Datasets:

allisonzz
/

embedding-migration-results

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 2 new columns ({'val_scores', 'best_alpha'})

This happened while the json dataset builder was generating data using

hf://datasets/allisonzz/embedding-migration-results/beir_fiqa_ridge.json (at revision 2c46ad92fc5aa04e6063849780de2504df39677e), ['hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/fewshot_curve.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/relative_repr.json'], ['hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/fewshot_curve.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/relative_repr.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)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1848, in _prepare_split_single
                  writer.write_table(table)
                  ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 765, in write_table
                  self._write_table(pa_table, writer_batch_size=writer_batch_size)
                  ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._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
              source: string
              target: string
              method: string
              method_label: string
              eval_dataset: string
              src_dim: int64
              tgt_dim: int64
              train_size: int64
              fit_time_s: double
              recall@10_translated: double
              recall@10_native_target: double
              ratio: double
              ndcg@10_translated: double
              ndcg@10_native_target: double
              ndcg_ratio: double
              best_alpha: double
              val_scores: struct<0.001: double, 0.01: double, 0.1: double, 1.0: double, 10.0: double, 100.0: double>
                child 0, 0.001: double
                child 1, 0.01: double
                child 2, 0.1: double
                child 3, 1.0: double
                child 4, 10.0: double
                child 5, 100.0: double
              to
              {'source': Value('string'), 'target': Value('string'), 'method': Value('string'), 'method_label': Value('string'), 'eval_dataset': Value('string'), 'src_dim': Value('int64'), 'tgt_dim': Value('int64'), 'train_size': Value('int64'), 'fit_time_s': Value('float64'), 'recall@10_translated': Value('float64'), 'recall@10_native_target': Value('float64'), 'ratio': Value('float64'), 'ndcg@10_translated': Value('float64'), 'ndcg@10_native_target': Value('float64'), 'ndcg_ratio': 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 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 1850, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
                  ...<4 lines>...
                  )
              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 ({'val_scores', 'best_alpha'})
              
              This happened while the json dataset builder was generating data using
              
              hf://datasets/allisonzz/embedding-migration-results/beir_fiqa_ridge.json (at revision 2c46ad92fc5aa04e6063849780de2504df39677e), ['hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/fewshot_curve.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/relative_repr.json'], ['hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_fiqa_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_nfcorpus_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/beir_scifact_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/fewshot_curve.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_procrustes.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/msmarco_ridge.json', 'hf://datasets/allisonzz/embedding-migration-results@2c46ad92fc5aa04e6063849780de2504df39677e/relative_repr.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)

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.

