Datasets:

nicemyeong
/

cacheblend-rag-extended

Error code:   DatasetGenerationError
Exception:    CastError
Message:      Couldn't cast
musique: struct<dataset: string, n_examples: int64, chunks_per_query_min: int64, chunks_per_query_max: int64, (... 622 chars omitted)
  child 0, dataset: string
  child 1, n_examples: int64
  child 2, chunks_per_query_min: int64
  child 3, chunks_per_query_max: int64
  child 4, chunks_per_query_target: int64
  child 5, chunks_per_query_uniform: bool
  child 6, gold_substring_hits_source: int64
  child 7, gold_substring_hits_extended: int64
  child 8, gold_preservation_ratio_vs_source: double
  child 9, structural_chunk_preservation: double
  child 10, avg_chunk_chars: double
  child 11, avg_chunk_tokens_estimated: double
  child 12, cacheblend_suitability: struct<checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, struc (... 138 chars omitted)
      child 0, checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, structural_chunk_pre (... 75 chars omitted)
          child 0, chunks_per_query_ge_10: bool
          child 1, gold_preservation_vs_source_ge_0.99: bool
          child 2, structural_chunk_preservation_eq_1.0: bool
          child 3, chunk_size_within_paper_range_300_800_tokens: bool
      child 1, verdict_pass: bool
      child 2, notes: list<item: string>
          child 0, item: string
  child 13, source_file: string
  child 14, extra_per_query: int64
  child 15, seed: int64
  child 16, output_file: string
wikimqa: struct<dataset: string, n_examples: int64, chunks_per_query_min: int64, chunks_per_query_max: int64, (... 622 chars omitted)
  child 0, dataset: string
  child 1, n_examples: int64
  child 2, chunks_per_query_min: int64
  child 3, chunks_per_query_max: int64
  child 4, chunks_per_query_target: int64
  child 5, chunks_per_query_uniform: bool
  child 6, gold_substring_hits_source: int64
  child 7, gold_substring_hits_extended: int64
  child 8, gold_preservation_ratio_vs_source: double
  child 9, structural_chunk_preservation: double
  child 10, avg_chunk_chars: double
  child 11, avg_chunk_tokens_estimated: double
  child 12, cacheblend_suitability: struct<checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, struc (... 138 chars omitted)
      child 0, checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, structural_chunk_pre (... 75 chars omitted)
          child 0, chunks_per_query_ge_10: bool
          child 1, gold_preservation_vs_source_ge_0.99: bool
          child 2, structural_chunk_preservation_eq_1.0: bool
          child 3, chunk_size_within_paper_range_300_800_tokens: bool
      child 1, verdict_pass: bool
      child 2, notes: list<item: string>
          child 0, item: string
  child 13, source_file: string
  child 14, extra_per_query: int64
  child 15, seed: int64
  child 16, output_file: string
to
{'question': Value('string'), 'ctxs': List({'title': Value('string'), 'text': Value('string')}), 'answers': List(Value('string'))}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1800, in _prepare_split_single
                  writer.write_table(table)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 764, in write_table
                  self.write_rows_on_file()  # in case there are buffered rows to write first
                  ^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 663, in write_rows_on_file
                  self._write_table(table)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2321, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2249, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              musique: struct<dataset: string, n_examples: int64, chunks_per_query_min: int64, chunks_per_query_max: int64, (... 622 chars omitted)
