![\"\"](\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/07\/63b6294451c0b3443eb897574a95873dbebdf0-300x199-1.png\")
<\/p>\n\u4eca\u5929\uff0c\u6211\u4eec\u5c06\u4f7f\u7528Hugging Face\u3001FastAPI\u548cDocker\u90e8\u7f72\u4e00\u4e2a\u7b80\u5355\u7684\u6a21\u578b\uff0c\u6f14\u793a\u5982\u4f55\u6709\u6548\u5730\u5b9e\u73b0\u8fd9\u4e2a\u76ee\u6807\u3002<\/p>\n
<\/p>\n
\u672c\u6587\u4ecb\u7ecd\u4e86\u4f7f\u7528FastAPI\u548cDocker\u751f\u6210\u4e00\u4e2a\u968f\u65f6\u53ef\u7528\u7684Hugging Face\u6a21\u578b\u3002<\/p>\n
![\"\"](\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/07\/e4f5955523d89ac4b97708e97ffb4789be89ce-300x118-1.png\")
<\/p>\n
\u60f3\u8c61\u4e00\u4e0b\uff0c\u5229\u7528Hugging Face\u6a21\u578b\u6765\u786e\u5b9a\u8bc4\u8bba\u7684\u60c5\u7eea\u3002\u5728\u8fc7\u53bb\uff0c\u7b2c\u4e00\u6b65\u662f\u5236\u4f5c\u8fd9\u6837\u4e00\u4e2a\u6a21\u578b\uff0c\u5e76\u786e\u4fdd\u5b83\u6b63\u5e38\u5de5\u4f5c\u3002<\/p>\n
\u7136\u800c\uff0c\u4eca\u5929\u7684\u9884\u8bad\u7ec3\u6a21\u578b\u8ba9\u6211\u4eec\u53ea\u9700\u82b1\u5f88\u5c11\u7684\u7cbe\u529b\uff0c\u5c31\u80fd\u51c6\u5907\u597d\u8fd9\u6837\u7684\u5927\u8bed\u8a00\u6a21\u578b(LLM)\u3002<\/p>\n
\u4e00\u65e6\u6211\u4eec\u51c6\u5907\u597d\u4f7f\u7528\u8fd9\u4e2a\u6a21\u578b\uff0c\u4e3b\u8981\u76ee\u6807\u662f\u8ba9\u516c\u53f8\u7684\u540c\u4e8b\u80fd\u591f\u4f7f\u7528\u8fd9\u4e2a\u6a21\u578b\uff0c\u800c\u4e0d\u9700\u8981\u4e0b\u8f7d\u6216\u4ece\u5934\u5f00\u59cb\u5b9e\u73b0\u5b83\u3002<\/p>\n
\u4e3a\u6b64\uff0c\u6211\u4eec\u5c06\u521b\u5efa\u4e00\u4e2a\u7aef\u70b9API\uff0c\u4f7f\u7528\u6237\u80fd\u591f\u72ec\u7acb\u5730\u8c03\u7528\u548c\u4f7f\u7528\u6a21\u578b\u3002\u8fd9\u5c31\u662f\u6211\u4eec\u6240\u8bf4\u7684\u4ece\u5934\u5230\u5c3e\u6784\u5efa\u7684\u7aef\u5230\u7aef\u9879\u76ee\u3002<\/p>\n
\u4eca\u5929\uff0c\u6211\u4eec\u5c06\u4f7f\u7528Hugging Face\u3001FastAPI\u548cDocker\u90e8\u7f72\u4e00\u4e2a\u7b80\u5355\u7684\u6a21\u578b\uff0c\u6f14\u793a\u5982\u4f55\u6709\u6548\u5730\u5b9e\u73b0\u8fd9\u4e2a\u76ee\u6807\u3002<\/p>\n
\u9996\u5148\u8981\u505a\u7684\u662f\u9009\u62e9\u4e00\u4e2a\u9002\u5408\u6211\u4eec\u9700\u8981\u7684Hugging Face\u6a21\u578b\u3002\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u5728\u6211\u4eec\u7684\u73af\u5883\u4e2d\u8f7b\u677e\u5b89\u88c5Hugging Face\uff1a<\/p>\n
