The dataset viewer is not available for this split.
Error code: FeaturesError
Exception: ArrowInvalid
Message: JSON parse error: Invalid value. in row 0
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 324, in _generate_tables
df = pandas_read_json(f)
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 38, in pandas_read_json
return pd.read_json(path_or_buf, **kwargs)
~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 815, in read_json
return json_reader.read()
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1014, in read
obj = self._get_object_parser(self.data)
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1040, in _get_object_parser
obj = FrameParser(json, **kwargs).parse()
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1176, in parse
self._parse()
~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1392, in _parse
ujson_loads(json, precise_float=self.precise_float), dtype=None
~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: Expected object or value
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 244, in compute_first_rows_from_streaming_response
iterable_dataset = iterable_dataset._resolve_features()
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 4523, in _resolve_features
features = _infer_features_from_batch(self.with_format(None)._head())
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2768, in _head
return next(iter(self.iter(batch_size=n)))
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2972, in iter
for key, pa_table in ex_iterable.iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2483, in _iter_arrow
yield from 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 327, in _generate_tables
raise e
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
return check_status(status)
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Invalid value. in row 0Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Episode 009 — Kandinsky 6.0: native video and audio test record
Watch the published episode: Kandinsky 6.0 Makes Its Own Sound. Does It Follow the Prompt?
31 complete outputs shown in Agentic Film Lab's episode, including the official samurai and dragon examples, recurring characters, physical actions, six visual styles, controlled prompting follow-ups, and four quantized tests. This is a dataset of experiment evidence, not model weights or a trained model.
What is included
Unmodified native MP4s with sound, source images, exact submitted/effective prompts, authoritative per-output runtime/settings receipts, sanitized execution traces, available RNG snapshots, technical failure records, checksums and generation/setup accounting. manifest.json links each output to its episode playback time. Individual runs are illustrative, not statistical quality or cost rankings.
Routes and controls
Originals and prompting follow-ups use non-distilled Pro-5s Diffusers: 864×480, 121 frames, 24 fps, 50 steps and guidance 5, on RTX PRO 6000 Blackwell Server Edition (96 GB class) with CPU offload. Original expansion was off; the inherited negative prompt discouraged several artistic styles. Follow-ups explicitly record changed expansion/negative treatment and effective text.
Quantized INT8/NVFP4 tests use Pro-distill through native ComfyUI, 10 steps/CFG 1 on NVIDIA L4 (24 GB class), dynamic CPU offload, expansion off, canonical native audio/speech syntax and empty negative text. NVFP4 on L4 uses an older-card fallback, not native FP4 acceleration. These are different checkpoint/routes, not a controlled quantization-only comparison. Native execution timing includes on-demand loading; Diffusers timing begins after pipeline load. See each receipt and workflow.
Cost scope
The grid allocates actual RunPod Serverless GPU bills to successful generation/execution windows: $60.79 total for the 31 shown outputs. Separately recorded setup/retries/disk allocation is $22.32, reconciling $83.11 for the selected-output lifecycle. Shared unitemized startup/idle is apportioned, so these are accounting allocations, not provider-itemized marginal prices. Initial download 40:45/$2.58 and verification 19:14/$1.22 are reported separately. No fal image charges or retained network-volume recurring storage. Archive-only K11–K15 footage and costs are omitted from this episode package.
Provenance and reproduction
Official model/examples: https://huggingface.co/kandinskylab/Kandinsky-6.0-Pro-5s-Diffusers and https://github.com/kandinskylab/kandinsky-6 . The official dragon reference is included with upstream attribution; original 4K source images were generated with GPT Image 2 through fal.ai. Consult upstream terms for upstream assets/software/weights; this record does not redistribute weights or relicense them. No blanket license is asserted for third-party material. Seeds, runtime revisions and exact prompts are in receipts; omitted example seeds and recorded RNG state do not promise bit-identical reproduction.
Download and verify
Download the complete evidence archive. It contains all 31 native clips with audio and the reproduction records. Extract it and use the included SHA256SUMS to verify the files. The archive is an immutable snapshot of the experiments; per-output receipts record the configurations actually executed.
SHA-256 of the archive: 3812b5baffe07af8f6221d5dd11a9d411ee4eeab5c42403e9d623c4566e801a7.
The record excludes credentials, voice references, private account payloads and the edited episode master. Model weights are not included. Optional PyTorch RNG snapshots are serialized records, so load them only with appropriate trust and tooling.
- Downloads last month
- 150