Shadowgraph
TrendingAbout
Sign in

lightonai/GTE-ModernColBERT-v1

Model

GTE-ModernColBERT-v1

178

15 commits

2 linked in READMEs

updated Aug 17, 2026

See the code

README

GTE-ModernColBERT-v1

Multi-vector embedding model based on Alibaba-NLP/gte-modernbert-base

This is a multi-vector (ColBERT-style late interaction) embedding model trained on the ms-marco-en-bge-gemma dataset. It maps sentences & paragraphs to sequences of 128-dimensional dense vectors and can be used for semantic textual similarity using the MaxSim operator.

Model Details

Model Description

  • Model Type: Multi-vector embedding model
  • Base model: Alibaba-NLP/gte-modernbert-base
  • Document Length: 300 tokens
  • Query Length: 32 tokens
  • Output Dimensionality: 128 dimensions
  • Similarity Function: MaxSim
  • Training Dataset:
  • Language: English
  • License: Apache 2.0

Document length

GTE-ModernColBERT has been trained with knowledge distillation on MS MARCO with a document length of 300 tokens, explaining its default value for documents length.

However, as illustrated in the ModernBERT paper, ColBERT models can generalize to documents lengths way beyond their training length and GTE-ModernColBERT actually yields results way above SOTA in long-context embedding benchmarks, see LongEmbed results.

Simply change adapt the document length parameter to your needs when loading the model:

model = models.ColBERT(
    model_name_or_path=lightonai/GTE-ModernColBERT-v1,
    document_length=8192,
)

ModernBERT itself has only been trained on 8K context length, but it seems that GTE-ModernColBERT can generalize to even bigger context sizes, though it is not guaranteed so please perform your own benches!

Model Sources

Full Model Architecture

ColBERT(
  (0): Transformer({'max_seq_length': 299, 'do_lower_case': False}) with Transformer model: ModernBertModel 
  (1): Dense({'in_features': 768, 'out_features': 128, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

Usage

Sentence Transformers

This model can be used with Sentence Transformers as a multi-vector (ColBERT-style late interaction) retriever via the MultiVectorEncoder:

pip install "sentence-transformers>=6.0.0"

from sentence_transformers import MultiVectorEncoder

model = MultiVectorEncoder("lightonai/GTE-ModernColBERT-v1")

query = "Which planet is known as the Red Planet?"
documents = [
    "Venus is often called Earth's twin because of its similar size and proximity.",
    "Mars, known for its reddish appearance, is often referred to as the Red Planet.",
    "Jupiter, the largest planet in our solar system, has a prominent red spot.",
    "Saturn, famous for its rings, is sometimes mistaken for the Red Planet.",
]

query_embeddings = model.encode_query(query)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings[0].shape)
# (12, 128) (18, 128)

# MaxSim late-interaction scoring (higher is more relevant)
scores = model.similarity(query_embeddings, document_embeddings)
print(scores)
# tensor([[11.2577, 11.5113, 11.3457, 11.4518]])

PyLate

First install the PyLate library:

pip install -U pylate

Retrieval

PyLate provides a streamlined interface to index and retrieve documents using ColBERT models. The index leverages the Voyager HNSW index to efficiently handle document embeddings and enable fast retrieval.

Indexing documents

First, load the ColBERT model and initialize the Voyager index, then encode and index your documents:

from pylate import indexes, models, retrieve

# Step 1: Load the ColBERT model
model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

# Step 2: Initialize the Voyager index
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
    override=True,  # This overwrites the existing index if any
)

# Step 3: Encode the documents
documents_ids = ["1", "2", "3"]
documents = ["document 1 text", "document 2 text", "document 3 text"]

documents_embeddings = model.encode(
    documents,
    batch_size=32,
    is_query=False,  # Ensure that it is set to False to indicate that these are documents, not queries
    show_progress_bar=True,
)

# Step 4: Add document embeddings to the index by providing embeddings and corresponding ids
index.add_documents(
    documents_ids=documents_ids,
    documents_embeddings=documents_embeddings,
)

Note that you do not have to recreate the index and encode the documents every time. Once you have created an index and added the documents, you can re-use the index later by loading it:

# To load an index, simply instantiate it with the correct folder/name and without overriding it
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
)

Retrieving top-k documents for queries

Once the documents are indexed, you can retrieve the top-k most relevant documents for a given set of queries. To do so, initialize the ColBERT retriever with the index you want to search in, encode the queries and then retrieve the top-k documents to get the top matches ids and relevance scores:

# Step 1: Initialize the ColBERT retriever
retriever = retrieve.ColBERT(index=index)

# Step 2: Encode the queries
queries_embeddings = model.encode(
    ["query for document 3", "query for document 1"],
    batch_size=32,
    is_query=True,  #  # Ensure that it is set to False to indicate that these are queries
    show_progress_bar=True,
)

# Step 3: Retrieve top-k documents
scores = retriever.retrieve(
    queries_embeddings=queries_embeddings, 
    k=10,  # Retrieve the top 10 matches for each query
)

Reranking

If you only want to use the ColBERT model to perform reranking on top of your first-stage retrieval pipeline without building an index, you can simply use rank function and pass the queries and documents to rerank:

from pylate import rank, models

queries = [
    "query A",
    "query B",
]

documents = [
    ["document A", "document B"],
    ["document 1", "document C", "document B"],
]

documents_ids = [
    [1, 2],
    [1, 3, 2],
]

model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

queries_embeddings = model.encode(
    queries,
    is_query=True,
)

documents_embeddings = model.encode(
    documents,
    is_query=False,
)

reranked_documents = rank.rerank(
    documents_ids=documents_ids,
    queries_embeddings=queries_embeddings,
    documents_embeddings=documents_embeddings,
)

Evaluation

Metrics

BEIR Benchmark

GTE-ModernColBERT is the first model to outpeform ColBERT-small on the BEIR benchmark. As reproduction in the IR domain is challenging, we worked closely with Benjamin Clavié, the author of ColBERT-small to reproduce the evaluation setup of this model. Despite all these efforts and reducing to the maximum the difference in scores in most of the datasets, some are still a bit different. For this reason, we also report the results of ColBERT-small in the same setup we used to evaluate GTE-ModernColBERT for completness and fair comparison.

ModelAverageFiQA2018NFCorpusTREC-COVIDTouche2020ArguAnaQuoraRetrievalSCIDOCSSciFactNQClimateFEVERHotpotQADBPediaCQADupstackFEVERMSMARCO
GTE-ModernColBERT54.6745.2837.9383.5931.2348.5186.6119.0676.3461.830.6277.3248.034187.4445.32
ColBERT-small (reported)53.7941.1537.384.5925.6950.0987.7218.4274.7759.133.0776.1145.5838.7590.9643.5
JinaColBERT-v240.834.683.427.436.688.718.667.86423.976.647.180.5
ColBERT-small (rerunned)53.3541.0136.8683.1424.9546.7687.8918.7274.0259.4232.8376.8846.3639.3688.6643.44

LongEmbed Benchmark

GTE-ModernColBERT has been trained with knowledge distillation on MS MARCO with a document length of 300 tokens, explaining its default value for documents length. However, as illustrated in the ModernBERT paper, ColBERT models can generalize to documents lengths way beyond their training length and GTE-ModernColBERT actually yields results way above SOTA (almost 10 points above previous SOTA) in long-context embedding benchmark:

ModelMeanLEMBNarrativeQARetrievalLEMBNeedleRetrievalLEMBPasskeyRetrievalLEMBQMSumRetrievalLEMBSummScreenFDRetrievalLEMBWikimQARetrieval
GTE-ModernColBERT (with 32k document length)88.3978.8292.59272.1794.9899.87
voyage-multilingual-279.1764.69475.259751.49599.10587.489
inf-retriever-v173.1960.70261.578.7555.07297.38785.751
snowflake-arctic-embed-l-v2,063.7343.63250.2577.2540.0496.38374.843
gte-multilingual-base62.1252.35842.2555.543.03395.49984.078
jasper_en_vision_language_v160.9337.9285562.2541.18697.20672.025
bge-m358.7345.76140.255935.54394.08977.726
jina-embeddings-v355.6634.297643839.33792.33466.018
e5-base-4k54.5130.0337.7565.2531.26893.86868.875
gte-Qwen2-7B-instruct47.2445.463138.531.27276.0861.151

ModernBERT itself has only been trained on 8K context length, but it seems that GTE-ModernColBERT can generalize to even bigger context sizes, though it is not guaranteed so please perform your own benches!

