The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 80, in _split_generators
raise ValueError(
...<2 lines>...
)
ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
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/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.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.
PubMedVision — grouped medical VQA TARs
Unofficial format conversion of FreedomIntelligence/PubMedVision at revision 3c84e04b38bceb5341419b9a4f8ca37ba790cb84. The upstream dataset card, citations, author credits and terms are preserved in SOURCE_README.md. Original authors retain credit and rights; this conversion adds no clinical annotations.
English Alignment and Instruction Tuning VQA plus the released Chinese QA version. Original-caption-only records are excluded. No official test split exists. Multi-image cases remain together. Whole QA cases with missing source images are excluded without substitution; see missing_image_exclusions in preparation_report.json for source IDs and counts.
| Split | Rows | Unique images | QA pairs |
|---|---|---|---|
| train | 631355 | 898912 | 1737674 |
Train and validation are merged where available. Test has priority: entire train cases sharing source image keys, decoded RGB pixels or exported JPEG pixels with test are excluded. This detects exact-image overlap, not patient-level overlap or visually similar derivatives. Each ordered image set has one row with all associated QA. Multi-image questions are never copied onto incomplete single-image rows.
JPEG quality 95, original dimensions, no padding/resizing. Conversation roles are human/gpt, with answers inside <answer> tags. No invented reasoning or boxes. Source metadata is aligned with each QA pair. partition: LLaVA-OneVision-2-Spatial is a compatibility field, not an upstream membership claim.
Each TAR contains one matching JSONL and an image/ directory, with at most 10,000 rows per shard. Train and test are sharded independently. JSONL image entries are basenames under image/, and tar_name identifies the archive. Each TAR member is verified against its prepared file; uploaded TAR SHA256 values are checked against the Hub. Extract shards to separate directories.
See preparation_report.json for exclusions and packaging_summary.json for exact counts and checksums.
- Downloads last month
- 411