\u590d\u5236<\/p>\n
pip install transformers # remember to work with transformers we need either tensorflow or pytorch installed as well pip install torch pip install tensorflow<\/p>\n
\u73b0\u5728\uff0c\u6211\u4eec\u9700\u8981\u5bfc\u5165Transformer\u5e93\u7684\u7ba1\u9053\u547d\u4ee4\u3002<\/p>\n
\u590d\u5236<\/p>\n
from transformers import pipeline<\/p>\n
\u7136\u540e\uff0c\u4f7f\u7528pipeline\u547d\u4ee4\uff0c\u6211\u4eec\u53ef\u4ee5\u8f7b\u677e\u751f\u6210\u4e00\u4e2a\u6a21\u578b\u6765\u5b9a\u4e49\u7279\u5b9a\u6587\u672c\u7684\u60c5\u7eea\u3002\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u4e24\u79cd\u4e0d\u540c\u7684\u65b9\u6cd5\u6765\u505a\u5230\u8fd9\u4e00\u70b9\uff1a\u901a\u8fc7\u5b9a\u4e49\u4efb\u52a1\u201c\u60c5\u7eea\u5206\u6790\u201d\u6216\u901a\u8fc7\u5b9a\u4e49\u6a21\u578b\uff0c\u5982\u4e0b\u9762\u7684\u4ee3\u7801\u6240\u793a\u3002<\/p>\n
\u590d\u5236<\/p>\n
# Defining directly the task we want to implement. pipe = pipeline(task=”sentiment-analysis”) # Defining the model we choose. pipe = pipeline(model=”model-to-be-used”)<\/p>\n
\u503c\u5f97\u4e00\u63d0\u7684\u662f\uff0c\u4e0d\u5efa\u8bae\u4f7f\u7528\u57fa\u4e8e\u4efb\u52a1\u7684\u65b9\u6cd5\uff0c\u56e0\u4e3a\u5b83\u9650\u5236\u4e86\u6211\u4eec\u5bf9\u6240\u4f7f\u7528\u7684\u7279\u5b9a\u6a21\u578b\u7684\u63a7\u5236\u3002<\/p>\n
\u5728\u672c\u6587\u4f8b\u5b50\u4e2d\uff0c\u6211\u9009\u62e9\u4e86\u201cdistilbert-base-uncase-fine tuned-sst-2-english\u201d\uff0c\u4f46\u4f60\u53ef\u4ee5\u968f\u610f\u6d4f\u89c8Hugging Face Hub\uff0c\u9009\u62e9\u9002\u5408\u9700\u8981\u7684\u4efb\u4f55\u578b\u53f7\u3002\u4f60\u53ef\u4ee5\u5728\u4e0b\u9762\u7684\u6587\u7ae0(https:\/\/www.datacamp.com\/tutorial\/what-is-hugging-face)\u4e2d\u627e\u5230Hugging Face\u7684\u7b80\u5355\u6307\u5357\u3002<\/p>\n
\u590d\u5236<\/p>\n
pipe = pipeline(model=”distilbert\/distilbert-base-uncased-finetuned-sst-2-english”)<\/p>\n
\u6211\u4eec\u5df2\u5b9a\u4e49\u4e86\u7ba1\u9053\u6a21\u578b\uff0c\u53ea\u9700\u53d1\u9001\u4e00\u4e2a\u7b80\u5355\u7684\u63d0\u793a\uff0c\u5c31\u53ef\u4ee5\u8fd4\u56de\u7ed3\u679c\u3002\u6bd4\u5982\u8bf4\uff0c\u8f93\u5165\u4ee5\u4e0b\u547d\u4ee4\uff1a<\/p>\n
\u590d\u5236<\/p>\n
print(pipe(“This tutorial is great!”))<\/p>\n
\u6211\u4eec\u5c06\u5f97\u5230[{‘label’: ‘POSITIVE’\uff0c ‘score’: 0.9998689889907837}]<\/p>\n