PyLate Information Retrieval

  • Datasets: NanoClimateFEVER, NanoDBPedia, NanoFEVER, NanoFiQA2018, NanoHotpotQA, NanoMSMARCO, NanoNFCorpus, NanoNQ, NanoQuoraRetrieval, NanoSCIDOCS, NanoArguAna, NanoSciFact and NanoTouche2020
  • Evaluated with pylate.evaluation.pylate_information_retrieval_evaluator.PyLateInformationRetrievalEvaluator
MetricNanoClimateFEVERNanoDBPediaNanoFEVERNanoFiQA2018NanoHotpotQANanoMSMARCONanoNFCorpusNanoNQNanoQuoraRetrievalNanoSCIDOCSNanoArguAnaNanoSciFactNanoTouche2020
MaxSim_accuracy@10.360.880.920.560.920.540.560.640.960.480.30.740.7755
MaxSim_accuracy@30.620.940.980.661.00.680.680.821.00.740.620.860.9388
MaxSim_accuracy@50.780.960.980.741.00.740.740.861.00.780.70.90.9796
MaxSim_accuracy@100.860.981.00.81.00.920.760.91.00.840.820.940.9796
MaxSim_precision@10.360.880.920.560.920.540.560.640.960.480.30.740.7755
MaxSim_precision@30.23330.71330.360.32670.580.22670.43330.28670.40.40.20670.30.6599
MaxSim_precision@50.2080.6560.2160.2560.360.1480.3920.180.2560.2920.140.1960.6571
MaxSim_precision@100.1280.5720.110.1520.1860.0920.3040.10.1340.1940.0820.1040.5184
MaxSim_recall@10.18330.1180.85670.30920.460.540.06640.610.84730.10070.30.7150.0518
MaxSim_recall@30.2890.23070.960.47840.870.680.1020.780.94530.24670.620.830.1362
MaxSim_recall@50.41570.29620.960.57520.90.740.12840.820.96930.29970.70.8850.2193
MaxSim_recall@100.49570.41460.980.64120.930.920.15660.880.98930.39670.820.930.334
MaxSim_ndcg@100.41480.72960.94520.5670.90120.70890.39570.76450.96910.39870.56090.83720.5927
MaxSim_mrr@100.52660.91690.95220.63590.960.64470.6270.7390.97670.61370.47750.81170.8629
MaxSim_map@1000.33470.58840.92710.50320.85920.64960.19180.72390.95520.31630.48240.80490.4257

Nano BEIR

  • Dataset: NanoBEIR_mean
  • Evaluated with pylate.evaluation.nano_beir_evaluator.NanoBEIREvaluator
MetricValue
MaxSim_accuracy@10.6643
MaxSim_accuracy@30.8107
MaxSim_accuracy@50.8584
MaxSim_accuracy@100.9077
MaxSim_precision@10.6643
MaxSim_precision@30.3943
MaxSim_precision@50.3044
MaxSim_precision@100.2059
MaxSim_recall@10.3968
MaxSim_recall@30.5514
MaxSim_recall@50.6084
MaxSim_recall@100.6837
MaxSim_ndcg@100.6758
MaxSim_mrr@100.7496
MaxSim_map@1000.5971

Training Details

Training Hyperparameters

Non-Default Hyperparameters

  • eval_strategy: steps
  • per_device_train_batch_size: 16
  • learning_rate: 3e-05
  • bf16: True

All Hyperparameters

Click to expand
  • overwrite_output_dir: False
  • do_predict: False
  • eval_strategy: steps
  • prediction_loss_only: True
  • per_device_train_batch_size: 16
  • per_device_eval_batch_size: 8
  • per_gpu_train_batch_size: None
  • per_gpu_eval_batch_size: None
  • gradient_accumulation_steps: 1
  • eval_accumulation_steps: None
  • torch_empty_cache_steps: None
  • learning_rate: 3e-05
  • weight_decay: 0.0
  • adam_beta1: 0.9
  • adam_beta2: 0.999
  • adam_epsilon: 1e-08
  • max_grad_norm: 1.0
  • num_train_epochs: 3
  • max_steps: -1
  • lr_scheduler_type: linear
  • lr_scheduler_kwargs: {}
  • warmup_ratio: 0.0
  • warmup_steps: 0
  • log_level: passive
  • log_level_replica: warning
  • log_on_each_node: True
  • logging_nan_inf_filter: True
  • save_safetensors: True
  • save_on_each_node: False
  • save_only_model: False
  • restore_callback_states_from_checkpoint: False
  • no_cuda: False
  • use_cpu: False
  • use_mps_device: False
  • seed: 42
  • data_seed: None
  • jit_mode_eval: False
  • use_ipex: False
  • bf16: True
  • fp16: False
  • fp16_opt_level: O1
  • half_precision_backend: auto
  • bf16_full_eval: False
  • fp16_full_eval: False
  • tf32: None
  • local_rank: 6
  • ddp_backend: None
  • tpu_num_cores: None
  • tpu_metrics_debug: False
  • debug: []
  • dataloader_drop_last: True
  • dataloader_num_workers: 0
  • dataloader_prefetch_factor: None
  • past_index: -1
  • disable_tqdm: False
  • remove_unused_columns: True
  • label_names: None
  • load_best_model_at_end: False
  • ignore_data_skip: False
  • fsdp: []
  • fsdp_min_num_params: 0
  • fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
  • fsdp_transformer_layer_cls_to_wrap: None
  • accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
  • deepspeed: None
  • label_smoothing_factor: 0.0
  • optim: adamw_torch
  • optim_args: None
  • adafactor: False
  • group_by_length: False
  • length_column_name: length
  • ddp_find_unused_parameters: None
  • ddp_bucket_cap_mb: None
  • ddp_broadcast_buffers: False
  • dataloader_pin_memory: True
  • dataloader_persistent_workers: False
  • skip_memory_metrics: True
  • use_legacy_prediction_loop: False
  • push_to_hub: False
  • resume_from_checkpoint: None
  • hub_model_id: None
  • hub_strategy: every_save
  • hub_private_repo: None
  • hub_always_push: False
  • gradient_checkpointing: False
  • gradient_checkpointing_kwargs: None
  • include_inputs_for_metrics: False
  • include_for_metrics: []
  • eval_do_concat_batches: True
  • fp16_backend: auto
  • push_to_hub_model_id: None
  • push_to_hub_organization: None
  • mp_parameters:
  • auto_find_batch_size: False
  • full_determinism: False
  • torchdynamo: None
  • ray_scope: last
  • ddp_timeout: 1800
  • torch_compile: False
  • torch_compile_backend: None
  • torch_compile_mode: None
  • dispatch_batches: None
  • split_batches: None
  • include_tokens_per_second: False
  • include_num_input_tokens_seen: False
  • neftune_noise_alpha: None
  • optim_target_modules: None
  • batch_eval_metrics: False
  • eval_on_start: False
  • use_liger_kernel: False
  • eval_use_gather_object: False
  • average_tokens_across_devices: False
  • prompts: None
  • batch_sampler: batch_sampler
  • multi_dataset_batch_sampler: proportional

