feature-information-dynamics
Collection
Prepared data for Feature Information Dynamics in Diffusion. • 2 items • Updated
Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code: DatasetGenerationError
Exception: ValueError
Message: Expected object or value
Traceback: Traceback (most recent call last):
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: Column() changed from object to string in row 0
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1827, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 613, in wrapped
for item in generator(*args, **kwargs):
~~~~~~~~~^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 304, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
examples = [ujson_loads(line) for line in original_batch.splitlines()]
~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
ValueError: Expected object or value
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
~~~~~~~~~~~~~~~~~~~~~~~~~^
builder, max_dataset_size_bytes=max_dataset_size_bytes
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
for job_id, done, content in self._prepare_split_single(
~~~~~~~~~~~~~~~~~~~~~~~~~~^
gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
):
^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1879, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
filenames string |
|---|
n01440764/n01440764_7055.JPEG |
n01440764/n01440764_13035.JPEG |
n01440764/n01440764_11745.JPEG |
n01440764/n01440764_11570.JPEG |
n01440764/n01440764_6471.JPEG |
n01440764/n01440764_1793.JPEG |
n01440764/n01440764_8748.JPEG |
n01440764/n01440764_14634.JPEG |
n01440764/n01440764_10542.JPEG |
n01440764/n01440764_6174.JPEG |
n01440764/n01440764_12314.JPEG |
n01440764/n01440764_12950.JPEG |
n01440764/n01440764_10733.JPEG |
n01440764/n01440764_7889.JPEG |
n01440764/n01440764_9523.JPEG |
n01440764/n01440764_5108.JPEG |
n01440764/n01440764_5066.JPEG |
n01440764/n01440764_9695.JPEG |
n01440764/n01440764_8384.JPEG |
n01440764/n01440764_3724.JPEG |
n01440764/n01440764_17058.JPEG |
n01440764/n01440764_11656.JPEG |
n01440764/n01440764_19807.JPEG |
n01440764/n01440764_2094.JPEG |
n01440764/n01440764_12111.JPEG |
n01440764/n01440764_7562.JPEG |
n01440764/n01440764_7714.JPEG |
n01440764/n01440764_8334.JPEG |
n01440764/n01440764_5256.JPEG |
n01440764/n01440764_1770.JPEG |
n01440764/n01440764_485.JPEG |
n01440764/n01440764_7677.JPEG |
n01440764/n01440764_5633.JPEG |
n01440764/n01440764_8466.JPEG |
n01440764/n01440764_13161.JPEG |
n01440764/n01440764_14555.JPEG |
n01440764/n01440764_5148.JPEG |
n01440764/n01440764_22963.JPEG |
n01440764/n01440764_11652.JPEG |
n01440764/n01440764_24502.JPEG |
n01440764/n01440764_14150.JPEG |
n01440764/n01440764_7379.JPEG |
n01440764/n01440764_9091.JPEG |
n01440764/n01440764_11593.JPEG |
n01440764/n01440764_7383.JPEG |
n01440764/n01440764_6641.JPEG |
n01440764/n01440764_13275.JPEG |
n01440764/n01440764_14294.JPEG |
n01440764/n01440764_19035.JPEG |
n01440764/n01440764_2433.JPEG |
n01440764/n01440764_20299.JPEG |
n01440764/n01440764_12898.JPEG |
n01440764/n01440764_416.JPEG |
n01440764/n01440764_8528.JPEG |
n01440764/n01440764_10319.JPEG |
n01440764/n01440764_2121.JPEG |
n01440764/n01440764_2118.JPEG |
n01440764/n01440764_13806.JPEG |
n01440764/n01440764_9846.JPEG |
n01440764/n01440764_1514.JPEG |
n01440764/n01440764_12040.JPEG |
n01440764/n01440764_10550.JPEG |
n01440764/n01440764_2919.JPEG |
n01440764/n01440764_1141.JPEG |
n01440764/n01440764_4506.JPEG |
n01440764/n01440764_8451.JPEG |
n01440764/n01440764_1284.JPEG |
n01440764/n01440764_10365.JPEG |
n01440764/n01440764_9788.JPEG |
n01440764/n01440764_12063.JPEG |
n01440764/n01440764_18128.JPEG |
n01440764/n01440764_6917.JPEG |
n01440764/n01440764_10787.JPEG |
n01440764/n01440764_17896.JPEG |
n01440764/n01440764_7752.JPEG |
n01440764/n01440764_1145.JPEG |