source string	target string	method string	method_label string	eval_dataset string	src_dim int64	tgt_dim int64	train_size int64	fit_time_s float64	recall@10_translated float64	recall@10_native_target float64	ratio float64	ndcg@10_translated float64	ndcg@10_native_target float64	ndcg_ratio float64
intfloat/e5-base-v2	BAAI/bge-base-en-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	6.423	0.2626	0.42673	0.61538	0.23583	0.40601	0.58085
intfloat/e5-base-v2	Alibaba-NLP/gte-base-en-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.045	0.03283	0.50703	0.06474	0.02442	0.48659	0.05018
intfloat/e5-base-v2	nomic-ai/nomic-embed-text-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.044	0.18933	0.3939	0.48065	0.17597	0.37417	0.47028
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	fiqa	768	1,024	5,000	0.047	0.22567	0.48124	0.46894	0.1965	0.46616	0.42154
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	fiqa	768	2,560	5,000	0.026	0.2544	0.59965	0.42424	0.22429	0.58402	0.38404
BAAI/bge-base-en-v1.5	intfloat/e5-base-v2	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.044	0.40563	0.41676	0.97328	0.38337	0.40052	0.95717
BAAI/bge-base-en-v1.5	Alibaba-NLP/gte-base-en-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.045	0.32884	0.50703	0.64855	0.30533	0.48659	0.62749
BAAI/bge-base-en-v1.5	nomic-ai/nomic-embed-text-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.045	0.32356	0.3939	0.82143	0.29553	0.37417	0.78982
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	fiqa	768	1,024	5,000	0.013	0.28664	0.48124	0.59562	0.25377	0.46616	0.54439
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	fiqa	768	2,560	5,000	0.025	0.32415	0.59965	0.54057	0.29326	0.58402	0.50213
Alibaba-NLP/gte-base-en-v1.5	intfloat/e5-base-v2	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.045	0.35932	0.41676	0.86217	0.32648	0.40052	0.81513
Alibaba-NLP/gte-base-en-v1.5	BAAI/bge-base-en-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.046	0.37339	0.42673	0.875	0.34012	0.40601	0.83771
Alibaba-NLP/gte-base-en-v1.5	nomic-ai/nomic-embed-text-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.046	0.34232	0.3939	0.86905	0.32071	0.37417	0.85712
Alibaba-NLP/gte-base-en-v1.5	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	fiqa	768	1,024	5,000	0.013	0.26202	0.48124	0.54446	0.22811	0.46616	0.48934
Alibaba-NLP/gte-base-en-v1.5	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	fiqa	768	2,560	5,000	0.025	0.28253	0.59965	0.47116	0.24931	0.58402	0.42689
nomic-ai/nomic-embed-text-v1.5	intfloat/e5-base-v2	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.044	0.3306	0.41676	0.79325	0.30866	0.40052	0.77064
nomic-ai/nomic-embed-text-v1.5	BAAI/bge-base-en-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.045	0.3306	0.42673	0.77473	0.30145	0.40601	0.74247
nomic-ai/nomic-embed-text-v1.5	Alibaba-NLP/gte-base-en-v1.5	procrustes	orthogonal_procrustes	fiqa	768	768	5,000	0.047	0.26143	0.50703	0.51561	0.23035	0.48659	0.47339
nomic-ai/nomic-embed-text-v1.5	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	fiqa	768	1,024	5,000	0.013	0.28312	0.48124	0.58831	0.25521	0.46616	0.54746
nomic-ai/nomic-embed-text-v1.5	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	fiqa	768	2,560	5,000	0.025	0.33587	0.59965	0.56012	0.30267	0.58402	0.51826
Qwen/Qwen3-Embedding-0.6B	intfloat/e5-base-v2	procrustes	least_squares	fiqa	1,024	768	5,000	0.015	0.27491	0.41676	0.65963	0.25859	0.40052	0.64564
Qwen/Qwen3-Embedding-0.6B	BAAI/bge-base-en-v1.5	procrustes	least_squares	fiqa	1,024	768	5,000	0.015	0.33353	0.42673	0.78159	0.30188	0.40601	0.74353
Qwen/Qwen3-Embedding-0.6B	Alibaba-NLP/gte-base-en-v1.5	procrustes	least_squares	fiqa	1,024	768	5,000	0.015	0.37925	0.50703	0.74798	0.35446	0.48659	0.72845