Training Logs

Click to expand
EpochStepTraining LossNanoClimateFEVER_MaxSim_ndcg@10NanoDBPedia_MaxSim_ndcg@10NanoFEVER_MaxSim_ndcg@10NanoFiQA2018_MaxSim_ndcg@10NanoHotpotQA_MaxSim_ndcg@10NanoMSMARCO_MaxSim_ndcg@10NanoNFCorpus_MaxSim_ndcg@10NanoNQ_MaxSim_ndcg@10NanoQuoraRetrieval_MaxSim_ndcg@10NanoSCIDOCS_MaxSim_ndcg@10NanoArguAna_MaxSim_ndcg@10NanoSciFact_MaxSim_ndcg@10NanoTouche2020_MaxSim_ndcg@10NanoBEIR_mean_MaxSim_ndcg@10
0.004200.0493--------------
0.008400.0434--------------
0.012600.0324--------------
0.016800.0238--------------
0.021000.0202--------------
0.0241200.0186--------------
0.0281400.0172--------------
0.0321600.0164--------------
0.0361800.0157--------------
0.042000.0153--------------
0.0442200.0145--------------
0.0482400.014--------------
0.0522600.0138--------------
0.0562800.0135--------------
0.063000.0132--------------
0.0643200.0129--------------
0.0683400.0126--------------
0.0723600.0123--------------
0.0763800.0122--------------
0.084000.012--------------
0.0844200.0121--------------
0.0884400.0115--------------
0.0924600.0113--------------
0.0964800.0112--------------
0.15000.01110.30850.63090.92060.53030.86180.68930.37030.71630.95480.38850.46820.79300.59820.6331
0.1045200.0109--------------
0.1085400.0109--------------
0.1125600.0109--------------
0.1165800.0105--------------
0.126000.0102--------------
0.1246200.0104--------------
0.1286400.0103--------------
0.1326600.01--------------
0.1366800.0101--------------
0.147000.0098--------------
0.1447200.0097--------------
0.1487400.0097--------------
0.1527600.0096--------------
0.1567800.0096--------------
0.168000.0094--------------
0.1648200.0096--------------
0.1688400.0095--------------
0.1728600.0093--------------
0.1768800.0092--------------
0.189000.0093--------------
0.1849200.009--------------
0.1889400.009--------------
0.1929600.0089--------------
0.1969800.0089--------------
0.210000.00890.31480.65860.93350.53740.88100.68050.37460.73680.94860.39550.48240.82190.60890.6442
0.20410200.0088--------------
0.20810400.0089--------------
0.21210600.0088--------------
0.21610800.0086--------------
0.2211000.0087--------------
0.22411200.0088--------------
0.22811400.0086--------------
0.23211600.0086--------------
0.23611800.0084--------------
0.2412000.0086--------------
0.24412200.0085--------------
0.24812400.0084--------------
0.25212600.0084--------------
0.25612800.0081--------------
0.2613000.0083--------------
0.26413200.0084--------------
0.26813400.0082--------------
0.27213600.0082--------------
0.27613800.008--------------
0.2814000.0078--------------
0.28414200.0079--------------
0.28814400.0078--------------
0.29214600.0081--------------
0.29614800.0081--------------
0.315000.00790.35100.65900.92850.54630.88930.68530.38000.73700.95130.39800.52680.82680.61300.6533
0.30415200.0078--------------
0.30815400.0078--------------
0.31215600.0077--------------
0.31615800.0078--------------
0.3216000.0078--------------
0.32416200.0078--------------
0.32816400.0078--------------
0.33216600.0076--------------
0.33616800.0076--------------
0.3417000.0077--------------
0.34417200.0076--------------
0.34817400.0074--------------
0.35217600.0074--------------
0.35617800.0075--------------
0.3618000.0076--------------
0.36418200.0075--------------
0.36818400.0073--------------
0.37218600.0075--------------
0.37618800.0073--------------
0.3819000.0074--------------
0.38419200.0072--------------
0.38819400.0072--------------
0.39219600.0071--------------
0.39619800.0073--------------
0.420000.00710.35510.68070.93110.53400.89510.70190.37670.74600.95590.39120.51210.82450.60580.6546
0.40420200.0073--------------
0.40820400.0072--------------
0.41220600.0071--------------
0.41620800.0073--------------
0.4221000.0069--------------
0.42421200.0071--------------
0.42821400.0069--------------
0.43221600.0071--------------
0.43621800.0071--------------
0.4422000.007--------------
0.44422200.0069--------------
0.44822400.0071--------------
0.45222600.0069--------------
0.45622800.0069--------------
0.4623000.0069--------------
0.46423200.0069--------------
0.46823400.0069--------------
0.47223600.0068--------------
0.47623800.0068--------------
0.4824000.0067--------------
0.48424200.0068--------------
0.48824400.0067--------------
0.49224600.0068--------------
0.49624800.0069--------------
0.525000.00680.36470.68830.94350.56240.89460.70650.38150.77090.96580.39930.56310.83710.60760.6681
0.50425200.0067--------------
0.50825400.0068--------------
0.51225600.0067--------------
0.51625800.0068--------------
0.5226000.0066--------------
0.52426200.0067--------------
0.52826400.0067--------------
0.53226600.0067--------------
0.53626800.0067--------------
0.5427000.0068--------------
0.54427200.0066--------------
0.54827400.0067--------------
0.55227600.0064--------------
0.55627800.0064--------------
0.5628000.0066--------------
0.56428200.0063--------------
0.56828400.0066--------------
0.57228600.0066--------------
0.57628800.0065--------------
0.5829000.0066--------------
0.58429200.0065--------------
0.58829400.0063--------------
0.59229600.0066--------------
0.59629800.0065--------------
0.630000.00640.35850.70810.94090.54740.89150.70370.37960.77630.95400.40380.56280.84240.60420.6672
0.60430200.0064--------------
0.60830400.0063--------------
0.61230600.0064--------------
0.61630800.0065--------------
0.6231000.0065--------------
0.62431200.0064--------------
0.62831400.0064--------------
0.63231600.0062--------------
0.63631800.0062--------------
0.6432000.0063--------------
0.64432200.0064--------------
0.64832400.0063--------------
0.65232600.0063--------------
0.65632800.0063--------------
0.6633000.0064--------------
0.66433200.0063--------------
0.66833400.0061--------------
0.67233600.0062--------------
0.67633800.0061--------------
0.6834000.0063--------------
0.68434200.006--------------
0.68834400.0061--------------
0.69234600.0062--------------
0.69634800.0062--------------
0.735000.00610.37830.70800.94410.56030.89020.70220.38240.77800.96120.39950.54140.84500.60490.6689
0.70435200.0062--------------
0.70835400.0061--------------
0.71235600.0061--------------
0.71635800.0062--------------
0.7236000.0061--------------
0.72436200.0061--------------
0.72836400.0061--------------
0.73236600.006--------------
0.73636800.006--------------
0.7437000.0061--------------
0.74437200.006--------------
0.74837400.0059--------------
0.75237600.006--------------
0.75637800.0061--------------
0.7638000.0061--------------
0.76438200.006--------------
0.76838400.0061--------------
0.77238600.0059--------------
0.77638800.006--------------
0.7839000.006--------------
0.78439200.0061--------------
0.78839400.006--------------
0.79239600.006--------------
0.79639800.0061--------------
0.840000.00590.38200.70280.94410.57220.88900.71350.38250.77900.96590.40120.54250.84460.60850.6714
0.80440200.0059--------------
0.80840400.0058--------------
0.81240600.0058--------------
0.81640800.0059--------------
0.8241000.0059--------------
0.82441200.0059--------------
0.82841400.0058--------------
0.83241600.006--------------
0.83641800.0059--------------
0.8442000.0059--------------
0.84442200.0059--------------
0.84842400.0058--------------
0.85242600.0059--------------
0.85642800.0057--------------
0.8643000.0058--------------
0.86443200.006--------------
0.86843400.0058--------------
0.87243600.0058--------------
0.87643800.0057--------------
0.8844000.0059--------------
0.88444200.0058--------------
0.88844400.0058--------------
0.89244600.0056--------------
0.89644800.0058--------------
0.945000.00590.37030.71110.94410.55550.88860.72510.39340.76320.96710.40520.53900.84420.60680.6703
0.90445200.0057--------------
0.90845400.0058--------------
0.91245600.0058--------------
0.91645800.0058--------------
0.9246000.0058--------------
0.92446200.0057--------------
0.92846400.0057--------------
0.93246600.0057--------------
0.93646800.0058--------------
0.9447000.0056--------------
0.94447200.0055--------------
0.94847400.0058--------------
0.95247600.0055--------------
0.95647800.0056--------------
0.9648000.0056--------------
0.96448200.0057--------------
0.96848400.0058--------------
0.97248600.0056--------------
0.97648800.0056--------------
0.9849000.0056--------------
0.98449200.0057--------------
0.98849400.0056--------------
0.99249600.0056--------------
0.99649800.0056--------------
1.050000.00560.37600.71310.94410.55220.88820.71570.39800.77390.97550.39870.54920.85010.59900.6718
1.00450200.0056--------------
1.00850400.0056--------------
1.01250600.0056--------------
1.01650800.0055--------------
1.0251000.0054--------------
1.02451200.0056--------------
1.02851400.0055--------------
1.03251600.0055--------------
1.03651800.0055--------------
1.0452000.0055--------------
1.04452200.0055--------------
1.04852400.0055--------------
1.05252600.0055--------------
1.05652800.0055--------------
1.0653000.0056--------------
1.06453200.0053--------------
1.06853400.0054--------------
1.07253600.0054--------------
1.07653800.0055--------------
1.0854000.0054--------------
1.08454200.0055--------------
1.08854400.0054--------------
1.09254600.0054--------------
1.09654800.0054--------------
1.155000.00540.37770.71090.93670.57050.89190.71360.39560.77500.95900.39470.53360.83680.60160.6690
1.10455200.0054--------------
1.10855400.0054--------------
1.11255600.0054--------------
1.11655800.0053--------------
1.1256000.0051--------------
1.12456200.0053--------------
1.128056400.0054--------------
1.132056600.0052--------------
1.136056800.0053--------------
1.140057000.0053--------------
1.14457200.0052--------------
1.14857400.0052--------------
1.15257600.0053--------------
1.15657800.0052--------------
1.1658000.0052--------------
1.16458200.0053--------------
1.16858400.0053--------------
1.17258600.0052--------------
1.17658800.0052--------------
1.1859000.0053--------------
1.18459200.0052--------------
1.18859400.0052--------------
1.19259600.0052--------------
1.19659800.0052--------------
1.260000.00510.39980.71710.94460.56990.88990.71940.40220.76310.96740.39600.53950.83890.60250.6731
1.20460200.0051--------------
1.20860400.0052--------------
1.21260600.0052--------------
1.21660800.0051--------------
1.2261000.0052--------------
1.22461200.0052--------------
1.22861400.0051--------------
1.23261600.0051--------------
1.23661800.0051--------------
1.2462000.0052--------------
1.24462200.0052--------------
1.24862400.0051--------------
1.25262600.0052--------------
1.25662800.0051--------------
1.2663000.0051--------------
1.26463200.0052--------------
1.26863400.0051--------------
1.27263600.0052--------------
1.27663800.005--------------
1.2864000.005--------------
1.28464200.005--------------
1.28864400.005--------------
1.29264600.0051--------------
1.29664800.0052--------------
1.365000.0050.40470.71370.94430.56900.89980.71200.39630.76890.98290.39560.55040.83630.59990.6749
1.30465200.005--------------
1.30865400.005--------------
1.31265600.0049--------------
1.31665800.005--------------
1.3266000.005--------------
1.32466200.0051--------------
1.32866400.005--------------
1.33266600.005--------------
1.33666800.005--------------
1.3467000.005--------------
1.344067200.005--------------
1.348067400.0048--------------
1.352067600.0049--------------
1.356067800.0049--------------
1.360068000.0051--------------
1.364068200.005--------------
1.368068400.0048--------------
1.372068600.005--------------
1.37668800.0049--------------
1.3869000.005--------------
1.38469200.0048--------------
1.38869400.0049--------------
1.39269600.0049--------------
1.39669800.0048--------------
1.470000.00490.40840.71560.94410.57000.89780.71340.40240.75570.97580.39970.55210.83660.59190.6741
1.40470200.0049--------------
1.40870400.0048--------------
1.41270600.0048--------------
1.41670800.0049--------------
1.4271000.0048--------------
1.42471200.0048--------------
1.42871400.0048--------------
1.43271600.0049--------------
1.43671800.0049--------------
1.4472000.0048--------------
1.44472200.0048--------------
1.44872400.0049--------------
1.45272600.0048--------------
1.45672800.0048--------------
1.4673000.0049--------------
1.46473200.0047--------------
1.46873400.0048--------------
1.47273600.0048--------------
1.47673800.0048--------------
1.4874000.0047--------------
1.48474200.0048--------------
1.48874400.0047--------------
1.49274600.0047--------------
1.49674800.0049--------------
1.575000.00470.41250.71860.94430.57230.89740.69410.39620.76760.96770.39900.54550.84330.59450.6733
1.50475200.0047--------------
1.50875400.0048--------------
1.51275600.0048--------------
1.51675800.0048--------------
1.5276000.0047--------------
1.52476200.0048--------------
1.52876400.0048--------------
1.53276600.0047--------------
1.53676800.0047--------------
1.5477000.0048--------------
1.54477200.0047--------------
1.54877400.0048--------------
1.55277600.0046--------------
1.55677800.0046--------------
1.5678000.0048--------------
1.56478200.0046------------

