SBERT Base Nli V2

🚀 SBERT-base-nli-v2

這個模型被用於論文 "SGPT: GPT Sentence Embeddings for Semantic Search" 和 "TSDAE: Using Transformer-based Sequential Denoising Auto-Encoder for Unsupervised Sentence Embedding Learning" 中，主要用於句子相似度計算、特徵提取等任務。

🚀 快速開始

若你想了解使用說明，請參考我們的代碼庫：https://github.com/Muennighoff/sgpt

若你想查看評估結果，請參考我們的論文：https://arxiv.org/abs/2202.08904

📦 安裝指南

當你安裝了 sentence-transformers 後，使用這個模型會變得很容易：

pip install -U sentence-transformers

💻 使用示例

基礎用法（Sentence-Transformers）

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

model = SentenceTransformer('{MODEL_NAME}')
embeddings = model.encode(sentences)
print(embeddings)

基礎用法（HuggingFace Transformers）

如果沒有安裝 sentence-transformers，你可以按如下方式使用該模型：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('{MODEL_NAME}')
model = AutoModel.from_pretrained('{MODEL_NAME}')

# 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)

評估結果

若要對這個模型進行自動評估，請查看 Sentence Embeddings Benchmark：https://seb.sbert.net

🔧 技術細節

訓練參數

該模型的訓練參數如下：

數據加載器： sentence_transformers.datasets.NoDuplicatesDataLoader.NoDuplicatesDataLoader，長度為 8807，參數如下：

{'batch_size': 64}

損失函數： sentence_transformers.losses.MultipleNegativesRankingLoss.MultipleNegativesRankingLoss，參數如下：

{'scale': 20.0, 'similarity_fct': 'cos_sim'}

fit() 方法的參數：

{
    "epochs": 1,
    "evaluation_steps": 880,
    "evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "<class 'transformers.optimization.AdamW'>",
    "optimizer_params": {
        "lr": 2e-05
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": null,
    "warmup_steps": 881,
    "weight_decay": 0.01
}

完整模型架構

SentenceTransformer(
  (0): Transformer({'max_seq_length': 75, '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})
)

📄 許可證

引用與作者

@article{muennighoff2022sgpt,
  title={SGPT: GPT Sentence Embeddings for Semantic Search},
  author={Muennighoff, Niklas},
  journal={arXiv preprint arXiv:2202.08904},
  year={2022}
}