You can deploy this model and start earning money today!

tomaarsen/
csr-mxbai-embed-large-v1-nq-cos-sim-scale-5-gamma-1-detach-2

Name: csr-mxbai-embed-large-v1-nq-cos-sim-scale-5-gamma-1-detach-2
Availability: InStock

Views

No views yet

Sparse CSR model trained on Natural Questions

This is a CSR Sparse Encoder model finetuned from mixedbread-ai/mxbai-embed-large-v1 on the natural-questions dataset using the sentence-transformers library. It maps sentences & paragraphs to a 4096-dimensional sparse vector space with 256 maximum active dimensions and can be used for semantic search and sparse retrieval.

Model Details

Model Description

Model Type: CSR Sparse Encoder
Base model: mixedbread-ai/mxbai-embed-large-v1
Maximum Sequence Length: 512 tokens
Output Dimensionality: 4096 dimensions (trained with 256 maximum active dimensions)
Similarity Function: Cosine Similarity
Training Dataset:
- natural-questions
Language: en
License: apache-2.0

Model Sources

Documentation: Sentence Transformers Documentation
Documentation: Sparse Encoder Documentation
Repository: Sentence Transformers on GitHub
Hugging Face: Sparse Encoders on Hugging Face

Full Model Architecture

SparseEncoder(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): CSRSparsity({'input_dim': 1024, 'hidden_dim': 4096, 'k': 256, 'k_aux': 512, 'normalize': False, 'dead_threshold': 30})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

python

from sentence_transformers import SparseEncoder

# Download from the 🤗 Hub
model = SparseEncoder("tomaarsen/csr-mxbai-embed-large-v1-nq-cos-sim-scale-5-gamma-1-detach-2")
# Run inference
queries = [
    "who is cornelius in the book of acts",
]
documents = [
    'Cornelius the Centurion Cornelius (Greek: Κορνήλιος) was a Roman centurion who is considered by Christians to be one of the first Gentiles to convert to the faith, as related in Acts of the Apostles.',
    "Joe Ranft Ranft reunited with Lasseter when he was hired by Pixar in 1991 as their head of story.[1] There he worked on all of their films produced up to 2006; this included Toy Story (for which he received an Academy Award nomination) and A Bug's Life, as the co-story writer and others as story supervisor. His final film was Cars. He also voiced characters in many of the films, including Heimlich the caterpillar in A Bug's Life, Wheezy the penguin in Toy Story 2, and Jacques the shrimp in Finding Nemo.[1]",
    'Wonderful Tonight "Wonderful Tonight" is a ballad written by Eric Clapton. It was included on Clapton\'s 1977 album Slowhand. Clapton wrote the song about Pattie Boyd.[1] The female vocal harmonies on the song are provided by Marcella Detroit (then Marcy Levy) and Yvonne Elliman.',
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 4096] [3, 4096]

# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[0.8907, 0.0410, 0.0237]])

Evaluation

Metrics

Sparse Information Retrieval

Dataset: nq_eval_4
Evaluated with SparseInformationRetrievalEvaluator with these parameters:
json
1{ 2 "max_active_dims": 4 3}

Metric	Value
cosine_accuracy@1	0.341
cosine_accuracy@3	0.53
cosine_accuracy@5	0.616
cosine_accuracy@10	0.71
cosine_precision@1	0.341
cosine_precision@3	0.1767
cosine_precision@5	0.1232
cosine_precision@10	0.071
cosine_recall@1	0.341
cosine_recall@3	0.53
cosine_recall@5	0.616
cosine_recall@10	0.71
cosine_ndcg@10	0.5178
cosine_mrr@10	0.457
cosine_map@100	0.4681
query_active_dims	4.0
query_sparsity_ratio	0.999
corpus_active_dims	4.0
corpus_sparsity_ratio	0.999

Sparse Information Retrieval

Dataset: nq_eval_8
Evaluated with SparseInformationRetrievalEvaluator with these parameters:
json
1{ 2 "max_active_dims": 8 3}