                child 0, dataset: string
                child 1, n_examples: int64
                child 2, chunks_per_query_min: int64
                child 3, chunks_per_query_max: int64
                child 4, chunks_per_query_target: int64
                child 5, chunks_per_query_uniform: bool
                child 6, gold_substring_hits_source: int64
                child 7, gold_substring_hits_extended: int64
                child 8, gold_preservation_ratio_vs_source: double
                child 9, structural_chunk_preservation: double
                child 10, avg_chunk_chars: double
                child 11, avg_chunk_tokens_estimated: double
                child 12, cacheblend_suitability: struct<checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, struc (... 138 chars omitted)
                    child 0, checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, structural_chunk_pre (... 75 chars omitted)
                        child 0, chunks_per_query_ge_10: bool
                        child 1, gold_preservation_vs_source_ge_0.99: bool
                        child 2, structural_chunk_preservation_eq_1.0: bool
                        child 3, chunk_size_within_paper_range_300_800_tokens: bool
                    child 1, verdict_pass: bool
                    child 2, notes: list<item: string>
                        child 0, item: string
                child 13, source_file: string
                child 14, extra_per_query: int64
                child 15, seed: int64
                child 16, output_file: string
              wikimqa: struct<dataset: string, n_examples: int64, chunks_per_query_min: int64, chunks_per_query_max: int64, (... 622 chars omitted)
                child 0, dataset: string
                child 1, n_examples: int64
                child 2, chunks_per_query_min: int64
                child 3, chunks_per_query_max: int64
                child 4, chunks_per_query_target: int64
                child 5, chunks_per_query_uniform: bool
                child 6, gold_substring_hits_source: int64
                child 7, gold_substring_hits_extended: int64
                child 8, gold_preservation_ratio_vs_source: double
                child 9, structural_chunk_preservation: double
                child 10, avg_chunk_chars: double
                child 11, avg_chunk_tokens_estimated: double
                child 12, cacheblend_suitability: struct<checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, struc (... 138 chars omitted)
                    child 0, checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, structural_chunk_pre (... 75 chars omitted)
                        child 0, chunks_per_query_ge_10: bool
                        child 1, gold_preservation_vs_source_ge_0.99: bool
                        child 2, structural_chunk_preservation_eq_1.0: bool
                        child 3, chunk_size_within_paper_range_300_800_tokens: bool
                    child 1, verdict_pass: bool
                    child 2, notes: list<item: string>
                        child 0, item: string
                child 13, source_file: string
                child 14, extra_per_query: int64
                child 15, seed: int64
                child 16, output_file: string
              to
              {'question': Value('string'), 'ctxs': List({'title': Value('string'), 'text': Value('string')}), 'answers': List(Value('string'))}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1802, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 2 new columns ({'wikimqa', 'musique'}) and 3 missing columns ({'question', 'answers', 'ctxs'}).
              
