Datasets:

BlueSourceJY
/

ERCBench-CineTrans-400

Languages:

English

Tags:

video-generation

cinetrans

eec-bench

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
schema_version: string
dimension_profile: string
reference_text: string
shot_ids: list<item: string>
  child 0, item: string
narrative: struct<facts: list<item: struct<fact_id: string, text: string>>>
  child 0, facts: list<item: struct<fact_id: string, text: string>>
      child 0, item: struct<fact_id: string, text: string>
          child 0, fact_id: string
          child 1, text: string
registry: struct<schema_version: string, shot_ids: list<item: string>, entities: list<item: struct<prompt_enti (... 197 chars omitted)
  child 0, schema_version: string
  child 1, shot_ids: list<item: string>
      child 0, item: string
  child 2, entities: list<item: struct<prompt_entity_id: string, name: string, description: string, evidence: list<item:  (... 125 chars omitted)
      child 0, item: struct<prompt_entity_id: string, name: string, description: string, evidence: list<item: struct<fact (... 113 chars omitted)
          child 0, prompt_entity_id: string
          child 1, name: string
          child 2, description: string
          child 3, evidence: list<item: struct<fact_id: string, quote: string>>
              child 0, item: struct<fact_id: string, quote: string>
                  child 0, fact_id: string
                  child 1, quote: string
          child 4, schedule: list<item: struct<shot_id: string, expectation: string, reason: string>>
              child 0, item: struct<shot_id: string, expectation: string, reason: string>
                  child 0, shot_id: s
...
, target_duration_seconds: double
      child 3, prompt: string
      child 4, prompt_sha256: string
source_snapshot_revision: string
sample_solver: string
filename: string
sample_steps: int64
validation: struct<status: string, codec: string, duration_seconds: double, fps: int64, width: int64, height: in (... 72 chars omitted)
  child 0, status: string
  child 1, codec: string
  child 2, duration_seconds: double
  child 3, fps: int64
  child 4, width: int64
  child 5, height: int64
  child 6, frame_count: int64
  child 7, full_file_decode: string
  child 8, visual_audit: string
source_generation_priority: int64
offload_model: bool
source_contract_sha256: string
source_episode_dir: string
width: int64
shot_count: int64
prompt_repo: string
target_duration_seconds: double
generated_at_utc: string
adapter: string
sample_shift: int64
prompt_path: string
guide_scale: int64
task: string
general_prompt: bool
generated_wall_seconds: double
frame_count: int64
mask: struct<mask_p: int64, start_masked_layer: int64, end_masked_layer: int64, mask_info: list<item: int6 (... 3 chars omitted)
  child 0, mask_p: int64
  child 1, start_masked_layer: int64
  child 2, end_masked_layer: int64
  child 3, mask_info: list<item: int64>
      child 0, item: int64
seed: int64
inference_code_revision: string
adapter_repo: string
base_model: string
model: string
cinetrans_priority: int64
video_bytes: int64
inference_code: string
global_prompt_sha256: string
video_sha256: string
base_model_revision: string
to
{'episode_id': Value('string'), 'cinetrans_priority': Value('int64'), 'source_generation_priority': Value('int64'), 'shot_count': Value('int64'), 'target_duration_seconds': Value('float64'), 'source_revision': Value('string'), 'source_snapshot_revision': Value('string'), 'source_episode_dir': Value('string'), 'source_contract_sha256': Value('string'), 'prompt_repo': Value('string'), 'prompt_repo_revision': Value('string'), 'prompt_path': Value('string'), 'model': Value('string'), 'base_model': Value('string'), 'base_model_revision': Value('string'), 'adapter': Value('string'), 'adapter_repo': Value('string'), 'adapter_revision': Value('string'), 'inference_code': Value('string'), 'inference_code_revision': Value('string'), 'inference_code_local_patch': Value('string'), 'task': Value('string'), 'width': Value('int64'), 'height': Value('int64'), 'fps': Value('int64'), 'frame_count': Value('int64'), 'sample_solver': Value('string'), 'sample_steps': Value('int64'), 'sample_shift': Value('int64'), 'guide_scale': Value('int64'), 'seed': Value('int64'), 'offload_model': Value('bool'), 't5_cpu': Value('bool'), 'general_prompt': Value('bool'), 'global_prompt': Value('string'), 'global_prompt_sha256': Value('string'), 'mask': {'mask_p': Value('int64'), 'start_masked_layer': Value('int64'), 'end_masked_layer': Value('int64'), 'mask_info': List(Value('int64'))}, 'shots': List({'shot_id': Value('string'), 'source_scene_id': Value('string'), 'target_duration_seconds': Value('float64'), 'prompt': Value('string'), 'prompt_sha256': Value('string')}), 'filename': Value('string'), 'video_sha256': Value('string'), 'video_bytes': Value('int64'), 'generated_wall_seconds': Value('float64'), 'generated_at_utc': Value('string'), 'validation': {'status': Value('string'), 'codec': Value('string'), 'duration_seconds': Value('float64'), 'fps': Value('int64'), 'width': Value('int64'), 'height': Value('int64'), 'frame_count': Value('int64'), 'full_file_decode': Value('string'), 'visual_audit': 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
              schema_version: string
              dimension_profile: string
              reference_text: string
              shot_ids: list<item: string>
                child 0, item: string
              narrative: struct<facts: list<item: struct<fact_id: string, text: string>>>
                child 0, facts: list<item: struct<fact_id: string, text: string>>
                    child 0, item: struct<fact_id: string, text: string>
                        child 0, fact_id: string