Truncated — view the full README on Hugging Face.

ColBERT
endpoints_compatible
feature-extraction
generated_from_trainer
model-index
modernbert
multi-vector
onnx
PyLate
safetensors
sentence-similarity
sentence-transformers
text-embeddings-inference

Similar repos

lightonai/ColBERT-Zero

Model

ColBERT-Zero

45

LiquidAI/LFM2-ColBERT-350M

Model

src="https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6…

153

thenlper/gte-large-zh

Model

gte-large-zh

121

nomic-ai/nomic-embed-text-v1-ablated

Model

nomic-embed-text-v1-ablated: A Reproducible Long Context (8192) Text Embedder

4

avsolatorio/GIST-all-MiniLM-L6-v2

Model

*GISTEmbed: Guided In-sample Selection of Training Negatives for Text Embedding Fine-tuning*

10

thenlper/gte-large

Model

gte-large

307

thenlper/gte-base-zh

Model

gte-base-zh

40

nomic-ai/nomic-embed-text-v1.5

Model

nomic-embed-text-v1.5: Resizable Production Embeddings with Matryoshka Representation Learning

943

Linked in READMEs

lightonai/pylate

Late Interaction Models Training & Retrieval

896

README badge

TusKANNy/awesome-multivector-retrieval

An extensive and commented list of resources on Late-Interaction Multivector Retrieval.