              This happened while the json dataset builder was generating data using
              
              hf://datasets/nicemyeong/cacheblend-rag-extended/wikimqa_extended.json (at revision e7039812fdb956c2955f291b38120944a66fe8bb), [/tmp/hf-datasets-cache/medium/datasets/93574789425726-config-parquet-and-info-nicemyeong-cacheblend-rag-724086b7/hub/datasets--nicemyeong--cacheblend-rag-extended/snapshots/e7039812fdb956c2955f291b38120944a66fe8bb/musique_extended.json (origin=hf://datasets/nicemyeong/cacheblend-rag-extended@e7039812fdb956c2955f291b38120944a66fe8bb/musique_extended.json), /tmp/hf-datasets-cache/medium/datasets/93574789425726-config-parquet-and-info-nicemyeong-cacheblend-rag-724086b7/hub/datasets--nicemyeong--cacheblend-rag-extended/snapshots/e7039812fdb956c2955f291b38120944a66fe8bb/suitability_report.json (origin=hf://datasets/nicemyeong/cacheblend-rag-extended@e7039812fdb956c2955f291b38120944a66fe8bb/suitability_report.json), /tmp/hf-datasets-cache/medium/datasets/93574789425726-config-parquet-and-info-nicemyeong-cacheblend-rag-724086b7/hub/datasets--nicemyeong--cacheblend-rag-extended/snapshots/e7039812fdb956c2955f291b38120944a66fe8bb/wikimqa_extended.json (origin=hf://datasets/nicemyeong/cacheblend-rag-extended@e7039812fdb956c2955f291b38120944a66fe8bb/wikimqa_extended.json)], ['hf://datasets/nicemyeong/cacheblend-rag-extended@e7039812fdb956c2955f291b38120944a66fe8bb/musique_extended.json', 'hf://datasets/nicemyeong/cacheblend-rag-extended@e7039812fdb956c2955f291b38120944a66fe8bb/suitability_report.json', 'hf://datasets/nicemyeong/cacheblend-rag-extended@e7039812fdb956c2955f291b38120944a66fe8bb/wikimqa_extended.json']
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1821, in _prepare_split_single
                  num_examples, num_bytes = writer.finalize()
                                            ^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 781, in finalize
                  self.write_rows_on_file()
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 663, in write_rows_on_file
                  self._write_table(table)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2321, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2249, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              musique: struct<dataset: string, n_examples: int64, chunks_per_query_min: int64, chunks_per_query_max: int64, (... 622 chars omitted)
                child 0, dataset: string
                child 1, n_examples: int64
                child 2, chunks_per_query_min: int64
                child 3, chunks_per_query_max: int64
                child 4, chunks_per_query_target: int64
                child 5, chunks_per_query_uniform: bool
                child 6, gold_substring_hits_source: int64
                child 7, gold_substring_hits_extended: int64
                child 8, gold_preservation_ratio_vs_source: double
                child 9, structural_chunk_preservation: double
                child 10, avg_chunk_chars: double
                child 11, avg_chunk_tokens_estimated: double
                child 12, cacheblend_suitability: struct<checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, struc (... 138 chars omitted)
                    child 0, checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, structural_chunk_pre (... 75 chars omitted)
                        child 0, chunks_per_query_ge_10: bool
                        child 1, gold_preservation_vs_source_ge_0.99: bool
                        child 2, structural_chunk_preservation_eq_1.0: bool
                        child 3, chunk_size_within_paper_range_300_800_tokens: bool
                    child 1, verdict_pass: bool
                    child 2, notes: list<item: string>
                        child 0, item: string
                child 13, source_file: string
                child 14, extra_per_query: int64
                child 15, seed: int64
                child 16, output_file: string
              wikimqa: struct<dataset: string, n_examples: int64, chunks_per_query_min: int64, chunks_per_query_max: int64, (... 622 chars omitted)
                child 0, dataset: string
                child 1, n_examples: int64
                child 2, chunks_per_query_min: int64
                child 3, chunks_per_query_max: int64
                child 4, chunks_per_query_target: int64
                child 5, chunks_per_query_uniform: bool
                child 6, gold_substring_hits_source: int64
                child 7, gold_substring_hits_extended: int64
                child 8, gold_preservation_ratio_vs_source: double
                child 9, structural_chunk_preservation: double
                child 10, avg_chunk_chars: double
                child 11, avg_chunk_tokens_estimated: double
                child 12, cacheblend_suitability: struct<checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, struc (... 138 chars omitted)
                    child 0, checks: struct<chunks_per_query_ge_10: bool, gold_preservation_vs_source_ge_0.99: bool, structural_chunk_pre (... 75 chars omitted)
                        child 0, chunks_per_query_ge_10: bool
                        child 1, gold_preservation_vs_source_ge_0.99: bool
                        child 2, structural_chunk_preservation_eq_1.0: bool
                        child 3, chunk_size_within_paper_range_300_800_tokens: bool
                    child 1, verdict_pass: bool
                    child 2, notes: list<item: string>
                        child 0, item: string
                child 13, source_file: string
                child 14, extra_per_query: int64
                child 15, seed: int64
                child 16, output_file: string
              to
              {'question': Value('string'), 'ctxs': List({'title': Value('string'), 'text': Value('string')}), 'answers': List(Value('string'))}
              because column names don't match
              
              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 1347, in compute_config_parquet_and_info_response
                  parquet_operations = convert_to_parquet(builder)
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 980, in convert_to_parquet
                  builder.download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 882, in download_and_prepare
                  self._download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 943, in _download_and_prepare
                  self._prepare_split(split_generator, **prepare_split_kwargs)
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1646, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1832, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

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.

