from transformers import AutoTokenizer, pipeline
from onnxruntime import InferenceSession

tokenizer = AutoTokenizer.from_pretrained("您的用户名/模型名")
session = InferenceSession("model.onnx")

# 预处理
inputs = tokenizer("I have a headache", return_tensors="np")

# 推理
outputs = session.run(None, dict(inputs))
predicted_id = outputs[0].argmax()


##前端使用方式
<!DOCTYPE html>
<html lang="en">
  <head>
    <meta charset="UTF-8" />
    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
    <title>Transformers.js Example</title>
  </head>
  <body>
    <h1>Transformers.js in Browser</h1>
    <script type="module">
      import { pipeline } from 'https://cdn.jsdelivr.net/npm/@xenova/transformers';
        // 创建一个pipeline
        const classifier = await pipeline('text-classification','xuxiaoda/drug_classification');
        // 推理
        const result = await classifier('What are the drugs for treating high blood pressure?');
        console.log(result[0].label);
    </script>
  </body>
</html>