79

README badge

lightonai/GTE-ModernColBERT-v1

Model

GTE-ModernColBERT-v1

178

15 commits

2 linked in READMEs

updated Aug 17, 2026

See the code

README

GTE-ModernColBERT-v1

Multi-vector embedding model based on Alibaba-NLP/gte-modernbert-base

This is a multi-vector (ColBERT-style late interaction) embedding model trained on the ms-marco-en-bge-gemma dataset. It maps sentences & paragraphs to sequences of 128-dimensional dense vectors and can be used for semantic textual similarity using the MaxSim operator.

Model Details

Model Description

  • Model Type: Multi-vector embedding model
  • Base model: Alibaba-NLP/gte-modernbert-base
  • Document Length: 300 tokens
  • Query Length: 32 tokens
  • Output Dimensionality: 128 dimensions
  • Similarity Function: MaxSim
  • Training Dataset:
  • Language: English
  • License: Apache 2.0

Document length

GTE-ModernColBERT has been trained with knowledge distillation on MS MARCO with a document length of 300 tokens, explaining its default value for documents length.

However, as illustrated in the ModernBERT paper, ColBERT models can generalize to documents lengths way beyond their training length and GTE-ModernColBERT actually yields results way above SOTA in long-context embedding benchmarks, see LongEmbed results.

Simply change adapt the document length parameter to your needs when loading the model:

model = models.ColBERT(
    model_name_or_path=lightonai/GTE-ModernColBERT-v1,
    document_length=8192,
)

ModernBERT itself has only been trained on 8K context length, but it seems that GTE-ModernColBERT can generalize to even bigger context sizes, though it is not guaranteed so please perform your own benches!

Model Sources

Full Model Architecture

ColBERT(
  (0): Transformer({'max_seq_length': 299, 'do_lower_case': False}) with Transformer model: ModernBertModel 
  (1): Dense({'in_features': 768, 'out_features': 128, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

Usage

Sentence Transformers

This model can be used with Sentence Transformers as a multi-vector (ColBERT-style late interaction) retriever via the MultiVectorEncoder:

pip install "sentence-transformers>=6.0.0"

from sentence_transformers import MultiVectorEncoder

model = MultiVectorEncoder("lightonai/GTE-ModernColBERT-v1")

query = "Which planet is known as the Red Planet?"
documents = [
    "Venus is often called Earth's twin because of its similar size and proximity.",
    "Mars, known for its reddish appearance, is often referred to as the Red Planet.",
    "Jupiter, the largest planet in our solar system, has a prominent red spot.",
    "Saturn, famous for its rings, is sometimes mistaken for the Red Planet.",
]

query_embeddings = model.encode_query(query)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings[0].shape)
# (12, 128) (18, 128)

# MaxSim late-interaction scoring (higher is more relevant)
scores = model.similarity(query_embeddings, document_embeddings)
print(scores)
# tensor([[11.2577, 11.5113, 11.3457, 11.4518]])

PyLate

First install the PyLate library:

pip install -U pylate

Retrieval

PyLate provides a streamlined interface to index and retrieve documents using ColBERT models. The index leverages the Voyager HNSW index to efficiently handle document embeddings and enable fast retrieval.

Indexing documents

First, load the ColBERT model and initialize the Voyager index, then encode and index your documents:

from pylate import indexes, models, retrieve

# Step 1: Load the ColBERT model
model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

# Step 2: Initialize the Voyager index
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
    override=True,  # This overwrites the existing index if any
)

# Step 3: Encode the documents
documents_ids = ["1", "2", "3"]
documents = ["document 1 text", "document 2 text", "document 3 text"]

documents_embeddings = model.encode(
    documents,
    batch_size=32,
    is_query=False,  # Ensure that it is set to False to indicate that these are documents, not queries
    show_progress_bar=True,
)

# Step 4: Add document embeddings to the index by providing embeddings and corresponding ids
index.add_documents(
    documents_ids=documents_ids,
    documents_embeddings=documents_embeddings,
)

Note that you do not have to recreate the index and encode the documents every time. Once you have created an index and added the documents, you can re-use the index later by loading it:

# To load an index, simply instantiate it with the correct folder/name and without overriding it
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
)

Retrieving top-k documents for queries

Once the documents are indexed, you can retrieve the top-k most relevant documents for a given set of queries. To do so, initialize the ColBERT retriever with the index you want to search in, encode the queries and then retrieve the top-k documents to get the top matches ids and relevance scores:

# Step 1: Initialize the ColBERT retriever
retriever = retrieve.ColBERT(index=index)

# Step 2: Encode the queries
queries_embeddings = model.encode(
    ["query for document 3", "query for document 1"],
    batch_size=32,
    is_query=True,  #  # Ensure that it is set to False to indicate that these are queries
    show_progress_bar=True,
)

# Step 3: Retrieve top-k documents
scores = retriever.retrieve(
    queries_embeddings=queries_embeddings, 
    k=10,  # Retrieve the top 10 matches for each query
)

Reranking

If you only want to use the ColBERT model to perform reranking on top of your first-stage retrieval pipeline without building an index, you can simply use rank function and pass the queries and documents to rerank:

from pylate import rank, models

queries = [
    "query A",
    "query B",
]

documents = [
    ["document A", "document B"],
    ["document 1", "document C", "document B"],
]

documents_ids = [
    [1, 2],
    [1, 3, 2],
]

model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

queries_embeddings = model.encode(
    queries,
    is_query=True,
)

documents_embeddings = model.encode(
    documents,
    is_query=False,
)

reranked_documents = rank.rerank(
    documents_ids=documents_ids,
    queries_embeddings=queries_embeddings,
    documents_embeddings=documents_embeddings,
)

Evaluation

Metrics

BEIR Benchmark

GTE-ModernColBERT is the first model to outpeform ColBERT-small on the BEIR benchmark. As reproduction in the IR domain is challenging, we worked closely with Benjamin Clavié, the author of ColBERT-small to reproduce the evaluation setup of this model. Despite all these efforts and reducing to the maximum the difference in scores in most of the datasets, some are still a bit different. For this reason, we also report the results of ColBERT-small in the same setup we used to evaluate GTE-ModernColBERT for completness and fair comparison.

ModelAverageFiQA2018NFCorpusTREC-COVIDTouche2020ArguAnaQuoraRetrievalSCIDOCSSciFactNQClimateFEVERHotpotQADBPediaCQADupstackFEVERMSMARCO
GTE-ModernColBERT54.6745.2837.9383.5931.2348.5186.6119.0676.3461.830.6277.3248.034187.4445.32
ColBERT-small (reported)53.7941.1537.384.5925.6950.0987.7218.4274.7759.133.0776.1145.5838.7590.9643.5
JinaColBERT-v240.834.683.427.436.688.718.667.86423.976.647.180.5
ColBERT-small (rerunned)53.3541.0136.8683.1424.9546.7687.8918.7274.0259.4232.8376.8846.3639.3688.6643.44

LongEmbed Benchmark

GTE-ModernColBERT has been trained with knowledge distillation on MS MARCO with a document length of 300 tokens, explaining its default value for documents length. However, as illustrated in the ModernBERT paper, ColBERT models can generalize to documents lengths way beyond their training length and GTE-ModernColBERT actually yields results way above SOTA (almost 10 points above previous SOTA) in long-context embedding benchmark:

ModelMeanLEMBNarrativeQARetrievalLEMBNeedleRetrievalLEMBPasskeyRetrievalLEMBQMSumRetrievalLEMBSummScreenFDRetrievalLEMBWikimQARetrieval
GTE-ModernColBERT (with 32k document length)88.3978.8292.59272.1794.9899.87
voyage-multilingual-279.1764.69475.259751.49599.10587.489
inf-retriever-v173.1960.70261.578.7555.07297.38785.751
snowflake-arctic-embed-l-v2,063.7343.63250.2577.2540.0496.38374.843
gte-multilingual-base62.1252.35842.2555.543.03395.49984.078
jasper_en_vision_language_v160.9337.9285562.2541.18697.20672.025
bge-m358.7345.76140.255935.54394.08977.726
jina-embeddings-v355.6634.297643839.33792.33466.018
e5-base-4k54.5130.0337.7565.2531.26893.86868.875
gte-Qwen2-7B-instruct47.2445.463138.531.27276.0861.151

ModernBERT itself has only been trained on 8K context length, but it seems that GTE-ModernColBERT can generalize to even bigger context sizes, though it is not guaranteed so please perform your own benches!