Metric	Value
cosine_accuracy@1	0.479
cosine_accuracy@3	0.683
cosine_accuracy@5	0.743
cosine_accuracy@10	0.827
cosine_precision@1	0.479
cosine_precision@3	0.2277
cosine_precision@5	0.1486
cosine_precision@10	0.0827
cosine_recall@1	0.479
cosine_recall@3	0.683
cosine_recall@5	0.743
cosine_recall@10	0.827
cosine_ndcg@10	0.6515
cosine_mrr@10	0.5954
cosine_map@100	0.6025
query_active_dims	8.0
query_sparsity_ratio	0.998
corpus_active_dims	8.0
corpus_sparsity_ratio	0.998

Sparse Information Retrieval

Dataset: nq_eval_16
Evaluated with SparseInformationRetrievalEvaluator with these parameters:
json
1{ 2 "max_active_dims": 16 3}

Metric	Value
cosine_accuracy@1	0.61
cosine_accuracy@3	0.792
cosine_accuracy@5	0.843
cosine_accuracy@10	0.9
cosine_precision@1	0.61
cosine_precision@3	0.264
cosine_precision@5	0.1686
cosine_precision@10	0.09
cosine_recall@1	0.61
cosine_recall@3	0.792
cosine_recall@5	0.843
cosine_recall@10	0.9
cosine_ndcg@10	0.7573
cosine_mrr@10	0.7115
cosine_map@100	0.716
query_active_dims	16.0
query_sparsity_ratio	0.9961
corpus_active_dims	16.0
corpus_sparsity_ratio	0.9961

Sparse Information Retrieval

Dataset: nq_eval_32
Evaluated with SparseInformationRetrievalEvaluator with these parameters:
json
1{ 2 "max_active_dims": 32 3}

Metric	Value
cosine_accuracy@1	0.739
cosine_accuracy@3	0.871
cosine_accuracy@5	0.899
cosine_accuracy@10	0.936
cosine_precision@1	0.739
cosine_precision@3	0.2903
cosine_precision@5	0.1798
cosine_precision@10	0.0936
cosine_recall@1	0.739
cosine_recall@3	0.871
cosine_recall@5	0.899
cosine_recall@10	0.936
cosine_ndcg@10	0.8407
cosine_mrr@10	0.8098
cosine_map@100	0.8124
query_active_dims	32.0
query_sparsity_ratio	0.9922
corpus_active_dims	32.0
corpus_sparsity_ratio	0.9922

Sparse Information Retrieval

Dataset: nq_eval_64
Evaluated with SparseInformationRetrievalEvaluator with these parameters:
json
1{ 2 "max_active_dims": 64 3}

Metric	Value
cosine_accuracy@1	0.775
cosine_accuracy@3	0.895
cosine_accuracy@5	0.925
cosine_accuracy@10	0.951
cosine_precision@1	0.775
cosine_precision@3	0.2983
cosine_precision@5	0.185
cosine_precision@10	0.0951
cosine_recall@1	0.775
cosine_recall@3	0.895
cosine_recall@5	0.925
cosine_recall@10	0.951
cosine_ndcg@10	0.8673
cosine_mrr@10	0.8399
cosine_map@100	0.8418
query_active_dims	63.992
query_sparsity_ratio	0.9844
corpus_active_dims	64.0
corpus_sparsity_ratio	0.9844

Sparse Information Retrieval

Dataset: nq_eval_128
Evaluated with SparseInformationRetrievalEvaluator with these parameters:
json
1{ 2 "max_active_dims": 128 3}

Metric	Value
cosine_accuracy@1	0.797
cosine_accuracy@3	0.901
cosine_accuracy@5	0.933
cosine_accuracy@10	0.951
cosine_precision@1	0.797
cosine_precision@3	0.3003
cosine_precision@5	0.1866
cosine_precision@10	0.0951
cosine_recall@1	0.797
cosine_recall@3	0.901
cosine_recall@5	0.933
cosine_recall@10	0.951
cosine_ndcg@10	0.8781
cosine_mrr@10	0.8542
cosine_map@100	0.8561
query_active_dims	119.217
query_sparsity_ratio	0.9709
corpus_active_dims	119.652
corpus_sparsity_ratio	0.9708

