Opensearch Neural Sparse Encoding Doc V2 Mini

🚀 opensearch-neural-sparse-encoding-doc-v2-mini

This project offers a learned sparse retrieval model, which encodes documents into sparse vectors for efficient search. It supports retrieval with OpenSearch and provides good search relevance and efficiency.

🚀 Quick Start

When selecting a model, search relevance, model inference, and retrieval efficiency (FLOPS) should be considered. We benchmarked the models' zero-shot performance on a subset of the BEIR benchmark, including TrecCovid, NFCorpus, NQ, HotpotQA, FiQA, ArguAna, Touche, DBPedia, SCIDOCS, FEVER, Climate FEVER, SciFact, and Quora.

Overall, the v2 series of models have better search relevance, efficiency, and inference speed than the v1 series. However, the specific advantages and disadvantages may vary across different datasets.

✨ Features

• Paper: Towards Competitive Search Relevance For Inference-Free Learned Sparse Retrievers
• Fine-tuning sample: opensearch-sparse-model-tuning-sample

This is a learned sparse retrieval model that encodes documents into 30522-dimensional sparse vectors. For queries, it uses a tokenizer and a weight look-up table to generate sparse vectors. The non-zero dimension index represents the corresponding token in the vocabulary, and the weight indicates the importance of the token. The similarity score is the inner product of the query and document sparse vectors.

The training datasets include MS MARCO, eli5_question_answer, squad_pairs, WikiAnswers, yahoo_answers_title_question, gooaq_pairs, stackexchange_duplicate_questions_body_body, wikihow, S2ORC_title_abstract, stackexchange_duplicate_questions_title-body_title-body, yahoo_answers_question_answer, searchQA_top5_snippets, stackexchange_duplicate_questions_title_title, and yahoo_answers_title_answer.

OpenSearch neural sparse feature supports learned sparse retrieval with the lucene inverted index. Link: https://opensearch.org/docs/latest/query-dsl/specialized/neural-sparse/. Indexing and search can be performed using the OpenSearch high-level API.

💻 Usage Examples

Basic Usage

import json
import itertools
import torch

from transformers import AutoModelForMaskedLM, AutoTokenizer


# get sparse vector from dense vectors with shape batch_size * seq_len * vocab_size
def get_sparse_vector(feature, output):
    values, _ = torch.max(output*feature["attention_mask"].unsqueeze(-1), dim=1)
    values = torch.log(1 + torch.relu(values))
    values[:,special_token_ids] = 0
    return values
    
# transform the sparse vector to a dict of (token, weight)
def transform_sparse_vector_to_dict(sparse_vector):
    sample_indices,token_indices=torch.nonzero(sparse_vector,as_tuple=True)
    non_zero_values = sparse_vector[(sample_indices,token_indices)].tolist()
    number_of_tokens_for_each_sample = torch.bincount(sample_indices).cpu().tolist()
    tokens = [transform_sparse_vector_to_dict.id_to_token[_id] for _id in token_indices.tolist()]

    output = []
    end_idxs = list(itertools.accumulate([0]+number_of_tokens_for_each_sample))
    for i in range(len(end_idxs)-1):
        token_strings = tokens[end_idxs[i]:end_idxs[i+1]]
        weights = non_zero_values[end_idxs[i]:end_idxs[i+1]]
        output.append(dict(zip(token_strings, weights)))
    return output
    
# download the idf file from model hub. idf is used to give weights for query tokens
def get_tokenizer_idf(tokenizer):
    from huggingface_hub import hf_hub_download
    local_cached_path = hf_hub_download(repo_id="opensearch-project/opensearch-neural-sparse-encoding-doc-v2-mini", filename="idf.json")
    with open(local_cached_path) as f:
        idf = json.load(f)
    idf_vector = [0]*tokenizer.vocab_size
    for token,weight in idf.items():
        _id = tokenizer._convert_token_to_id_with_added_voc(token)
        idf_vector[_id]=weight
    return torch.tensor(idf_vector)

# load the model
model = AutoModelForMaskedLM.from_pretrained("opensearch-project/opensearch-neural-sparse-encoding-doc-v2-mini")
tokenizer = AutoTokenizer.from_pretrained("opensearch-project/opensearch-neural-sparse-encoding-doc-v2-mini")
idf = get_tokenizer_idf(tokenizer)

# set the special tokens and id_to_token transform for post-process
special_token_ids = [tokenizer.vocab[token] for token in tokenizer.special_tokens_map.values()]
get_sparse_vector.special_token_ids = special_token_ids
id_to_token = ["" for i in range(tokenizer.vocab_size)]
for token, _id in tokenizer.vocab.items():
    id_to_token[_id] = token
transform_sparse_vector_to_dict.id_to_token = id_to_token



query = "What's the weather in ny now?"
document = "Currently New York is rainy."

# encode the query
feature_query = tokenizer([query], padding=True, truncation=True, return_tensors='pt', return_token_type_ids=False)
input_ids = feature_query["input_ids"]
batch_size = input_ids.shape[0]
query_vector = torch.zeros(batch_size, tokenizer.vocab_size)
query_vector[torch.arange(batch_size).unsqueeze(-1), input_ids] = 1
query_sparse_vector = query_vector*idf

# encode the document
feature_document = tokenizer([document], padding=True, truncation=True, return_tensors='pt', return_token_type_ids=False)
output = model(**feature_document)[0]
document_sparse_vector = get_sparse_vector(feature_document, output)


# get similarity score
sim_score = torch.matmul(query_sparse_vector[0],document_sparse_vector[0])
print(sim_score)   # tensor(13.8344, grad_fn=<DotBackward0>)


query_token_weight = transform_sparse_vector_to_dict(query_sparse_vector)[0]
document_query_token_weight = transform_sparse_vector_to_dict(document_sparse_vector)[0]
for token in sorted(query_token_weight, key=lambda x:query_token_weight[x], reverse=True):
    if token in document_query_token_weight:
        print("score in query: %.4f, score in document: %.4f, token: %s"%(query_token_weight[token],document_query_token_weight[token],token))
        

        
# result:
# score in query: 5.7729, score in document: 1.0251, token: ny
# score in query: 4.5684, score in document: 1.1145, token: weather
# score in query: 3.5895, score in document: 0.5356, token: now
# score in query: 3.3313, score in document: 0.2710, token: ?

The above code sample shows an example of neural sparse search. Although there is no overlap token in the original query and document, this model performs a good match.

📚 Documentation

Detailed Search Relevance

📄 License

This project is licensed under the Apache v2.0 License.

Copyright

Copyright OpenSearch Contributors. See NOTICE for details.