PyLate Information Retrieval

  • Datasets: NanoClimateFEVER, NanoDBPedia, NanoFEVER, NanoFiQA2018, NanoHotpotQA, NanoMSMARCO, NanoNFCorpus, NanoNQ, NanoQuoraRetrieval, NanoSCIDOCS, NanoArguAna, NanoSciFact and NanoTouche2020
  • Evaluated with pylate.evaluation.pylate_information_retrieval_evaluator.PyLateInformationRetrievalEvaluator
MetricNanoClimateFEVERNanoDBPediaNanoFEVERNanoFiQA2018NanoHotpotQANanoMSMARCONanoNFCorpusNanoNQNanoQuoraRetrievalNanoSCIDOCSNanoArguAnaNanoSciFactNanoTouche2020
MaxSim_accuracy@10.360.880.920.560.920.540.560.640.960.480.30.740.7755
MaxSim_accuracy@30.620.940.980.661.00.680.680.821.00.740.620.860.9388
MaxSim_accuracy@50.780.960.980.741.00.740.740.861.00.780.70.90.9796
MaxSim_accuracy@100.860.981.00.81.00.920.760.91.00.840.820.940.9796
MaxSim_precision@10.360.880.920.560.920.540.560.640.960.480.30.740.7755
MaxSim_precision@30.23330.71330.360.32670.580.22670.43330.28670.40.40.20670.30.6599
MaxSim_precision@50.2080.6560.2160.2560.360.1480.3920.180.2560.2920.140.1960.6571
MaxSim_precision@100.1280.5720.110.1520.1860.0920.3040.10.1340.1940.0820.1040.5184
MaxSim_recall@10.18330.1180.85670.30920.460.540.06640.610.84730.10070.30.7150.0518
MaxSim_recall@30.2890.23070.960.47840.870.680.1020.780.94530.24670.620.830.1362
MaxSim_recall@50.41570.29620.960.57520.90.740.12840.820.96930.29970.70.8850.2193
MaxSim_recall@100.49570.41460.980.64120.930.920.15660.880.98930.39670.820.930.334
MaxSim_ndcg@100.41480.72960.94520.5670.90120.70890.39570.76450.96910.39870.56090.83720.5927
MaxSim_mrr@100.52660.91690.95220.63590.960.64470.6270.7390.97670.61370.47750.81170.8629
MaxSim_map@1000.33470.58840.92710.50320.85920.64960.19180.72390.95520.31630.48240.80490.4257

Nano BEIR

  • Dataset: NanoBEIR_mean
  • Evaluated with pylate.evaluation.nano_beir_evaluator.NanoBEIREvaluator
MetricValue
MaxSim_accuracy@10.6643
MaxSim_accuracy@30.8107
MaxSim_accuracy@50.8584
MaxSim_accuracy@100.9077
MaxSim_precision@10.6643
MaxSim_precision@30.3943
MaxSim_precision@50.3044
MaxSim_precision@100.2059
MaxSim_recall@10.3968
MaxSim_recall@30.5514
MaxSim_recall@50.6084
MaxSim_recall@100.6837
MaxSim_ndcg@100.6758
MaxSim_mrr@100.7496
MaxSim_map@1000.5971

Training Details

Training Hyperparameters

Non-Default Hyperparameters

  • eval_strategy: steps
  • per_device_train_batch_size: 16
  • learning_rate: 3e-05
  • bf16: True

All Hyperparameters

Click to expand
  • overwrite_output_dir: False
  • do_predict: False
  • eval_strategy: steps
  • prediction_loss_only: True
  • per_device_train_batch_size: 16
  • per_device_eval_batch_size: 8
  • per_gpu_train_batch_size: None
  • per_gpu_eval_batch_size: None
  • gradient_accumulation_steps: 1
  • eval_accumulation_steps: None
  • torch_empty_cache_steps: None
  • learning_rate: 3e-05
  • weight_decay: 0.0
  • adam_beta1: 0.9
  • adam_beta2: 0.999
  • adam_epsilon: 1e-08
  • max_grad_norm: 1.0
  • num_train_epochs: 3
  • max_steps: -1
  • lr_scheduler_type: linear
  • lr_scheduler_kwargs: {}
  • warmup_ratio: 0.0
  • warmup_steps: 0
  • log_level: passive
  • log_level_replica: warning
  • log_on_each_node: True
  • logging_nan_inf_filter: True
  • save_safetensors: True
  • save_on_each_node: False
  • save_only_model: False
  • restore_callback_states_from_checkpoint: False
  • no_cuda: False
  • use_cpu: False
  • use_mps_device: False
  • seed: 42
  • data_seed: None
  • jit_mode_eval: False
  • use_ipex: False
  • bf16: True
  • fp16: False
  • fp16_opt_level: O1
  • half_precision_backend: auto
  • bf16_full_eval: False
  • fp16_full_eval: False
  • tf32: None
  • local_rank: 6
  • ddp_backend: None
  • tpu_num_cores: None
  • tpu_metrics_debug: False
  • debug: []
  • dataloader_drop_last: True
  • dataloader_num_workers: 0
  • dataloader_prefetch_factor: None
  • past_index: -1
  • disable_tqdm: False
  • remove_unused_columns: True
  • label_names: None
  • load_best_model_at_end: False
  • ignore_data_skip: False
  • fsdp: []
  • fsdp_min_num_params: 0
  • fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
  • fsdp_transformer_layer_cls_to_wrap: None
  • accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
  • deepspeed: None
  • label_smoothing_factor: 0.0
  • optim: adamw_torch
  • optim_args: None
  • adafactor: False
  • group_by_length: False
  • length_column_name: length
  • ddp_find_unused_parameters: None
  • ddp_bucket_cap_mb: None
  • ddp_broadcast_buffers: False
  • dataloader_pin_memory: True
  • dataloader_persistent_workers: False
  • skip_memory_metrics: True
  • use_legacy_prediction_loop: False
  • push_to_hub: False
  • resume_from_checkpoint: None
  • hub_model_id: None
  • hub_strategy: every_save
  • hub_private_repo: None
  • hub_always_push: False
  • gradient_checkpointing: False
  • gradient_checkpointing_kwargs: None
  • include_inputs_for_metrics: False
  • include_for_metrics: []
  • eval_do_concat_batches: True
  • fp16_backend: auto
  • push_to_hub_model_id: None
  • push_to_hub_organization: None
  • mp_parameters:
  • auto_find_batch_size: False
  • full_determinism: False
  • torchdynamo: None
  • ray_scope: last
  • ddp_timeout: 1800
  • torch_compile: False
  • torch_compile_backend: None
  • torch_compile_mode: None
  • dispatch_batches: None
  • split_batches: None
  • include_tokens_per_second: False
  • include_num_input_tokens_seen: False
  • neftune_noise_alpha: None
  • optim_target_modules: None
  • batch_eval_metrics: False
  • eval_on_start: False
  • use_liger_kernel: False
  • eval_use_gather_object: False
  • average_tokens_across_devices: False
  • prompts: None
  • batch_sampler: batch_sampler
  • multi_dataset_batch_sampler: proportional