n01440764/n01440764_4852.JPEG |
n01440764/n01440764_7428.JPEG |
n01440764/n01440764_10470.JPEG |
n01440764/n01440764_9898.JPEG |
n01440764/n01440764_26006.JPEG |
n01440764/n01440764_3778.JPEG |
n01440764/n01440764_9217.JPEG |
n01440764/n01440764_11011.JPEG |
n01440764/n01440764_4510.JPEG |
n01440764/n01440764_12659.JPEG |
n01440764/n01440764_3259.JPEG |
n01440764/n01440764_11887.JPEG |
n01440764/n01440764_14435.JPEG |
n01440764/n01440764_2835.JPEG |
n01440764/n01440764_200.JPEG |
n01440764/n01440764_7362.JPEG |
n01440764/n01440764_993.JPEG |
n01440764/n01440764_8150.JPEG |
n01440764/n01440764_16051.JPEG |
n01440764/n01440764_5410.JPEG |
n01440764/n01440764_14034.JPEG |
n01440764/n01440764_2729.JPEG |
n01440764/n01440764_10029.JPEG |
n01440764/n01440764_13757.JPEG |
End of preview.
Feature Information Dynamics: prepared ImageNet conditions
Prepared spatial conditions and paired sample metadata for Feature Information Dynamics in Diffusion. The common component contains segmentation masks, masked-Canny maps, filename sidecars, paired train/validation lists and a full ImageNet class-ID mapping. It contains no RGB images or VAVAE latent arrays. The optional VAVAE component contains sampled original/flip latents, globally correct labels, filename sidecars and channel mean/std statistics.
Both common and optional VAVAE components are published. All 325 payload identities were verified. Companion code and actual readers were checked using existing data.
Usage
Obtain ImageNet training images separately and retain all 1000 class directories. Install the companion Feature Information Dynamics repository, then use:
python -m pip install -e '.[data]'
feature-information data download --manifest https://huggingface.co/datasets/AI4Science-WestlakeU/feature-information-dynamics/resolve/main/release.json --output data/prepared
The release manifest is published after its payloads have been uploaded and verified. Common is installed by default;
add --component vavae for the offline VAVAE cache. Pixel, RAE and online SDVAE
do not require the VAVAE component. Follow the code repository's docs/data.md
for weight resources, configuration generation and stage-chain execution.
Files and identity
Each release file has a relative installed path, byte count and SHA256. Compressed payloads additionally identify the source path, compressed byte count and SHA256. The installer restores the original tensor bytes. Keep the supplied filename order and global class IDs; do not renumber retained SR95 classes or substitute a random image-level split.
The common component has 196 payloads: 3,344,352,304 download bytes and 85,089,831,558 installed bytes. VAVAE has 129 payloads: 39,500,888,780 download bytes and 42,548,211,383 installed bytes. Actual payload sizes and split counts must be taken from the published release manifest, not these rounded historical totals. The release contains 547448 training and 66685 validation images (614133 unique). It retains the 980 invalid mask sentinels and removes the validation occurrence of the historical one-image train/validation overlap; the manifest records this correction.
Attribution
The images originate from ImageNet. Masks were generated using SAM3.1; VAVAE encoding follows LightningDiT. Please cite the original dataset/models and Feature Information Dynamics when using this release. The companion code's MIT license does not replace upstream dataset/model terms.
Release verification
Each release records original and compressed SHA256 checksums and exact split counts. The installer validates those identities and spatial/latent alignment. Repository code checks and bounded real-data reader checks are documented in the companion code repository. They do not claim a new full training rerun.
- Downloads last month
- 96