You can deploy this model and start earning money today!

tomaarsen/
inference-free-splade-bert-tiny-nq-3e-3-lambda-corpus

Name: inference-free-splade-bert-tiny-nq-3e-3-lambda-corpus
Availability: InStock

Views

No views yet

Inference-free SPLADE BERT-tiny trained on Natural-Questions tuples

This is a Asymmetric Inference-free SPLADE Sparse Encoder model trained on the natural-questions dataset using the sentence-transformers library. It maps sentences & paragraphs to a 30522-dimensional sparse vector space and can be used for semantic search and sparse retrieval.

Model Details

Model Description

Model Type: Asymmetric Inference-free SPLADE Sparse Encoder

Maximum Sequence Length: 512 tokens
Output Dimensionality: 30522 dimensions
Similarity Function: Dot Product
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): Asym(
    (query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast)
    (corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM 
    (corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522})
  )
)

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/inference-free-splade-bert-tiny-nq-3e-3-lambda-corpus")
# Run inference
sentences = [
    'when did the american civil rights movement end',
    'African-American civil rights movement (1954–1968) The Civil Rights Movement (also known as the American civil rights movement, African-American civil rights movement, and other terms,[b]) was a human rights movement from 1954â€“1968 that encompassed strategies, groups, and social movements to accomplish its goal of ending legalized racial segregation and discrimination laws in the United States. The movement secured the legal recognition and federal protection of black Americans in the United States Constitution and federal law.',
    'Paleolithic Paleolithic humans made tools of stone, bone, and wood.[23] The early paleolithic hominins, Australopithecus, were the first users of stone tools. Excavations in Gona, Ethiopia have produced thousands of artifacts, and through radioisotopic dating and magnetostratigraphy, the sites can be firmly dated to 2.6Â\xa0million years ago. Evidence shows these early hominins intentionally selected raw materials with good flaking qualities and chose appropriate sized stones for their needs to produce sharp-edged tools for cutting.[29]',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# (3, 30522)

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Sparse Information Retrieval

Datasets: NanoMSMARCO, NanoNFCorpus and NanoNQ
Evaluated with SparseInformationRetrievalEvaluator

Metric	NanoMSMARCO	NanoNFCorpus	NanoNQ
dot_accuracy@1	0.3	0.46	0.24
dot_accuracy@3	0.46	0.56	0.5
dot_accuracy@5	0.52	0.6	0.64
dot_accuracy@10	0.66	0.66	0.74
dot_precision@1	0.3	0.46	0.24
dot_precision@3	0.1533	0.36	0.1667
dot_precision@5	0.104	0.312	0.128
dot_precision@10	0.066	0.248	0.074
dot_recall@1	0.3	0.0438	0.24
dot_recall@3	0.46	0.0729	0.48
dot_recall@5	0.52	0.1097	0.61
dot_recall@10	0.66	0.1332	0.7
dot_ndcg@10	0.4614	0.3231	0.4673
dot_mrr@10	0.3999	0.5245	0.4007
dot_map@100	0.4174	0.1394	0.3953

Sparse Nano BEIR

Dataset: NanoBEIR_mean
Evaluated with SparseNanoBEIREvaluator with these parameters:
json
1{ 2 "dataset_names": [ 3 "msmarco", 4 "nfcorpus", 5 "nq" 6 ] 7}

Metric	Value
dot_accuracy@1	0.3333
dot_accuracy@3	0.5067
dot_accuracy@5	0.5867
dot_accuracy@10	0.6867
dot_precision@1	0.3333
dot_precision@3	0.2267
dot_precision@5	0.1813
dot_precision@10	0.1293
dot_recall@1	0.1946
dot_recall@3	0.3376
dot_recall@5	0.4132
dot_recall@10	0.4977
dot_ndcg@10	0.4172
dot_mrr@10	0.4417
dot_map@100	0.3174

Training Details

Training Dataset

natural-questions

Dataset: natural-questions at f9e894e
Size: 99,000 training samples
Columns: query and corpus
Approximate statistics based on the first 1000 samples:
query corpus
type dict dict
details

