S DagoBERT STSb

🚀 {s-DagoBERT-STSb}

{s-DagoBERT-STSb} 是一個 sentence-transformers 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。

🚀 快速開始

安裝依賴

若要使用此模型，需先安裝 sentence-transformers：

pip install -U sentence-transformers

基本使用

安裝完成後，即可按以下方式使用該模型：

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)

高級使用（不使用 sentence-transformers）

若未安裝 sentence-transformers，可按以下步驟使用該模型：首先，將輸入數據傳入 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)

✨ 主要特性

• 向量映射：能夠將句子和段落映射到 768 維的密集向量空間。
• 多任務適用性：可用於聚類、語義搜索等多種自然語言處理任務。

📦 安裝指南

使用此模型前，需安裝 sentence-transformers，安裝命令如下：

pip install -U sentence-transformers

💻 使用示例

基礎用法

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)

高級用法

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)

📚 詳細文檔

評估結果

若要對該模型進行自動化評估，可參考 Sentence Embeddings Benchmark：https://seb.sbert.net

訓練參數

該模型的訓練參數如下：

數據加載器

torch.utils.data.dataloader.DataLoader，長度為 44，參數如下：

{'batch_size': 128, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}

損失函數

sentence_transformers.losses.CosineSimilarityLoss.CosineSimilarityLoss

訓練方法參數

{
    "epochs": 15,
    "evaluation_steps": 360,
    "evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
    "optimizer_params": {
        "lr": 3e-05
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": null,
    "warmup_steps": 50,
    "weight_decay": 0
}

完整模型架構

SentenceTransformer(
  (0): Transformer({'max_seq_length': 80, 'do_lower_case': False}) with Transformer model: RobertaModel 
  (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})
)

🔧 技術細節

該模型基於 sentence-transformers 框架構建，利用 Transformer 架構將句子和段落映射到 768 維的密集向量空間。在訓練過程中，使用了 CosineSimilarityLoss 損失函數，並結合 AdamW 優化器進行參數更新。通過池化操作（如均值池化）對上下文詞嵌入進行處理，以獲得句子級別的向量表示。

📄 許可證

文檔中未提及許可證相關信息。

Citing & Authors

文檔中未提供相關信息。