Efficient SPLADE
Efficient SPLADE model for passage retrieval. This architecture uses two distinct models for query and document inference. This is the query one, please also download the doc one (https://huggingface.co/naver/efficient-splade-V-large-doc). For additional details, please visit:
- paper: https://dl.acm.org/doi/10.1145/3477495.3531833
- code: https://github.com/naver/splade
| | MRR@10 (MS MARCO dev) | R@1000 (MS MARCO dev) | Latency (PISA) ms | Latency (Inference) ms
| --- | --- | --- | --- | --- |
|
naver/efficient-splade-V-large| 38.8 | 98.0 | 29.0 | 45.3 |naver/efficient-splade-VI-BT-large| 38.0 | 97.8 | 31.1 | 0.7
Model Details
This is a Asymmetric SPLADE Sparse Encoder model. It maps sentences & paragraphs to a 30522-dimensional sparse vector space and can be used for semantic search and sparse retrieval.
Model Description
- Model Type: SPLADE Sparse Encoder
- Maximum Sequence Length: 512 tokens (256 for evaluation reproduction)
- Output Dimensionality: 30522 dimensions
- Similarity Function: Dot Product
Full Model Architecture
SparseEncoder(
(0): Router(
(query_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: DistilBertForMaskedLM
(query_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522})
(document_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: DistilBertForMaskedLM
(document_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. Note that with Sentence Transformers you load the entire model, i.e. the doc and query part.
from sentence_transformers import SparseEncoder
# Download from the 🤗 Hub
model = SparseEncoder("naver/efficient-splade-V-large-query")
# Run inference
queries = ["what causes aging fast"]
documents = [
"UV-A light, specifically, is what mainly causes tanning, skin aging, and cataracts, UV-B causes sunburn, skin aging and skin cancer, and UV-C is the strongest, and therefore most effective at killing microorganisms. Again â\x80\x93 single words and multiple bullets.",
"Answers from Ronald Petersen, M.D. Yes, Alzheimer's disease usually worsens slowly. But its speed of progression varies, depending on a person's genetic makeup, environmental factors, age at diagnosis and other medical conditions. Still, anyone diagnosed with Alzheimer's whose symptoms seem to be progressing quickly â\x80\x94 or who experiences a sudden decline â\x80\x94 should see his or her doctor.",
"Bell's palsy and Extreme tiredness and Extreme fatigue (2 causes) Bell's palsy and Extreme tiredness and Hepatitis (2 causes) Bell's palsy and Extreme tiredness and Liver pain (2 causes) Bell's palsy and Extreme tiredness and Lymph node swelling in children (2 causes)",
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 30522] [3, 30522]
# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[9.0047, 8.1454, 2.5808]])
Citation
If you use our checkpoint, please cite our work (need to update):
@inproceedings{10.1145/3477495.3531833,
author = {Lassance, Carlos and Clinchant, St\'{e}phane},
title = {An Efficiency Study for SPLADE Models},
year = {2022},
isbn = {9781450387323},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3477495.3531833},
doi = {10.1145/3477495.3531833},
abstract = {Latency and efficiency issues are often overlooked when evaluating IR models based on Pretrained Language Models (PLMs) in reason of multiple hardware and software testing scenarios. Nevertheless, efficiency is an important part of such systems and should not be overlooked. In this paper, we focus on improving the efficiency of the SPLADE model since it has achieved state-of-the-art zero-shot performance and competitive results on TREC collections. SPLADE efficiency can be controlled via a regularization factor, but solely controlling this regularization has been shown to not be efficient enough. In order to reduce the latency gap between SPLADE and traditional retrieval systems, we propose several techniques including L1 regularization for queries, a separation of document/query encoders, a FLOPS-regularized middle-training, and the use of faster query encoders. Our benchmark demonstrates that we can drastically improve the efficiency of these models while increasing the performance metrics on in-domain data. To our knowledge, we propose the first neural models that, under the same computing constraints, achieve similar latency (less than 4ms difference) as traditional BM25, while having similar performance (less than 10% MRR@10 reduction) as the state-of-the-art single-stage neural rankers on in-domain data.},
booktitle = {Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval},
pages = {2220–2226},
numpages = {7},
keywords = {splade, latency, information retrieval, sparse representations},
location = {Madrid, Spain},
series = {SIGIR '22}
}