Sbert Chinese Qmc Finance V1

🚀 DMetaSoul/sbert-chinese-qmc-finance-v1

This model is based on the BERT model of the bert-base-chinese version. It is trained and tuned on a large-scale bank question matching dataset (BQCorpus), and is suitable for the scenario of question matching in the financial field. For example:

• 8 thousand with 400 yuan of daily interest? VS How much is the daily interest for 10,000 yuan?
• Early repayment is calculated based on the full amount VS How to repay when the repayment deduction fails?
• Why did my borrowing transaction fail? VS Why did the newly applied borrowing fail?

Note: The lightweight version of this model has also been open-sourced!

🚀 Quick Start

✨ Features

• Based on the BERT model of the bert-base-chinese version.
• Trained and tuned on a large-scale bank question matching dataset.
• Suitable for question matching in the financial field.

📦 Installation

You can install the necessary libraries according to the usage method you choose.

Install sentence-transformers

pip install -U sentence-transformers

💻 Usage Examples

Basic Usage

If you use the sentence-transformers framework:

from sentence_transformers import SentenceTransformer
sentences = ["到期不能按时还款怎么办", "剩余欠款还有多少？"]

model = SentenceTransformer('DMetaSoul/sbert-chinese-qmc-finance-v1')
embeddings = model.encode(sentences)
print(embeddings)

Advanced Usage

If you use the HuggingFace Transformers:

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 = ["到期不能按时还款怎么办", "剩余欠款还有多少？"]

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('DMetaSoul/sbert-chinese-qmc-finance-v1')
model = AutoModel.from_pretrained('DMetaSoul/sbert-chinese-qmc-finance-v1')

# 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)

📚 Documentation

Evaluation

This model is evaluated on several public semantic matching datasets, and the correlation coefficient between the vector similarity and the real label is calculated:

📄 License

No license information is provided in the original document.

Citing & Authors

E-mail: xiaowenbin@dmetasoul.com