Sparse Information Retrieval

Dataset: nq_eval_256
Evaluated with SparseInformationRetrievalEvaluator with these parameters:
json
1{ 2 "max_active_dims": 256 3}

Metric	Value
cosine_accuracy@1	0.8
cosine_accuracy@3	0.901
cosine_accuracy@5	0.933
cosine_accuracy@10	0.951
cosine_precision@1	0.8
cosine_precision@3	0.3003
cosine_precision@5	0.1866
cosine_precision@10	0.0951
cosine_recall@1	0.8
cosine_recall@3	0.901
cosine_recall@5	0.933
cosine_recall@10	0.951
cosine_ndcg@10	0.8789
cosine_mrr@10	0.8553
cosine_map@100	0.8573
query_active_dims	133.43
query_sparsity_ratio	0.9674
corpus_active_dims	129.169
corpus_sparsity_ratio	0.9685

Training Details

Training Dataset

natural-questions

Dataset: natural-questions at f9e894e
Size: 99,000 training samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 10 tokens
mean: 11.71 tokens
max: 26 tokens
min: 4 tokens
mean: 131.81 tokens
max: 450 tokens

Samples:

query	answer
`who played the father in papa don't preach`	`Alex McArthur Alex McArthur (born March 6, 1957) is an American actor.`
`where was the location of the battle of hastings`	`Battle of Hastings The Battle of Hastings[a] was fought on 14 October 1066 between the Norman-French army of William, the Duke of Normandy, and an English army under the Anglo-Saxon King Harold Godwinson, beginning the Norman conquest of England. It took place approximately 7 miles (11 kilometres) northwest of Hastings, close to the present-day town of Battle, East Sussex, and was a decisive Norman victory.`
`how many puppies can a dog give birth to`	`Canine reproduction The largest litter size to date was set by a Neapolitan Mastiff in Manea, Cambridgeshire, UK on November 29, 2004; the litter was 24 puppies.[22]`

Loss: CSRLoss with these parameters:
json
1{ 2 "beta": 0.1, 3 "gamma": 1.0, 4 "loss": "SparseMultipleNegativesRankingLoss(scale=5.0, similarity_fct='cos_sim')" 5}

Evaluation Dataset

natural-questions

Dataset: natural-questions at f9e894e
Size: 1,000 evaluation samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 10 tokens
mean: 11.69 tokens
max: 23 tokens
min: 15 tokens
mean: 134.01 tokens
max: 512 tokens

Samples:

query	answer
`where is the tiber river located in italy`	Tiber The Tiber (/ˈtaɪbər/, Latin: Tiberis,[1] Italian: Tevere [ˈteːvere])[2] is the third-longest river in Italy, rising in the Apennine Mountains in Emilia-Romagna and flowing 406 kilometres (252 mi) through Tuscany, Umbria and Lazio, where it is joined by the river Aniene, to the Tyrrhenian Sea, between Ostia and Fiumicino.[3] It drains a basin estimated at 17,375 square kilometres (6,709 sq mi). The river has achieved lasting fame as the main watercourse of the city of Rome, founded on its eastern banks.
`what kind of car does jay gatsby drive`	`Jay Gatsby At the Buchanan home, Jordan Baker, Nick, Jay, and the Buchanans decide to visit New York City. Tom borrows Gatsby's yellow Rolls Royce to drive up to the city. On the way to New York City, Tom makes a detour at a gas station in "the Valley of Ashes", a run-down part of Long Island. The owner, George Wilson, shares his concern that his wife, Myrtle, may be having an affair. This unnerves Tom, who has been having an affair with Myrtle, and he leaves in a hurry.`
`who sings if i can dream about you`	`I Can Dream About You "I Can Dream About You" is a song performed by American singer Dan Hartman on the soundtrack album of the film Streets of Fire. Released in 1984 as a single from the soundtrack, and included on Hartman's album I Can Dream About You, it reached number 6 on the Billboard Hot 100.[1]`