Qwen/Qwen3-Embedding-0.6B	nomic-ai/nomic-embed-text-v1.5	procrustes	least_squares	fiqa	1,024	768	5,000	0.015	0.28664	0.3939	0.72768	0.26093	0.37417	0.69737
Qwen/Qwen3-Embedding-0.6B	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	fiqa	1,024	2,560	5,000	0.041	0.48652	0.59965	0.81134	0.46139	0.58402	0.79002
Qwen/Qwen3-Embedding-4B	intfloat/e5-base-v2	procrustes	least_squares	fiqa	2,560	768	5,000	0.046	0.3347	0.41676	0.80309	0.29857	0.40052	0.74547
Qwen/Qwen3-Embedding-4B	BAAI/bge-base-en-v1.5	procrustes	least_squares	fiqa	2,560	768	5,000	0.046	0.33411	0.42673	0.78297	0.29871	0.40601	0.73573
Qwen/Qwen3-Embedding-4B	Alibaba-NLP/gte-base-en-v1.5	procrustes	least_squares	fiqa	2,560	768	5,000	0.046	0.39683	0.50703	0.78266	0.35856	0.48659	0.73688
Qwen/Qwen3-Embedding-4B	nomic-ai/nomic-embed-text-v1.5	procrustes	least_squares	fiqa	2,560	768	5,000	0.046	0.29543	0.3939	0.75	0.27777	0.37417	0.74237
Qwen/Qwen3-Embedding-4B	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	fiqa	2,560	1,024	5,000	0.046	0.45311	0.48124	0.94153	0.4345	0.46616	0.93209
intfloat/e5-base-v2	BAAI/bge-base-en-v1.5	ridge	ridge(α=0.1)	fiqa	768	768	5,000	1.425	0.35815	0.42673	0.83929	0.32161	0.40601	0.79213
intfloat/e5-base-v2	Alibaba-NLP/gte-base-en-v1.5	ridge	ridge(α=1.0)	fiqa	768	768	5,000	1.043	0.29426	0.50703	0.58035	0.25776	0.48659	0.52972
intfloat/e5-base-v2	nomic-ai/nomic-embed-text-v1.5	ridge	ridge(α=0.1)	fiqa	768	768	5,000	0.99	0.23154	0.3939	0.5878	0.20867	0.37417	0.55769
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-0.6B	ridge	ridge(α=1.0)	fiqa	768	1,024	5,000	1.006	0.1524	0.48124	0.31669	0.12337	0.46616	0.26464
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-4B	ridge	ridge(α=1.0)	fiqa	768	2,560	5,000	1.133	0.14771	0.59965	0.24633	0.1254	0.58402	0.21472
BAAI/bge-base-en-v1.5	intfloat/e5-base-v2	ridge	ridge(α=0.1)	fiqa	768	768	5,000	1.053	0.36108	0.41676	0.86639	0.33896	0.40052	0.84631
BAAI/bge-base-en-v1.5	Alibaba-NLP/gte-base-en-v1.5	ridge	ridge(α=1.0)	fiqa	768	768	5,000	1.165	0.35053	0.50703	0.69133	0.32358	0.48659	0.66499
BAAI/bge-base-en-v1.5	nomic-ai/nomic-embed-text-v1.5	ridge	ridge(α=1.0)	fiqa	768	768	5,000	0.983	0.25674	0.3939	0.65179	0.22333	0.37417	0.59687
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-0.6B	ridge	ridge(α=1.0)	fiqa	768	1,024	5,000	1.058	0.25615	0.48124	0.53228	0.22121	0.46616	0.47453
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-4B	ridge	ridge(α=1.0)	fiqa	768	2,560	5,000	1.295	0.27081	0.59965	0.45161	0.24447	0.58402	0.4186
Alibaba-NLP/gte-base-en-v1.5	intfloat/e5-base-v2	ridge	ridge(α=1.0)	fiqa	768	768	5,000	1.008	0.23564	0.41676	0.5654	0.21595	0.40052	0.53918
Alibaba-NLP/gte-base-en-v1.5	BAAI/bge-base-en-v1.5	ridge	ridge(α=1.0)	fiqa	768	768	5,000	1.137	0.30891	0.42673	0.7239	0.27482	0.40601	0.67689
Alibaba-NLP/gte-base-en-v1.5	nomic-ai/nomic-embed-text-v1.5	ridge	ridge(α=1.0)	fiqa	768	768	5,000	0.909	0.22392	0.3939	0.56845	0.20997	0.37417	0.56116
Alibaba-NLP/gte-base-en-v1.5	Qwen/Qwen3-Embedding-0.6B	ridge	ridge(α=1.0)	fiqa	768	1,024	5,000	1.198	0.21805	0.48124	0.45311	0.18563	0.46616	0.39822
Alibaba-NLP/gte-base-en-v1.5	Qwen/Qwen3-Embedding-4B	ridge	ridge(α=1.0)	fiqa	768	2,560	5,000	1.367	0.21981	0.59965	0.36657	0.19028	0.58402	0.32581
nomic-ai/nomic-embed-text-v1.5	intfloat/e5-base-v2	ridge	ridge(α=0.1)	fiqa	768	768	5,000	1.332	0.2925	0.41676	0.70183	0.26748	0.40052	0.66784
nomic-ai/nomic-embed-text-v1.5	BAAI/bge-base-en-v1.5	ridge	ridge(α=0.1)	fiqa	768	768	5,000	1.233	0.32063	0.42673	0.75137	0.29319	0.40601	0.72212
nomic-ai/nomic-embed-text-v1.5	Alibaba-NLP/gte-base-en-v1.5	ridge	ridge(α=0.1)	fiqa	768	768	5,000	1.15	0.35463	0.50703	0.69942	0.32321	0.48659	0.66423
nomic-ai/nomic-embed-text-v1.5	Qwen/Qwen3-Embedding-0.6B	ridge	ridge(α=1.0)	fiqa	768	1,024	5,000	0.915	0.24267	0.48124	0.50426	0.21353	0.46616	0.45805