\u60f3\u8c61\u4e00\u4e0b\uff0c\u6211\u4eec\u5e0c\u671b\u6211\u4eec\u7684\u7528\u6237\u5f97\u5230\u4e00\u4e2a\u5173\u4e8e\u8fd9\u4e2a\u5206\u7c7b\u7684\u81ea\u7136\u8bed\u8a00\u53e5\u5b50\u3002\u6211\u4eec\u4e5f\u53ef\u4ee5\u5b9e\u65bd\u7b80\u5355\u7684Python\u4ee3\u7801\u540c\u6837\u5b9e\u73b0\u8fd9\u4e2a\u76ee\u7684\uff1a<\/p>\n
\u590d\u5236<\/p>\n
def generate_response(prompt:str): response = pipe(“This is a great tutorial!”) label = response[0][“label”] score = response[0][“score”] return f”The ‘{prompt}’ input is {label} with a score of {score}” print(generate_response(“This tutorial is great!”))<\/p>\n
\u91cd\u590d\u540c\u6837\u7684\u8bd5\u9a8c\uff0c\u6211\u4eec\u4f1a\u5f97\u5230\uff1a<\/p>\n
\u590d\u5236<\/p>\n
The ‘This tutorial is great!’ input is POSITIVE with a score of 0.9997909665107727<\/p>\n
\u73b0\u5728\u6211\u4eec\u6709\u4e86\u4e00\u4e2a\u5207\u5b9e\u53ef\u884c\u7684\u6a21\u578b\uff0c\u53ef\u4ee5\u7ee7\u7eed\u5b9a\u4e49\u6211\u4eec\u7684API\u3002<\/p>\n
\u4e3a\u4e86\u5b9a\u4e49API\uff0c\u6211\u4eec\u5c06\u4f7f\u7528FastAPI\u3002\u5b83\u662f\u4e00\u4e2a\u7528\u4e8e\u6784\u5efa\u9ad8\u6027\u80fdWeb API\u7684Python\u6846\u67b6\u3002\u9996\u5148\uff0c\u4f7f\u7528pip\u547d\u4ee4\u5b89\u88c5FastAPI\u5e93\uff0c\u5e76\u5c06\u5176\u5bfc\u5165\u5230\u6211\u4eec\u7684\u73af\u5883\u4e2d\u3002\u6b64\u5916\uff0c\u6211\u4eec\u5c06\u5229\u7528pydantic\u5e93\u6765\u786e\u4fdd\u8f93\u5165\u662f\u6240\u9700\u7684\u7c7b\u578b\u3002<\/p>\n
\u4e0b\u9762\u7684\u4ee3\u7801\u5c06\u751f\u6210\u5207\u5b9e\u53ef\u884c\u7684API\uff0c\u6211\u4eec\u7684\u540c\u4e8b\u53ef\u4ee5\u76f4\u63a5\u4f7f\u7528\u3002<\/p>\n
\u590d\u5236<\/p>\n
from fastapi import FastAPI from pydantic import BaseModel from transformers import pipeline # You can check any other model in the Hugging Face Hub pipe = pipeline(model=”distilbert\/distilbert-base-uncased-finetuned-sst-2-english”) # We define the app app = FastAPI() # We define that we expect our input to be a string class RequestModel(BaseModel): input: str # Now we define that we accept post requests @app.post(“\/sentiment”) def get_response(request: RequestModel): prompt = request.input response = pipe(prompt) label = response[0][“label”] score = response[0][“score”] return f”The ‘{prompt}’ input is {label} with a score of {score}”<\/p>\n