                        child 1, text: string
              registry: struct<schema_version: string, shot_ids: list<item: string>, entities: list<item: struct<prompt_enti (... 197 chars omitted)
                child 0, schema_version: string
                child 1, shot_ids: list<item: string>
                    child 0, item: string
                child 2, entities: list<item: struct<prompt_entity_id: string, name: string, description: string, evidence: list<item:  (... 125 chars omitted)
                    child 0, item: struct<prompt_entity_id: string, name: string, description: string, evidence: list<item: struct<fact (... 113 chars omitted)
                        child 0, prompt_entity_id: string
                        child 1, name: string
                        child 2, description: string
                        child 3, evidence: list<item: struct<fact_id: string, quote: string>>
                            child 0, item: struct<fact_id: string, quote: string>
                                child 0, fact_id: string
                                child 1, quote: string
                        child 4, schedule: list<item: struct<shot_id: string, expectation: string, reason: string>>
                            child 0, item: struct<shot_id: string, expectation: string, reason: string>
                                child 0, shot_id: s
              ...
              , target_duration_seconds: double
                    child 3, prompt: string
                    child 4, prompt_sha256: string
              source_snapshot_revision: string
              sample_solver: string
              filename: string
              sample_steps: int64
              validation: struct<status: string, codec: string, duration_seconds: double, fps: int64, width: int64, height: in (... 72 chars omitted)
                child 0, status: string
                child 1, codec: string
                child 2, duration_seconds: double
                child 3, fps: int64
                child 4, width: int64
                child 5, height: int64
                child 6, frame_count: int64
                child 7, full_file_decode: string
                child 8, visual_audit: string
              source_generation_priority: int64
              offload_model: bool
              source_contract_sha256: string
              source_episode_dir: string
              width: int64
              shot_count: int64
              prompt_repo: string
              target_duration_seconds: double
              generated_at_utc: string
              adapter: string
              sample_shift: int64
              prompt_path: string
              guide_scale: int64
              task: string
              general_prompt: bool
              generated_wall_seconds: double
              frame_count: int64
              mask: struct<mask_p: int64, start_masked_layer: int64, end_masked_layer: int64, mask_info: list<item: int6 (... 3 chars omitted)
                child 0, mask_p: int64
                child 1, start_masked_layer: int64
                child 2, end_masked_layer: int64
                child 3, mask_info: list<item: int64>
                    child 0, item: int64
              seed: int64
              inference_code_revision: string
              adapter_repo: string
              base_model: string
              model: string
              cinetrans_priority: int64
              video_bytes: int64
              inference_code: string
              global_prompt_sha256: string
              video_sha256: string
              base_model_revision: string
              to
              {'episode_id': Value('string'), 'cinetrans_priority': Value('int64'), 'source_generation_priority': Value('int64'), 'shot_count': Value('int64'), 'target_duration_seconds': Value('float64'), 'source_revision': Value('string'), 'source_snapshot_revision': Value('string'), 'source_episode_dir': Value('string'), 'source_contract_sha256': Value('string'), 'prompt_repo': Value('string'), 'prompt_repo_revision': Value('string'), 'prompt_path': Value('string'), 'model': Value('string'), 'base_model': Value('string'), 'base_model_revision': Value('string'), 'adapter': Value('string'), 'adapter_repo': Value('string'), 'adapter_revision': Value('string'), 'inference_code': Value('string'), 'inference_code_revision': Value('string'), 'inference_code_local_patch': Value('string'), 'task': Value('string'), 'width': Value('int64'), 'height': Value('int64'), 'fps': Value('int64'), 'frame_count': Value('int64'), 'sample_solver': Value('string'), 'sample_steps': Value('int64'), 'sample_shift': Value('int64'), 'guide_scale': Value('int64'), 'seed': Value('int64'), 'offload_model': Value('bool'), 't5_cpu': Value('bool'), 'general_prompt': Value('bool'), 'global_prompt': Value('string'), 'global_prompt_sha256': Value('string'), 'mask': {'mask_p': Value('int64'), 'start_masked_layer': Value('int64'), 'end_masked_layer': Value('int64'), 'mask_info': List(Value('int64'))}, 'shots': List({'shot_id': Value('string'), 'source_scene_id': Value('string'), 'target_duration_seconds': Value('float64'), 'prompt': Value('string'), 'prompt_sha256': Value('string')}), 'filename': Value('string'), 'video_sha256': Value('string'), 'video_bytes': Value('int64'), 'generated_wall_seconds': Value('float64'), 'generated_at_utc': Value('string'), 'validation': {'status': Value('string'), 'codec': Value('string'), 'duration_seconds': Value('float64'), 'fps': Value('int64'), 'width': Value('int64'), 'height': Value('int64'), 'frame_count': Value('int64'), 'full_file_decode': Value('string'), 'visual_audit': 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.