Training Logs

Click to expand
EpochStepTraining LossNanoClimateFEVER_MaxSim_ndcg@10NanoDBPedia_MaxSim_ndcg@10NanoFEVER_MaxSim_ndcg@10NanoFiQA2018_MaxSim_ndcg@10NanoHotpotQA_MaxSim_ndcg@10NanoMSMARCO_MaxSim_ndcg@10NanoNFCorpus_MaxSim_ndcg@10NanoNQ_MaxSim_ndcg@10NanoQuoraRetrieval_MaxSim_ndcg@10NanoSCIDOCS_MaxSim_ndcg@10NanoArguAna_MaxSim_ndcg@10NanoSciFact_MaxSim_ndcg@10NanoTouche2020_MaxSim_ndcg@10NanoBEIR_mean_MaxSim_ndcg@10
0.004200.0493--------------
0.008400.0434--------------
0.012600.0324--------------
0.016800.0238--------------
0.021000.0202--------------
0.0241200.0186--------------
0.0281400.0172--------------
0.0321600.0164--------------
0.0361800.0157--------------
0.042000.0153--------------
0.0442200.0145--------------
0.0482400.014--------------
0.0522600.0138--------------
0.0562800.0135--------------
0.063000.0132--------------
0.0643200.0129--------------
0.0683400.0126--------------
0.0723600.0123--------------
0.0763800.0122--------------
0.084000.012--------------
0.0844200.0121--------------
0.0884400.0115--------------
0.0924600.0113--------------
0.0964800.0112--------------
0.15000.01110.30850.63090.92060.53030.86180.68930.37030.71630.95480.38850.46820.79300.59820.6331
0.1045200.0109--------------
0.1085400.0109--------------
0.1125600.0109--------------
0.1165800.0105--------------
0.126000.0102--------------
0.1246200.0104--------------
0.1286400.0103--------------
0.1326600.01--------------
0.1366800.0101--------------
0.147000.0098--------------
0.1447200.0097--------------
0.1487400.0097--------------
0.1527600.0096--------------
0.1567800.0096--------------
0.168000.0094--------------
0.1648200.0096--------------
0.1688400.0095--------------
0.1728600.0093--------------
0.1768800.0092--------------
0.189000.0093--------------
0.1849200.009--------------
0.1889400.009--------------
0.1929600.0089--------------
0.1969800.0089--------------
0.210000.00890.31480.65860.93350.53740.88100.68050.37460.73680.94860.39550.48240.82190.60890.6442
0.20410200.0088--------------
0.20810400.0089--------------
0.21210600.0088--------------
0.21610800.0086--------------
0.2211000.0087--------------
0.22411200.0088--------------
0.22811400.0086--------------
0.23211600.0086--------------
0.23611800.0084--------------
0.2412000.0086--------------
0.24412200.0085--------------
0.24812400.0084--------------
0.25212600.0084--------------
0.25612800.0081--------------
0.2613000.0083--------------
0.26413200.0084--------------
0.26813400.0082--------------
0.27213600.0082--------------
0.27613800.008--------------
0.2814000.0078--------------
0.28414200.0079--------------
0.28814400.0078--------------
0.29214600.0081--------------
0.29614800.0081--------------
0.315000.00790.35100.65900.92850.54630.88930.68530.38000.73700.95130.39800.52680.82680.61300.6533
0.30415200.0078--------------
0.30815400.0078--------------
0.31215600.0077--------------
0.31615800.0078--------------
0.3216000.0078--------------
0.32416200.0078--------------
0.32816400.0078--------------
0.33216600.0076--------------
0.33616800.0076--------------
0.3417000.0077--------------
0.34417200.0076--------------
0.34817400.0074--------------
0.35217600.0074--------------
0.35617800.0075--------------
0.3618000.0076--------------
0.36418200.0075--------------
0.36818400.0073--------------
0.37218600.0075--------------
0.37618800.0073--------------
0.3819000.0074--------------
0.38419200.0072--------------
0.38819400.0072--------------
0.39219600.0071--------------
0.39619800.0073--------------
0.420000.00710.35510.68070.93110.53400.89510.70190.37670.74600.95590.39120.51210.82450.60580.6546
0.40420200.0073--------------
0.40820400.0072--------------
0.41220600.0071--------------
0.41620800.0073--------------
0.4221000.0069--------------
0.42421200.0071--------------
0.42821400.0069--------------
0.43221600.0071--------------
0.43621800.0071--------------
0.4422000.007--------------
0.44422200.0069--------------
0.44822400.0071--------------
0.45222600.0069--------------
0.45622800.0069--------------
0.4623000.0069--------------
0.46423200.0069--------------
0.46823400.0069--------------
0.47223600.0068--------------
0.47623800.0068--------------
0.4824000.0067--------------
0.48424200.0068--------------
0.48824400.0067--------------
0.49224600.0068--------------
0.49624800.0069--------------
0.525000.00680.36470.68830.94350.56240.89460.70650.38150.77090.96580.39930.56310.83710.60760.6681
0.50425200.0067--------------
0.50825400.0068--------------
0.51225600.0067--------------
0.51625800.0068--------------
0.5226000.0066--------------
0.52426200.0067--------------
0.52826400.0067--------------
0.53226600.0067--------------
0.53626800.0067--------------
0.5427000.0068--------------
0.54427200.0066--------------
0.54827400.0067--------------
0.55227600.0064--------------
0.55627800.0064--------------
0.5628000.0066--------------
0.56428200.0063--------------
0.56828400.0066--------------
0.57228600.0066--------------
0.57628800.0065--------------
0.5829000.0066--------------
0.58429200.0065--------------
0.58829400.0063--------------
0.59229600.0066--------------
0.59629800.0065--------------
0.630000.00640.35850.70810.94090.54740.89150.70370.37960.77630.95400.40380.56280.84240.60420.6672
0.60430200.0064--------------
0.60830400.0063--------------
0.61230600.0064--------------
0.61630800.0065--------------
0.6231000.0065--------------
0.62431200.0064--------------
0.62831400.0064--------------
0.63231600.0062--------------
0.63631800.0062--------------
0.6432000.0063--------------
0.64432200.0064--------------
0.64832400.0063--------------
0.65232600.0063--------------
0.65632800.0063--------------
0.6633000.0064--------------
0.66433200.0063--------------
0.66833400.0061--------------
0.67233600.0062--------------
0.67633800.0061--------------
0.6834000.0063--------------
0.68434200.006--------------
0.68834400.0061--------------
0.69234600.0062--------------
0.69634800.0062--------------
0.735000.00610.37830.70800.94410.56030.89020.70220.38240.77800.96120.39950.54140.84500.60490.6689
0.70435200.0062--------------
0.70835400.0061--------------
0.71235600.0061--------------
0.71635800.0062--------------
0.7236000.0061--------------
0.72436200.0061--------------
0.72836400.0061--------------
0.73236600.006--------------
0.73636800.006--------------
0.7437000.0061--------------
0.74437200.006--------------
0.74837400.0059--------------
0.75237600.006--------------
0.75637800.0061--------------
0.7638000.0061--------------
0.76438200.006--------------
0.76838400.0061--------------
0.77238600.0059--------------
0.77638800.006--------------
0.7839000.006--------------
0.78439200.0061--------------
0.78839400.006--------------
0.79239600.006--------------
0.79639800.0061--------------
0.840000.00590.38200.70280.94410.57220.88900.71350.38250.77900.96590.40120.54250.84460.60850.6714
0.80440200.0059--------------
0.80840400.0058--------------
0.81240600.0058--------------
0.81640800.0059--------------
0.8241000.0059--------------
0.82441200.0059--------------
0.82841400.0058--------------
0.83241600.006--------------
0.83641800.0059--------------
0.8442000.0059--------------
0.84442200.0059--------------
0.84842400.0058--------------
0.85242600.0059--------------
0.85642800.0057--------------
0.8643000.0058--------------
0.86443200.006--------------
0.86843400.0058--------------
0.87243600.0058--------------
0.87643800.0057--------------
0.8844000.0059--------------
0.88444200.0058--------------
0.88844400.0058--------------
0.89244600.0056--------------
0.89644800.0058--------------
0.945000.00590.37030.71110.94410.55550.88860.72510.39340.76320.96710.40520.53900.84420.60680.6703
0.90445200.0057--------------
0.90845400.0058--------------
0.91245600.0058--------------
0.91645800.0058--------------
0.9246000.0058--------------
0.92446200.0057--------------
0.92846400.0057--------------
0.93246600.0057--------------
0.93646800.0058--------------
0.9447000.0056--------------
0.94447200.0055--------------
0.94847400.0058--------------
0.95247600.0055--------------
0.95647800.0056--------------
0.9648000.0056--------------
0.96448200.0057--------------
0.96848400.0058--------------
0.97248600.0056--------------
0.97648800.0056--------------
0.9849000.0056--------------
0.98449200.0057--------------
0.98849400.0056--------------
0.99249600.0056--------------
0.99649800.0056--------------
1.050000.00560.37600.71310.94410.55220.88820.71570.39800.77390.97550.39870.54920.85010.59900.6718
1.00450200.0056--------------
1.00850400.0056--------------
1.01250600.0056--------------
1.01650800.0055--------------
1.0251000.0054--------------
1.02451200.0056--------------
1.02851400.0055--------------
1.03251600.0055--------------
1.03651800.0055--------------
1.0452000.0055--------------
1.04452200.0055--------------
1.04852400.0055--------------
1.05252600.0055--------------
1.05652800.0055--------------
1.0653000.0056--------------
1.06453200.0053--------------
1.06853400.0054--------------
1.07253600.0054--------------
1.07653800.0055--------------
1.0854000.0054--------------
1.08454200.0055--------------
1.08854400.0054--------------
1.09254600.0054--------------
1.09654800.0054--------------
1.155000.00540.37770.71090.93670.57050.89190.71360.39560.77500.95900.39470.53360.83680.60160.6690
1.10455200.0054--------------
1.10855400.0054--------------
1.11255600.0054--------------
1.11655800.0053--------------
1.1256000.0051--------------
1.12456200.0053--------------
1.128056400.0054--------------
1.132056600.0052--------------
1.136056800.0053--------------
1.140057000.0053--------------
1.14457200.0052--------------
1.14857400.0052--------------
1.15257600.0053--------------
1.15657800.0052--------------
1.1658000.0052--------------
1.16458200.0053--------------
1.16858400.0053--------------
1.17258600.0052--------------
1.17658800.0052--------------
1.1859000.0053--------------
1.18459200.0052--------------
1.18859400.0052--------------
1.19259600.0052--------------
1.19659800.0052--------------
1.260000.00510.39980.71710.94460.56990.88990.71940.40220.76310.96740.39600.53950.83890.60250.6731
1.20460200.0051--------------
1.20860400.0052--------------
1.21260600.0052--------------
1.21660800.0051--------------
1.2261000.0052--------------
1.22461200.0052--------------
1.22861400.0051--------------
1.23261600.0051--------------
1.23661800.0051--------------
1.2462000.0052--------------
1.24462200.0052--------------
1.24862400.0051--------------
1.25262600.0052--------------
1.25662800.0051--------------
1.2663000.0051--------------
1.26463200.0052--------------
1.26863400.0051--------------
1.27263600.0052--------------
1.27663800.005--------------
1.2864000.005--------------
1.28464200.005--------------
1.28864400.005--------------
1.29264600.0051--------------
1.29664800.0052--------------
1.365000.0050.40470.71370.94430.56900.89980.71200.39630.76890.98290.39560.55040.83630.59990.6749
1.30465200.005--------------
1.30865400.005--------------
1.31265600.0049--------------
1.31665800.005--------------
1.3266000.005--------------
1.32466200.0051--------------
1.32866400.005--------------
1.33266600.005--------------
1.33666800.005--------------
1.3467000.005--------------
1.344067200.005--------------
1.348067400.0048--------------
1.352067600.0049--------------
1.356067800.0049--------------
1.360068000.0051--------------
1.364068200.005--------------
1.368068400.0048--------------
1.372068600.005--------------
1.37668800.0049--------------
1.3869000.005--------------
1.38469200.0048--------------
1.38869400.0049--------------
1.39269600.0049--------------
1.39669800.0048--------------
1.470000.00490.40840.71560.94410.57000.89780.71340.40240.75570.97580.39970.55210.83660.59190.6741
1.40470200.0049--------------
1.40870400.0048--------------
1.41270600.0048--------------
1.41670800.0049--------------
1.4271000.0048--------------
1.42471200.0048--------------
1.42871400.0048--------------
1.43271600.0049--------------
1.43671800.0049--------------
1.4472000.0048--------------
1.44472200.0048--------------
1.44872400.0049--------------
1.45272600.0048--------------
1.45672800.0048--------------
1.4673000.0049--------------
1.46473200.0047--------------
1.46873400.0048--------------
1.47273600.0048--------------
1.47673800.0048--------------
1.4874000.0047--------------
1.48474200.0048--------------
1.48874400.0047--------------
1.49274600.0047--------------
1.49674800.0049--------------
1.575000.00470.41250.71860.94430.57230.89740.69410.39620.76760.96770.39900.54550.84330.59450.6733
1.50475200.0047--------------
1.50875400.0048--------------
1.51275600.0048--------------
1.51675800.0048--------------
1.5276000.0047--------------
1.52476200.0048--------------
1.52876400.0048--------------
1.53276600.0047--------------
1.53676800.0047--------------
1.5477000.0048--------------
1.54477200.0047--------------
1.54877400.0048--------------
1.55277600.0046--------------
1.55677800.0046--------------
1.5678000.0048--------------
1.56478200.0046------------