\u4e0b\u9762\u662f\u4ee3\u7801\u4e2d\u9010\u6b65\u53d1\u751f\u7684\u4e8b\u60c5\uff1a<\/p>\n
\u5982\u679c\u6211\u4eec\u6267\u884c\u4ee3\u7801\uff0cAPI\u5c06\u5728\u672c\u5730\u4e3b\u673a\u4e2d\u53ef\u7528\uff0c\u5982\u4e0b\u56fe\u6240\u793a\uff1a<\/p>\n
![\"\"](\"https:\/\/aiforumimage.oss-cn-shanghai.aliyuncs.com\/wp-content\/uploads\/2024\/07\/28fb42c302a8cc0dcbe419de6f84a4255539c2-300x99-1.png\")
<\/p>\n
\u7b80\u800c\u8a00\u4e4b\uff0c\u8fd9\u6bb5\u4ee3\u7801\u8bbe\u7f6e\u7b80\u5355\u7684Web\u670d\u52a1\uff0c\u4f60\u53ef\u4ee5\u5f80\u8be5\u670d\u52a1\u53d1\u9001\u4e00\u6bb5\u6587\u672c\uff0c\u5176\u7ed9\u51fa\u7684\u56de\u590d\u662f\u5206\u6790\u8be5\u6587\u672c\u7684\u60c5\u7eea\uff0c\u901a\u8fc7FastAPI\u5145\u5206\u5229\u7528Hugging Face\u6a21\u578b\u7684\u5f3a\u5927\u529f\u80fd\u3002<\/p>\n
\u63a5\u4e0b\u6765\uff0c\u6211\u4eec\u5e94\u8be5\u5c06\u5e94\u7528\u7a0b\u5e8f\u5bb9\u5668\u5316\uff0c\u4ee5\u4fbf\u53ef\u4ee5\u5728\u4efb\u4f55\u5730\u65b9\u6267\u884c\uff0c\u800c\u4e0d\u4ec5\u4ec5\u662f\u5728\u672c\u5730\u8ba1\u7b97\u673a\u4e0a\u6267\u884c\u3002\u8fd9\u5c06\u786e\u4fdd\u66f4\u597d\u7684\u53ef\u79fb\u690d\u6027\u548c\u6613\u4e8e\u90e8\u7f72\u3002<\/p>\n
\u5bb9\u5668\u5316\u9700\u8981\u5c06\u5e94\u7528\u7a0b\u5e8f\u653e\u5165\u5bb9\u5668\u4e2d\u3002Docker\u5bb9\u5668\u8fd0\u884cDocker\u955c\u50cf\u7684\u5b9e\u4f8b\uff0c\u8fd9\u5305\u62ec\u5b83\u81ea\u5df1\u7684\u64cd\u4f5c\u7cfb\u7edf\u548c\u5e94\u7528\u7a0b\u5e8f\u6240\u9700\u7684\u6240\u6709\u4f9d\u8d56\u9879\u3002<\/p>\n
\u6bd4\u5982\u8bf4\uff0c\u4f60\u53ef\u4ee5\u5728\u5bb9\u5668\u4e2d\u5b89\u88c5Python\u548c\u6240\u6709\u5fc5\u9700\u7684\u5305\uff0c\u8fd9\u6837\u5b83\u53ef\u4ee5\u5728\u4efb\u4f55\u5730\u65b9\u8fd0\u884c\uff0c\u4e0d\u9700\u8981\u5b89\u88c5\u8fd9\u4e9b\u5e93\u3002<\/p>\n
\u4e3a\u4e86\u5728Docker\u5bb9\u5668\u4e2d\u8fd0\u884c\u6211\u4eec\u7684\u60c5\u7eea\u5206\u6790\u5e94\u7528\u7a0b\u5e8f\uff0c\u6211\u4eec\u5148\u9700\u8981\u521b\u5efaDocker\u955c\u50cf\u3002\u8fd9\u4e2a\u8fc7\u7a0b\u5305\u62ec\u5199\u4e00\u4e2aDockerfile\uff0c\u6307\u5b9aDocker\u955c\u50cf\u5e94\u8be5\u542b\u6709\u4ec0\u4e48\u3002<\/p>\n
\u5982\u679c\u4f60\u7684\u7cfb\u7edf\u6ca1\u6709\u5b89\u88c5Docker\uff0c\u53ef\u4ee5\u4eceDocker\u7684\u7f51\u7ad9\u4e0a\u4e0b\u8f7d\u3002\u8fd9\u662f\u6211\u4eec\u5c06\u5728\u8fd9\u4e2a\u9879\u76ee\u4e2d\u4f7f\u7528\u7684Dockerfile\uff0c\u5728\u5b58\u50a8\u5e93\u4e2d\u540d\u4e3aDockerfile\u3002<\/p>\n