Samples:

query	corpus
`{'query': "who played the father in papa don't preach"}`	`{'corpus': 'Alex McArthur Alex McArthur (born March 6, 1957) is an American actor.'}`
`{'query': 'where was the location of the battle of hastings'}`	`{'corpus': '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.'}`
`{'query': 'how many puppies can a dog give birth to'}`	`{'corpus': '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: SpladeLoss with these parameters:
json
1{'loss': SparseMultipleNegativesRankingLoss( 2 (model): SparseEncoder( 3 (0): Asym( 4 (query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast) 5 (corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM 6 (corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522}) 7 ) 8 ) 9 (cross_entropy_loss): CrossEntropyLoss() 10), 'lambda_corpus': 0.003, 'lambda_query': 0, 'corpus_regularizer': FlopsLoss( 11 (model): SparseEncoder( 12 (0): Asym( 13 (query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast) 14 (corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM 15 (corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522}) 16 ) 17 ) 18), 'query_regularizer': None}

Evaluation Dataset

natural-questions

Dataset: natural-questions at f9e894e
Size: 1,000 evaluation samples
Columns: query and corpus
Approximate statistics based on the first 1000 samples:
query corpus
type dict dict
details

Samples:

query	corpus
`{'query': 'where is the tiber river located in italy'}`	{'corpus': '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\xa0mi) 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\xa0sq\xa0mi). The river has achieved lasting fame as the main watercourse of the city of Rome, founded on its eastern banks.'}
`{'query': 'what kind of car does jay gatsby drive'}`	`{'corpus': '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.'}`
`{'query': 'who sings if i can dream about you'}`	`{'corpus': '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: SpladeLoss with these parameters:
json
1{'loss': SparseMultipleNegativesRankingLoss( 2 (model): SparseEncoder( 3 (0): Asym( 4 (query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast) 5 (corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM 6 (corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522}) 7 ) 8 ) 9 (cross_entropy_loss): CrossEntropyLoss() 10), 'lambda_corpus': 0.003, 'lambda_query': 0, 'corpus_regularizer': FlopsLoss( 11 (model): SparseEncoder( 12 (0): Asym( 13 (query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast) 14 (corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM 15 (corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522}) 16 ) 17 ) 18), 'query_regularizer': None}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 64
per_device_eval_batch_size: 64
learning_rate: 2e-05
num_train_epochs: 1
warmup_ratio: 0.1
fp16: 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: 2e-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: 0.1
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: False
fp16: True
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
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: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	NanoMSMARCO_dot_ndcg@10	NanoNFCorpus_dot_ndcg@10	NanoNQ_dot_ndcg@10	NanoBEIR_mean_dot_ndcg@10