CineTrans-DiT prompts and generations for EEC-Bench

Private release of the exact CineTrans generation prompts for all 400 episodes in the frozen EEC-Bench queue, plus the completed videos that passed QA. Every prompt record is keyed by its original episode_id.

Coverage

prompts/episodes/<episode_id>.json: exact prompt inputs for all 400 episodes. Each file includes the original episode ID, queue priority, source provenance, global_prompt, each per-shot prompt, and SHA-256 hashes. Per-shot prompt text is the frozen wan_prompt_en; the global text is global_prompt_original.
The generation run covered only the first 100 queue entries. It produced 26 validated videos; 22 entries have failed terminal attempts and 52 have no completion event. The remaining 300 have prompts in this dataset but were not attempted in that run.
episodes/<episode_id>/video.mp4 and metadata.json exist only for the 26 validated videos. Original source artifacts for those videos remain under each episode's source/ directory.
runs/priority_1_400_manifest.jsonl maps all 400 original IDs to prompt files and generation/video status. runs/priority_1_100_manifest.jsonl preserves the original partial-run manifest.

Inference and validation

CineTrans-DiT uses the Wan2.1-T2V-1.3B backbone with the released CineTrans LoRA. For the completed priority-100 run, per-shot wan_prompt_en and the episode's global_prompt_original were used unchanged. The 26 included videos passed H.264, 832x480, 16 fps, and expected-duration checks; FFmpeg decoded every full file, and a midpoint frame from each shot was reviewed. This is a coarse visual correspondence audit, not frame-by-frame semantic scoring.

Downloads last month: 9

Total file size:

3.53 GB