The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
dim: int64
dtype: string
format: string
num_queries: int64
qas: list<item: string>
child 0, item: string
qrels_rows: int64
query_maxlen: int64
questions: list<item: string>
child 0, item: string
source: string
max_querylen: int64
num_docs: int64
num_query_tokens: int64
chunks: string
queries_copied: bool
max_doclen: int64
num_chunks: int64
docs_f32_materialized: bool
chunk_ids: list<item: int64>
child 0, item: int64
num_doc_tokens: int64
to
{'chunk_ids': List(Value('int64')), 'chunks': Value('string'), 'dim': Value('int64'), 'docs_f32_materialized': Value('bool'), 'format': Value('string'), 'max_doclen': Value('int64'), 'max_querylen': Value('int64'), 'num_chunks': Value('int64'), 'num_doc_tokens': Value('int64'), 'num_docs': Value('int64'), 'num_queries': Value('int64'), 'num_query_tokens': Value('int64'), 'queries_copied': Value('bool'), 'source': 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 2951, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, 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 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 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
dim: int64
dtype: string
format: string
num_queries: int64
qas: list<item: string>
child 0, item: string
qrels_rows: int64
query_maxlen: int64
questions: list<item: string>
child 0, item: string
source: string
max_querylen: int64
num_docs: int64
num_query_tokens: int64
chunks: string
queries_copied: bool
max_doclen: int64
num_chunks: int64
docs_f32_materialized: bool
chunk_ids: list<item: int64>
child 0, item: int64
num_doc_tokens: int64
to
{'chunk_ids': List(Value('int64')), 'chunks': Value('string'), 'dim': Value('int64'), 'docs_f32_materialized': Value('bool'), 'format': Value('string'), 'max_doclen': Value('int64'), 'max_querylen': Value('int64'), 'num_chunks': Value('int64'), 'num_doc_tokens': Value('int64'), 'num_docs': Value('int64'), 'num_queries': Value('int64'), 'num_query_tokens': Value('int64'), 'queries_copied': Value('bool'), 'source': Value('string')}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Multi-vector retrieval inputs
Inputs are organized under workloads/<dataset>/<embedding-model>/<split-or-scale>/.
Within each split, shared/ contains common inputs and system directories contain
indexes and representation variants. verum/ is the current name of TSUMUGI.
Variant names preserve parameters such as quantization bits, centroid counts,
native indexes, and physical order. Do not combine query orders or document IDs
across splits. These research inputs retain their original dataset/model licenses.
| Dataset | Directory |
|---|---|
| activitynet | workloads/activitynet |
| audiocaps | workloads/audiocaps |
| evqa | workloads/evqa |
| inquire | workloads/inquire |
| lotte-lifestyle | workloads/lotte-lifestyle |
| msmarco-v1 | workloads/msmarco-v1 |
| msrvtt | workloads/msrvtt |
| okvqa | workloads/okvqa |
Restore inputs
catalog/current.json lists paths, fixed source commits, sizes and checksums.
Use releases/dataset-layout-20260917-v1/restore_inputs.py to select inputs,
download through HF Mirror or the official endpoint, and verify their contents.
See the restoration guide.
Original payloads are unchanged. Original metadata may describe historical local
paths; use the catalog for downloads and provide local paths to runners. Historical
basenames such as lifestyle.test are retained where loaders expect them: they do
not change the dataset identified by the enclosing workload directory.
releases/ contains frozen experiment/recovery references, not new copies of data.
Older fixed revisions remain available. Scripts with old paths must pin the old
revision or use the compatibility catalog rather than assume old paths exist on main.
Other material
msmarco_v2/, other LoTTe subsets under datasets/ and indexes/, and checkpoints/
are outside the eight-workload reorganization and were left unchanged.
The private research archive and anonymous code repository are separate.
- Downloads last month
- 66