Truncated — view the full README on Hugging Face.

ColBERT
endpoints_compatible
feature-extraction
generated_from_trainer
model-index
modernbert
multi-vector
onnx
PyLate
safetensors
sentence-similarity
sentence-transformers
text-embeddings-inference

Similar repos

lightonai/ColBERT-Zero

Model

ColBERT-Zero

45

LiquidAI/LFM2-ColBERT-350M

Model

src="https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/2b08LKpev0DNEk6…

153

thenlper/gte-large-zh

Model

gte-large-zh

121

nomic-ai/nomic-embed-text-v1-ablated

Model

nomic-embed-text-v1-ablated: A Reproducible Long Context (8192) Text Embedder

4

avsolatorio/GIST-all-MiniLM-L6-v2

Model

*GISTEmbed: Guided In-sample Selection of Training Negatives for Text Embedding Fine-tuning*

10

thenlper/gte-large

Model

gte-large

307

thenlper/gte-base-zh

Model

gte-base-zh

40

nomic-ai/nomic-embed-text-v1.5

Model

nomic-embed-text-v1.5: Resizable Production Embeddings with Matryoshka Representation Learning

943

Linked in READMEs

lightonai/pylate

Late Interaction Models Training & Retrieval

896

README badge

TusKANNy/awesome-multivector-retrieval

An extensive and commented list of resources on Late-Interaction Multivector Retrieval.

79

README badge