0.0129	20	1.5874	-	-	-	-	-
0.0259	40	1.7845	-	-	-	-	-
0.0388	60	1.924	-	-	-	-	-
0.0517	80	1.6441	-	-	-	-	-
0.0646	100	1.1842	-	-	-	-	-
0.0776	120	0.9556	-	-	-	-	-
0.0905	140	0.8623	-	-	-	-	-
0.1034	160	0.7888	-	-	-	-	-
0.1164	180	0.7923	-	-	-	-	-
0.1293	200	0.7464	0.6235	0.4629	0.3201	0.4511	0.4113
0.1422	220	0.7121	-	-	-	-	-
0.1551	240	0.6795	-	-	-	-	-
0.1681	260	0.7187	-	-	-	-	-
0.1810	280	0.6786	-	-	-	-	-
0.1939	300	0.6608	-	-	-	-	-
0.2069	320	0.6625	-	-	-	-	-
0.2198	340	0.651	-	-	-	-	-
0.2327	360	0.6671	-	-	-	-	-
0.2456	380	0.6732	-	-	-	-	-
0.2586	400	0.6301	0.5740	0.4243	0.3278	0.4273	0.3931
0.2715	420	0.6375	-	-	-	-	-
0.2844	440	0.6651	-	-	-	-	-
0.2973	460	0.6378	-	-	-	-	-
0.3103	480	0.6592	-	-	-	-	-
0.3232	500	0.6404	-	-	-	-	-
0.3361	520	0.6216	-	-	-	-	-
0.3491	540	0.6072	-	-	-	-	-
0.3620	560	0.6508	-	-	-	-	-
0.3749	580	0.5645	-	-	-	-	-
0.3878	600	0.6275	0.4993	0.4352	0.3241	0.4227	0.3940
0.4008	620	0.566	-	-	-	-	-
0.4137	640	0.5063	-	-	-	-	-
0.4266	660	0.5297	-	-	-	-	-
0.4396	680	0.5448	-	-	-	-	-
0.4525	700	0.5436	-	-	-	-	-
0.4654	720	0.4771	-	-	-	-	-
0.4783	740	0.5035	-	-	-	-	-
0.4913	760	0.5005	-	-	-	-	-
0.5042	780	0.4509	-	-	-	-	-
0.5171	800	0.4956	0.4341	0.4596	0.3280	0.4357	0.4078
0.5301	820	0.4876	-	-	-	-	-
0.5430	840	0.4622	-	-	-	-	-
0.5559	860	0.4791	-	-	-	-	-
0.5688	880	0.4608	-	-	-	-	-
0.5818	900	0.451	-	-	-	-	-
0.5947	920	0.4537	-	-	-	-	-
0.6076	940	0.4233	-	-	-	-	-
0.6206	960	0.4534	-	-	-	-	-
0.6335	980	0.4701	-	-	-	-	-
0.6464	1000	0.4017	0.4052	0.4692	0.3271	0.4452	0.4138
0.6593	1020	0.4518	-	-	-	-	-
0.6723	1040	0.4173	-	-	-	-	-
0.6852	1060	0.4369	-	-	-	-	-
0.6981	1080	0.456	-	-	-	-	-
0.7111	1100	0.448	-	-	-	-	-
0.7240	1120	0.4369	-	-	-	-	-
0.7369	1140	0.4394	-	-	-	-	-
0.7498	1160	0.437	-	-	-	-	-
0.7628	1180	0.4402	-	-	-	-	-
0.7757	1200	0.4382	0.3901	0.4623	0.3238	0.4664	0.4175
0.7886	1220	0.4111	-	-	-	-	-
0.8016	1240	0.4386	-	-	-	-	-
0.8145	1260	0.4136	-	-	-	-	-
0.8274	1280	0.4439	-	-	-	-	-
0.8403	1300	0.4423	-	-	-	-	-
0.8533	1320	0.4339	-	-	-	-	-
0.8662	1340	0.4124	-	-	-	-	-
0.8791	1360	0.417	-	-	-	-	-
0.8920	1380	0.4067	-	-	-	-	-
0.9050	1400	0.414	0.3854	0.4591	0.3234	0.4660	0.4162
0.9179	1420	0.4153	-	-	-	-	-
0.9308	1440	0.3889	-	-	-	-	-
0.9438	1460	0.4368	-	-	-	-	-
0.9567	1480	0.4241	-	-	-	-	-
0.9696	1500	0.423	-	-	-	-	-
0.9825	1520	0.4287	-	-	-	-	-
0.9955	1540	0.4282	-	-	-	-	-
-1	-1	-	-	0.4614	0.3231	0.4673	0.4172

Environmental Impact

Carbon emissions were measured using CodeCarbon.

Energy Consumed: 0.030 kWh
Carbon Emitted: 0.011 kg of CO2
Hours Used: 0.087 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: 4.49.0
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",
}

SpladeLoss

bibtex

@misc{formal2022distillationhardnegativesampling,
      title={From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective},
      author={Thibault Formal and Carlos Lassance and Benjamin Piwowarski and Stéphane Clinchant},
      year={2022},
      eprint={2205.04733},
      archivePrefix={arXiv},
      primaryClass={cs.IR},
      url={https://arxiv.org/abs/2205.04733},
}

	query	corpus
type	dict	dict
details

	query	corpus
type	dict	dict
details