Datasets:

moorcheh
/

mair-floating-point-vs-quantized-benchmarks

Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 12 new columns ({'Moorcheh All', 'Provider', 'Rerank_Mean_ms', 'Elasticsearch', 'Pinecone (Search + Rerank)', 'Cohere (Rerank)', 'Search_Mean_ms', 'PGVector', 'Moorcheh Distance Only', 'Qdrant', 'Pinecone (Search)', 'Search_Server_Mean_ms'}) and 5 missing columns ({'PGVector - Vector (Floating-Point)', 'Qdrant - Vector (Floating-Point)', 'Moorcheh - Vector (Quantized)', 'Elasticsearch - Vector (Floating-Point)', 'Pinecone with Cohere - Vector (Floating-Point)'}).

This happened while the csv dataset builder was generating data using

hf://datasets/moorcheh/mair-floating-point-vs-quantized-benchmarks/latency_vector_fp.csv (at revision 09cf874987ffa43796b5e512444fcb330f7a38cb)

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)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1831, in _prepare_split_single
                  writer.write_table(table)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 714, in write_table
                  pa_table = table_cast(pa_table, self._schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2272, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2218, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              Dataset: string
              Category: string
              Dataset Size: int64
              Provider: string
              Search_Server_Mean_ms: double
              Search_Mean_ms: double
              Rerank_Mean_ms: double
              Moorcheh Distance Only: double
              Elasticsearch: double
              PGVector: double
              Qdrant: double
              Moorcheh All: double
              Pinecone (Search + Rerank): double
              Pinecone (Search): double
              Cohere (Rerank): double
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 2174
              to
              {'Dataset': Value('string'), 'Category': Value('string'), 'Dataset Size': Value('int64'), 'Moorcheh - Vector (Quantized)': Value('float64'), 'Elasticsearch - Vector (Floating-Point)': Value('float64'), 'Pinecone with Cohere - Vector (Floating-Point)': Value('float64'), 'PGVector - Vector (Floating-Point)': Value('float64'), 'Qdrant - Vector (Floating-Point)': Value('float64')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1339, 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 972, in convert_to_parquet
                  builder.download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 894, in download_and_prepare
                  self._download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 970, in _download_and_prepare
                  self._prepare_split(split_generator, **prepare_split_kwargs)
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1702, 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 1833, 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 12 new columns ({'Moorcheh All', 'Provider', 'Rerank_Mean_ms', 'Elasticsearch', 'Pinecone (Search + Rerank)', 'Cohere (Rerank)', 'Search_Mean_ms', 'PGVector', 'Moorcheh Distance Only', 'Qdrant', 'Pinecone (Search)', 'Search_Server_Mean_ms'}) and 5 missing columns ({'PGVector - Vector (Floating-Point)', 'Qdrant - Vector (Floating-Point)', 'Moorcheh - Vector (Quantized)', 'Elasticsearch - Vector (Floating-Point)', 'Pinecone with Cohere - Vector (Floating-Point)'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/moorcheh/mair-floating-point-vs-quantized-benchmarks/latency_vector_fp.csv (at revision 09cf874987ffa43796b5e512444fcb330f7a38cb)
              
              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)

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.

Dataset string	Category string	Dataset Size int64	Moorcheh - Vector (Quantized) float64	Elasticsearch - Vector (Floating-Point) float64	Pinecone with Cohere - Vector (Floating-Point) float64	PGVector - Vector (Floating-Point) float64	Qdrant - Vector (Floating-Point) float64
AILA2019-Statutes	Legal & Regulatory	197	0.228303	0.222806	0.209694	0.2222	0.221341
Apple	API Documentation	678	0.642484	0.63617	0.652698	0.6362	0.636662
AILA2019-Case	Legal & Regulatory	2,914	0.137408	0.169442	0.112211	0.1693	0.169729
LeCaRDv2	Legal & Regulatory	3,000	0.661731	0.694548	0.540271	0.6938	0.696224
NFCorpus	Medical & Clinical	3,633	0.361107	0.394753	0.360817	0.3947	0.393532
REGIR-UK2EU	Legal & Regulatory	3,930	0.578363	0.667594	0.485664	0.6676	0.667251
HC3Finance	Financial	3,933	0.400558	0.427735	0.378612	0.4277	0.432085
ConvFinQA	Financial	6,503	0.742992	0.768612	0.78801	0.7691	0.764325
REGIR-EU2UK	Legal & Regulatory	10,000	0.621456	0.641452	0.645327	0.6315	0.636338
FinQA	Financial	11,865	0.683209	0.692795	0.740212	0.6816	0.6916
FinanceBench	Financial	15,325	0.578197	0.673586	0.63823	0.6636	0.671871
LegalQuAD	Legal & Regulatory	17,702	0.667329	0.684662	0.789092	0.6685	0.685607
ACORDAR	Legal & Regulatory	31,589	0.308483	0.341478	0.301194	0.3402	0.321282
FiQA	Financial	57,638	0.539874	0.56898	0.533496	0.5633	0.568467

Dataset Coverage

The benchmarks evaluate performance across 14+ specialized datasets covering:

Legal & Regulatory: ACORDAR, AILA2019-Case, AILA2019-Statutes, LeCaRDv2, LegalQuAD, REGIR-EU2UK, REGIR-UK2EU.
Financial: ConvFinQA, FinanceBench, FinQA, FiQA, HC3Finance.
Medical & Clinical: NFCorpus.
General/API: Apple Documentation.

Metrics

NDCG@10: Normalized Discounted Cumulative Gain at rank 10, measuring retrieval quality.
Latency (ms): Mean search latency measured on the server-side and end-to-end.

Benchmark Configurations

The data is organized into four primary splits, which you can switch between using the "Viewer" tab above:

NDCG@10 - Floating Point (ndcg_vector_fp): Evaluation of retrieval accuracy using standard 32-bit floating-point vectors.
Latency - Floating Point (latency_vector_fp): Search performance (ms) for floating-point vector retrieval.
NDCG@10 - Quantized (ndcg_vector_quantized): Evaluation of retrieval accuracy using optimized quantized/binary embedding paths.
Latency - Quantized (latency_vector_quantized): Search performance (ms) for optimized quantized vector retrieval.

Tested Providers

Results include comparisons across:

Moorcheh
Elasticsearch
Pinecone (with Cohere Rerank)
PGVector (PostgreSQL)
Qdrant

How to use in Python

You can load these results directly into a Pandas DataFrame using the Hugging Face datasets library:

from datasets import load_dataset

# Load the Latency results for Floating Point vectors
dataset = load_dataset("moorcheh/Benchmarks", split="latency_vector_fp")
df = dataset.to_pandas()

print(df.head())

Downloads last month: 11