nomic-ai/nomic-embed-text-v1.5	Qwen/Qwen3-Embedding-4B	ridge	ridge(α=1.0)	fiqa	768	2,560	5,000	2.795	0.25264	0.59965	0.42131	0.23418	0.58402	0.40098
Qwen/Qwen3-Embedding-0.6B	intfloat/e5-base-v2	ridge	ridge(α=1.0)	fiqa	1,024	768	5,000	2.937	0.26671	0.41676	0.63994	0.24603	0.40052	0.61429
Qwen/Qwen3-Embedding-0.6B	BAAI/bge-base-en-v1.5	ridge	ridge(α=1.0)	fiqa	1,024	768	5,000	2.623	0.31712	0.42673	0.74313	0.28719	0.40601	0.70736
Qwen/Qwen3-Embedding-0.6B	Alibaba-NLP/gte-base-en-v1.5	ridge	ridge(α=1.0)	fiqa	1,024	768	5,000	2.328	0.37515	0.50703	0.73988	0.34678	0.48659	0.71266
Qwen/Qwen3-Embedding-0.6B	nomic-ai/nomic-embed-text-v1.5	ridge	ridge(α=1.0)	fiqa	1,024	768	5,000	1.397	0.28312	0.3939	0.71875	0.25042	0.37417	0.66926
Qwen/Qwen3-Embedding-0.6B	Qwen/Qwen3-Embedding-4B	ridge	ridge(α=1.0)	fiqa	1,024	2,560	5,000	1.564	0.45838	0.59965	0.76442	0.43182	0.58402	0.73939
Qwen/Qwen3-Embedding-4B	intfloat/e5-base-v2	ridge	ridge(α=1.0)	fiqa	2,560	768	5,000	2.008	0.31477	0.41676	0.75527	0.28326	0.40052	0.70723
Qwen/Qwen3-Embedding-4B	BAAI/bge-base-en-v1.5	ridge	ridge(α=1.0)	fiqa	2,560	768	5,000	2.073	0.36342	0.42673	0.85165	0.32525	0.40601	0.80109
Qwen/Qwen3-Embedding-4B	Alibaba-NLP/gte-base-en-v1.5	ridge	ridge(α=1.0)	fiqa	2,560	768	5,000	2.084	0.42497	0.50703	0.83815	0.38634	0.48659	0.79396
Qwen/Qwen3-Embedding-4B	nomic-ai/nomic-embed-text-v1.5	ridge	ridge(α=1.0)	fiqa	2,560	768	5,000	2.066	0.29894	0.3939	0.75893	0.27807	0.37417	0.74316
Qwen/Qwen3-Embedding-4B	Qwen/Qwen3-Embedding-0.6B	ridge	ridge(α=1.0)	fiqa	2,560	1,024	5,000	2.212	0.45662	0.48124	0.94884	0.43553	0.46616	0.9343
intfloat/e5-base-v2	BAAI/bge-base-en-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	6.484	0.05319	0.07346	0.72406	0.27193	0.3757	0.72379
intfloat/e5-base-v2	Alibaba-NLP/gte-base-en-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.045	0.02246	0.06875	0.32665	0.10784	0.35949	0.29997
intfloat/e5-base-v2	nomic-ai/nomic-embed-text-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.044	0.04086	0.06737	0.6065	0.20313	0.34794	0.58381
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	nfcorpus	768	1,024	5,000	0.041	0.04759	0.06875	0.69222	0.23156	0.35642	0.64968
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	nfcorpus	768	2,560	5,000	0.025	0.0514	0.07864	0.65361	0.25396	0.40618	0.62524
BAAI/bge-base-en-v1.5	intfloat/e5-base-v2	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.044	0.06478	0.06924	0.9356	0.32902	0.35503	0.92673
BAAI/bge-base-en-v1.5	Alibaba-NLP/gte-base-en-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.045	0.04929	0.06875	0.71698	0.25274	0.35949	0.70304
BAAI/bge-base-en-v1.5	nomic-ai/nomic-embed-text-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.044	0.05886	0.06737	0.87365	0.29972	0.34794	0.86141
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	nfcorpus	768	1,024	5,000	0.013	0.051	0.06875	0.74175	0.25406	0.35642	0.71281
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	nfcorpus	768	2,560	5,000	0.024	0.05586	0.07864	0.71031	0.27433	0.40618	0.67539
Alibaba-NLP/gte-base-en-v1.5	intfloat/e5-base-v2	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.045	0.05278	0.06924	0.7623	0.27091	0.35503	0.76307
Alibaba-NLP/gte-base-en-v1.5	BAAI/bge-base-en-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.046	0.05659	0.07346	0.77042	0.28476	0.3757	0.75794
Alibaba-NLP/gte-base-en-v1.5	nomic-ai/nomic-embed-text-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.046	0.05343	0.06737	0.79302	0.26811	0.34794	0.77055
Alibaba-NLP/gte-base-en-v1.5	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	nfcorpus	768	1,024	5,000	0.013	0.03932	0.06875	0.57193	0.18741	0.35642	0.5258
Alibaba-NLP/gte-base-en-v1.5	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	nfcorpus	768	2,560	5,000	0.025	0.04378	0.07864	0.5567	0.21613	0.40618	0.53211
