Mcontriever Base Msmarco

🚀 mcontriever-base-msmarco

mcontriever-base-msmarco是一個句子轉換器模型，它可以將句子和段落映射到768維的密集向量空間，可用於聚類或語義搜索等任務。

該模型是從facebook的mcontriever-msmarco模型轉換而來。使用此模型時，請參考相關論文：Unsupervised Dense Information Retrieval with Contrastive Learning。

🚀 快速開始

本部分將介紹如何使用mcontriever-base-msmarco模型進行句子嵌入。

📦 安裝指南

若要使用該模型，需先安裝sentence-transformers庫：

pip install -U sentence-transformers

💻 使用示例

基礎用法

使用sentence-transformers庫調用模型：

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('nthakur/mcontriever-base-msmarco')
embeddings = model.encode(sentences)
print(embeddings)

高級用法

若未安裝sentence-transformers庫，可使用HuggingFace Transformers庫調用模型。首先，將輸入傳遞給Transformer模型，然後對上下文詞嵌入應用正確的池化操作。

from transformers import AutoTokenizer, AutoModel
import torch


#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)


# Sentences we want sentence embeddings for
sentences = ['This is an example sentence', 'Each sentence is converted']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('nthakur/mcontriever-base-msmarco')
model = AutoModel.from_pretrained('nthakur/mcontriever-base-msmarco')

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)

# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Sentence embeddings:")
print(sentence_embeddings)

📚 詳細文檔

評估結果

要對該模型進行自動評估，請參考句子嵌入基準測試：https://seb.sbert.net

完整模型架構

SentenceTransformer(
  (0): Transformer({'max_seq_length': 509, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
)

引用與作者

使用該模型時，請參考相關論文：Unsupervised Dense Information Retrieval with Contrastive Learning。