Loss: CSRLoss with these parameters:
json
1{ 2 "beta": 0.1, 3 "gamma": 1.0, 4 "loss": "SparseMultipleNegativesRankingLoss(scale=5.0, similarity_fct='cos_sim')" 5}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 64
per_device_eval_batch_size: 64
learning_rate: 4e-05
num_train_epochs: 1
bf16: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 64
per_device_eval_batch_size: 64
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: 4e-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: 1
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: 0
ddp_backend: None
tpu_num_cores: None
tpu_metrics_debug: False
debug: []
dataloader_drop_last: False
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
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: no_duplicates
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss	nq_eval_4_cosine_ndcg@10	nq_eval_8_cosine_ndcg@10	nq_eval_16_cosine_ndcg@10	nq_eval_32_cosine_ndcg@10	nq_eval_64_cosine_ndcg@10	nq_eval_128_cosine_ndcg@10	nq_eval_256_cosine_ndcg@10
-1	-1	-	-	0.2566	0.4513	0.6853	0.8617	0.9369	0.9685	0.9757
0.0646	100	2.9836	-	-	-	-	-	-	-	-
0.1293	200	2.7758	-	-	-	-	-	-	-	-
0.1939	300	2.6386	2.3891	0.4003	0.5884	0.7387	0.8220	0.8695	0.9164	0.9372
0.2586	400	2.5466	-	-	-	-	-	-	-	-
0.3232	500	2.4711	-	-	-	-	-	-	-	-
0.3878	600	2.3918	2.1817	0.4580	0.6189	0.7230	0.7986	0.8554	0.8939	0.9146
0.4525	700	2.2802	-	-	-	-	-	-	-	-
0.5171	800	2.1309	-	-	-	-	-	-	-	-
0.5818	900	2.0585	1.8844	0.4932	0.6402	0.7482	0.8361	0.8665	0.8857	0.8895
0.6464	1000	2.0203	-	-	-	-	-	-	-	-
0.7111	1100	1.9934	-	-	-	-	-	-	-	-
0.7757	1200	1.9734	1.8208	0.5168	0.6452	0.7592	0.8371	0.8690	0.8775	0.8804
0.8403	1300	1.9583	-	-	-	-	-	-	-	-
0.9050	1400	1.9496	-	-	-	-	-	-	-	-
0.9696	1500	1.9499	1.8020	0.5159	0.6536	0.7568	0.8399	0.8670	0.8785	0.8778
-1	-1	-	-	0.5178	0.6515	0.7573	0.8407	0.8673	0.8781	0.8789

Environmental Impact

Carbon emissions were measured using CodeCarbon.

Energy Consumed: 0.110 kWh
Carbon Emitted: 0.043 kg of CO2
Hours Used: 0.274 hours

Training Hardware

On Cloud: No
GPU Model: 1 x NVIDIA GeForce RTX 3090
CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
RAM Size: 31.78 GB

Framework Versions

Python: 3.11.6
Sentence Transformers: 4.2.0.dev0
Transformers: 4.52.4
PyTorch: 2.6.0+cu124
Accelerate: 1.5.1
Datasets: 2.21.0
Tokenizers: 0.21.1

Citation

BibTeX

Sentence Transformers

bibtex

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

CSRLoss

bibtex

@misc{wen2025matryoshkarevisitingsparsecoding,
      title={Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation},
      author={Tiansheng Wen and Yifei Wang and Zequn Zeng and Zhong Peng and Yudi Su and Xinyang Liu and Bo Chen and Hongwei Liu and Stefanie Jegelka and Chenyu You},
      year={2025},
      eprint={2503.01776},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2503.01776},
}

SparseMultipleNegativesRankingLoss

bibtex

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}

	query	answer
type	string	string
details	min: 10 tokens mean: 11.71 tokens max: 26 tokens	min: 4 tokens mean: 131.81 tokens max: 450 tokens