float64'), 'lj_coarse': Value('float64'), 'label_latent_smooth': Value('int64'), 'n_fine_calls': Value('int64'), 'n_coarse_calls': Value('int64'), 'prefill_vnorm': Value('float64'), 'prefill_vabsmean': Value('float64')}), 'status': Value('string'), 'gpu': Value('string'), 'upstream_rcm_commit': Value('string'), 'dataset_paths': {'fine': Value('string'), 'reference': Value('string'), 'coarse': Value('string'), 'vae': Value('string'), 't5': Value('string'), 'tokenizer': Value('string')}, 'latent': {'frames': Value('int64'), 'h': Value('int64'), 'w': Value('int64'), 'c': Value('int64')}, 'block_names': List(Value('string')), 'n_blocks': Value('int64'), 'n_segments': Value('int64'), 'n_positive': Value('int64'), 'n_positive_latent': Value('int64'), 'n_positive_smooth': Value('int64'), 'feature_shape': List(Value('int64')), 'timing': {'deps_s': Value('float64'), 'clone_s': Value('float64'), 'encode_s': Value('float64')}, 'elapsed_s': Value('float64')}
              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.

JevGAG SelfAware-DiT Intrinsic Probes

Intrinsic DiT activation probes for pre-generation granularity routing

JevGAG SelfAware-DiT Intrinsic Probes

Captured Causal-rCM diffusion-transformer (DiT) activations and probe results for SelfAware-DiT. An intrinsic pre-generation probe on the DiT's own block activations predicts coarse-unsafe temporal segments at AUROC 0.798 (block 11), versus 0.491 for hand-crafted context signals and 0.788 for the external Jev text oracle.

This supports adaptive temporal granularity routing in causal video generation and video world models without an external oracle.

features.npz: prefill/generation pooled activations, labels, groups
manifest.json: protocol and per-segment records
probe_results.json: layer sweep, confidence intervals, permutation nulls
fig_probe_layer_sweep.png: probe AUROC by layer
DATASET_NOTES.md: original experiment notes

Related resources

Code (JevGAG programme): https://github.com/j1s4nn/JevGAG
Code (Jev-Routed Temporal Granularity): https://github.com/j1s4nn/jev-adaptive-granularity-WM
Project page (JevGAG): https://j1s4nn.github.io/JevGAG/
Project page (Paper 1): https://j1s4nn.github.io/jev-adaptive-granularity-WM/
Portfolio: https://j1s4nn.github.io
Source (Kaggle): https://www.kaggle.com/datasets/ajjisan/jevgag-selfaware-dit
Preprint (JevGAG): https://doi.org/10.5281/zenodo.23693517
Preprint (Jev-Routed Temporal Granularity): https://doi.org/10.5281/zenodo.23395414

Citation

@misc{hossen_jevgag_ddag,
  author       = {Hossen, MD Jisan},
  title        = {Regime-Aware Temporal Granularity for Causal Video World Models: When Does Adaptivity Pay?},
  year         = {2026},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.23693517},
  url          = {https://doi.org/10.5281/zenodo.23693517}
}