question string	ctxs list	answers list
Where was the author of Hannibal and Scipio educated at?	[ { "title": "", "text": "\" ) whose fissile material was highly enriched uranium, and a plutonium - based device ( see trinity test and \" fat man \" ) whose plutonium was derived from uranium - 238. the uranium - based little boy device became the first nuclear weapon used in war when it was detonated over ...	[ "Exeter College" ]
In which county is Southern Maryland Electric Cooperative headquartered?	[ { "title": "", "text": "and a newly formed corporation decided to merge to form fuji heavy industries. these companies were : fuji kogyo, a scooter manufacturer ; coachbuilders fuji jidosha ; engine manufacturers omiya fuji kogyo ; chassis builders utsunomiya sharyo and the tokyo fuji dangyo trading company...	[ "Charles County" ]
Who is the child of the Victim of Romance performer?	[ { "title": "", "text": "still have feelings for each other. but nicole is engaged to daniel jonas, and when eric and nicole escape the basement, nicole breaks things off with eric, leaving him devastated. in august 2015, eric finds serena strangled to death in the park ( she is the first victim of the neckt...	[ "Chynna Phillips" ]
What county was Tim Dubois born in?	[ { "title": "", "text": "\", which quickly became the bestselling spanish newspaper, a crown it holds to this day. he was survived by his wife, simone ortega, and three children, one of whom works as a journalist for \" el pais \". enos bronson ( march 31, 1774 – april 22, 1823 ) was an american writer and n...	[ "McDonald County" ]
What record label did the person who is part of The Bruce Lee Band start?	[{"title":"","text":"with his touring band, the noisemakers, and his last album for rca records. fis(...TRUNCATED)	[ "Asian Man Records" ]
What is another notable work made by the author of Miss Sara Sampson?	[{"title":"","text":"and adolphe menjou. sara sidner ( born may 31, 1972 ) is an american journalist(...TRUNCATED)	[ "Emilia Galotti" ]
What instrument is played by the person from The Blackout All-Stars?	[{"title":"","text":"is taken from a famous quotation by meher baba. the ` ` instruments'' in the a (...TRUNCATED)	[ "conga" ]
What is the seat of the county where Van Hook Township is located?	[{"title":"","text":"in the united states. it is approximately long and flows through warton townshi(...TRUNCATED)	[ "Stanley" ]
Who is the father of Edward Baring, 1st Baron Revelstoke's father?	[{"title":"","text":"kingdom of sardinia, if the nobility of italy had not been legally abolished af(...TRUNCATED)	[ "Sir Francis Baring, 1st Baronet" ]
What group was the performer of Be the One a member of?	[{"title":"","text":", tito puente, tito nieves, paquito d'rivera, dave valentin, grover washington,(...TRUNCATED)	[ "Jackson 5" ]

End of preview.

CacheBlend RAG Extended

Multi-hop QA splits derived from MuSiQue and 2WikiMQA (as bundled in the official CacheBlend repo), augmented with extra distractor chunks per query so that each example carries 20 context passages instead of the original 10.

Designed to amplify the contrast between full_reuse (no cross-chunk attention -> quality drop) and cacheblend (selective KV recompute -> quality recovers) while preserving the multi-hop questions and gold short answers.

Splits

split	n	chunks/query	~tokens/chunk	gold preserved vs source	structural preservation	CacheBlend ready
`musique`	150	20	598	100.00%	100%	PASS
`wikimqa`	200	20	560	103.81%	100%	PASS

Schema

Each row matches the bundled CacheBlend JSON format:

{
  "question": "Where was the author of Hannibal and Scipio educated at?",
  "ctxs": [
    {"title": "<wiki title>", "text": "<512-token passage>"},
    ...  // 20 entries
  ],
  "answers": ["Exeter College"]
}

How it was built

Start from musique_s.json (150 ex) and wikimqa_s.json (200 ex) — both already have the multi-hop questions and 10 gold/distractor passages per query that the CacheBlend paper used.
For each query, sample 10 additional chunks from OTHER queries in the same source (no chunk is reused twice within a query; titles deduped).
Shuffle the resulting 20 chunks so gold evidence is not always first.
Verify (per row): (i) chunks/query == 20, (ii) at least one gold-answer substring is preserved across the 20 chunks.

Reproduce with scripts/build_cacheblend_dataset.py --extra 10 in mkim0628/experiment-reproducer.

Suitability gates

The build script runs three CacheBlend-relevance checks. All splits must pass before upload:

chunks_per_query_ge_10: at least 10 passages so KV reuse pays off
gold_preservation_ratio_ge_0.95: ≥95 % of queries still have the gold-answer substring after distractor injection
chunk_size_within_paper_range_300_800_tokens: chunk length matches the paper's 512-token spec

See suitability_report.json for the per-split numbers.

Downloads last month: 38

Total file size:

16.7 MB