nomic-ai/nomic-embed-text-v1.5	intfloat/e5-base-v2	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.044	0.05602	0.06924	0.80913	0.27732	0.35503	0.78111
nomic-ai/nomic-embed-text-v1.5	BAAI/bge-base-en-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.045	0.05473	0.07346	0.74503	0.26897	0.3757	0.71592
nomic-ai/nomic-embed-text-v1.5	Alibaba-NLP/gte-base-en-v1.5	procrustes	orthogonal_procrustes	nfcorpus	768	768	5,000	0.046	0.04086	0.06875	0.59434	0.20769	0.35949	0.57773
nomic-ai/nomic-embed-text-v1.5	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	nfcorpus	768	1,024	5,000	0.013	0.04929	0.06875	0.71698	0.23881	0.35642	0.67001
nomic-ai/nomic-embed-text-v1.5	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	nfcorpus	768	2,560	5,000	0.024	0.05197	0.07864	0.66082	0.25037	0.40618	0.6164
Qwen/Qwen3-Embedding-0.6B	intfloat/e5-base-v2	procrustes	least_squares	nfcorpus	1,024	768	5,000	0.015	0.05035	0.06924	0.72717	0.25067	0.35503	0.70604
Qwen/Qwen3-Embedding-0.6B	BAAI/bge-base-en-v1.5	procrustes	least_squares	nfcorpus	1,024	768	5,000	0.015	0.05197	0.07346	0.70751	0.26277	0.3757	0.69941
Qwen/Qwen3-Embedding-0.6B	Alibaba-NLP/gte-base-en-v1.5	procrustes	least_squares	nfcorpus	1,024	768	5,000	0.015	0.05002	0.06875	0.72759	0.25114	0.35949	0.69859
Qwen/Qwen3-Embedding-0.6B	nomic-ai/nomic-embed-text-v1.5	procrustes	least_squares	nfcorpus	1,024	768	5,000	0.015	0.04881	0.06737	0.72443	0.24499	0.34794	0.70412
Qwen/Qwen3-Embedding-0.6B	Qwen/Qwen3-Embedding-4B	procrustes	least_squares	nfcorpus	1,024	2,560	5,000	0.035	0.06446	0.07864	0.81959	0.33832	0.40618	0.83294
Qwen/Qwen3-Embedding-4B	intfloat/e5-base-v2	procrustes	least_squares	nfcorpus	2,560	768	5,000	0.051	0.0604	0.06924	0.87237	0.30249	0.35503	0.852
Qwen/Qwen3-Embedding-4B	BAAI/bge-base-en-v1.5	procrustes	least_squares	nfcorpus	2,560	768	5,000	0.046	0.06275	0.07346	0.8543	0.31819	0.3757	0.84693
Qwen/Qwen3-Embedding-4B	Alibaba-NLP/gte-base-en-v1.5	procrustes	least_squares	nfcorpus	2,560	768	5,000	0.046	0.05692	0.06875	0.82783	0.28241	0.35949	0.78558
Qwen/Qwen3-Embedding-4B	nomic-ai/nomic-embed-text-v1.5	procrustes	least_squares	nfcorpus	2,560	768	5,000	0.046	0.05554	0.06737	0.82431	0.27922	0.34794	0.80251
Qwen/Qwen3-Embedding-4B	Qwen/Qwen3-Embedding-0.6B	procrustes	least_squares	nfcorpus	2,560	1,024	5,000	0.047	0.06948	0.06875	1.01061	0.35913	0.35642	1.00759
intfloat/e5-base-v2	BAAI/bge-base-en-v1.5	ridge	ridge(α=0.1)	nfcorpus	768	768	5,000	1.537	0.06243	0.07346	0.84989	0.31893	0.3757	0.84891
intfloat/e5-base-v2	Alibaba-NLP/gte-base-en-v1.5	ridge	ridge(α=1.0)	nfcorpus	768	768	5,000	1.108	0.04524	0.06875	0.65802	0.2266	0.35949	0.63034
intfloat/e5-base-v2	nomic-ai/nomic-embed-text-v1.5	ridge	ridge(α=0.1)	nfcorpus	768	768	5,000	1.264	0.04678	0.06737	0.69434	0.23952	0.34794	0.68841
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-0.6B	ridge	ridge(α=1.0)	nfcorpus	768	1,024	5,000	1.068	0.04378	0.06875	0.63679	0.21083	0.35642	0.59151
intfloat/e5-base-v2	Qwen/Qwen3-Embedding-4B	ridge	ridge(α=1.0)	nfcorpus	768	2,560	5,000	1.1	0.04597	0.07864	0.58454	0.22791	0.40618	0.56111
BAAI/bge-base-en-v1.5	intfloat/e5-base-v2	ridge	ridge(α=0.1)	nfcorpus	768	768	5,000	1.057	0.06227	0.06924	0.8993	0.32094	0.35503	0.90396
BAAI/bge-base-en-v1.5	Alibaba-NLP/gte-base-en-v1.5	ridge	ridge(α=1.0)	nfcorpus	768	768	5,000	0.973	0.05213	0.06875	0.75825	0.26888	0.35949	0.74793
BAAI/bge-base-en-v1.5	nomic-ai/nomic-embed-text-v1.5	ridge	ridge(α=1.0)	nfcorpus	768	768	5,000	0.966	0.0497	0.06737	0.73767	0.25727	0.34794	0.7394
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-0.6B	ridge	ridge(α=1.0)	nfcorpus	768	1,024	5,000	0.879	0.05059	0.06875	0.73585	0.24903	0.35642	0.69868
BAAI/bge-base-en-v1.5	Qwen/Qwen3-Embedding-4B	ridge	ridge(α=1.0)	nfcorpus	768	2,560	5,000	1.295	0.05383	0.07864	0.68454	0.26657	0.40618	0.65629

End of preview.

Embedding Migration Results

Experimental results from testing whether vector databases can be migrated to new embedding models without re-embedding the entire corpus.

Overview

We embedded ~1M MS MARCO passages with 6 embedding models across 3 dimensionalities (768, 1024, 2560), trained linear translators between every pair of embedding spaces, and measured recall@10 ratio (translated / native ceiling).

Models

Model	Dim	Prefix
intfloat/e5-base-v2	768	query/passage
BAAI/bge-base-en-v1.5	768	query only
Alibaba-NLP/gte-base-en-v1.5	768	none
nomic-ai/nomic-embed-text-v1.5	768	search_query/search_document
Qwen/Qwen3-Embedding-0.6B	1024	instruction format
Qwen/Qwen3-Embedding-4B	2560	instruction format

Files

File	Evaluations	Description
`msmarco_procrustes.json`	30	Orthogonal Procrustes / least-squares on MS MARCO 1M
`msmarco_ridge.json`	30	Ridge regression on MS MARCO 1M
`beir_scifact_procrustes.json`	30	Procrustes/LS evaluated on BEIR SciFact
`beir_scifact_ridge.json`	30	Ridge evaluated on BEIR SciFact
`beir_fiqa_procrustes.json`	30	Procrustes/LS evaluated on BEIR FiQA
`beir_fiqa_ridge.json`	30	Ridge evaluated on BEIR FiQA
`beir_nfcorpus_procrustes.json`	30	Procrustes/LS evaluated on BEIR NFCorpus
`beir_nfcorpus_ridge.json`	30	Ridge evaluated on BEIR NFCorpus
`fewshot_curve.json`	84	Few-shot learning curve (50-5000 training examples)
`relative_repr.json`	30	Relative representations baseline (Moschella et al. 2023)

Total: 354 evaluations

Key Results

50 paired examples is enough for 95%+ native performance on compatible model pairs
Prefix mismatch (not dimension mismatch) is the dominant failure mode
Ridge regression rescues prefix-mismatched pairs (E5→GTE: 0.090 → 0.814)
Cross-domain generalization holds for same-family pairs, degrades 10-30% for mismatched pairs
Relative representations fail at retrieval scale (negative result)

Schema

Each JSON file contains an array of evaluation records. Common fields:

{
  "source": "model/name",
  "target": "model/name",
  "method": "procrustes|ridge|relative_repr",
  "src_dim": 768,
  "tgt_dim": 768,
  "recall@10_translated": 0.862,
  "recall@10_native_target": 0.885,
  "ratio": 0.974,
  "train_size": 5000
}

Citation

If you use these results, please link the GitHub repository: https://github.com/allison-stack/embedding-migration

License

MIT

Downloads last month: 21

Total file size:

207 kB