\u590d\u5236<\/p>\n
# Use an official Python runtime as a parent image FROM python:3.10-slim # Set the working directory in the container WORKDIR \/sentiment # Copy the requirements.txt file into the root COPY requirements.txt . # Copy the current directory contents into the container at \/app as well COPY .\/app .\/app # Install any needed packages specified in requirements.txt RUN pip install -r requirements.txt # Make port 8000 available to the world outside this container EXPOSE 8000 # Run main.py when the container launches, as it is contained under the app folder, we define app.main CMD [“uvicorn”, “app.main:app”, “–host”, “0.0.0.0”, “–port”, “8000”]<\/p>\n
\u7136\u540e\uff0c\u6211\u4eec\u53ea\u9700\u8981\u5728\u7ec8\u7aef\u4e2d\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u6765\u6784\u5efaDocker\u955c\u50cf\u3002<\/p>\n
\u590d\u5236<\/p>\n
docker build -t sentit-app<\/p>\n
\u7136\u540e\u4e3a\u4e86\u6267\u884c\uff0c\u6211\u4eec\u6709\u4e24\u4e2a\u9009\u9879\uff1a<\/p>\n
\u4f7f\u7528\u5e26\u6709\u547d\u4ee4\u7684\u7ec8\u7aef\u3002<\/p>\n
\u590d\u5236<\/p>\n
docker run -p 8000:8000 –name name_of_cointainer sentiment-hf<\/p>\n
\u4f7f\u7528docker hub\u3002\u6211\u4eec\u5f88\u5bb9\u6613\u8fdb\u5165docker hub\uff0c\u70b9\u51fb\u955c\u50cf\u7684\u8fd0\u884c\u6309\u94ae\u3002<\/p>\n
\u8fd9\u5c31\u662f\u5168\u90e8\u7ec6\u8282!\u73b0\u5728\uff0c\u6211\u4eec\u6709\u4e86\u4e00\u4e2a\u5207\u5b9e\u53ef\u884c\u7684\u60c5\u7eea\u5206\u7c7b\u6a21\u578b\uff0c\u5b83\u53ef\u4ee5\u5728\u4efb\u4f55\u5730\u65b9\u5de5\u4f5c\uff0c\u5e76\u4e14\u53ef\u4ee5\u4f7f\u7528API\u6765\u6267\u884c\u3002<\/p>\n
\u5177\u4f53\u6d41\u7a0b\u5982\u4e0b\uff1a<\/p>\n
\u4f60\u53ef\u4ee5\u5728\u4e0b\u9762\u7684GitHub\u4ee3\u7801\u5e93\u4e2d\u67e5\u770b\u6211\u7684\u5168\u90e8\u4ee3\u7801\uff1ahttps:\/\/github.com\/rfeers\/data-science-portfolio\/tree\/main\/end-to-end-projects\/simple-docker-hf-model\u3002<\/p>\n
\u539f\u6587\u6807\u9898\uff1aA Simple to Implement End-to-End Project with HuggingFace\uff0c\u4f5c\u8005\uff1aJosep Ferrer<\/p>\n
\u94fe\u63a5\uff1a<\/p>\n
https:\/\/www.kdnuggets.com\/a-simple-to-implement-end-to-end-project-with-huggingface\u3002<\/p>\n
\u60f3\u4e86\u89e3\u66f4\u591aAIGC\u7684\u5185\u5bb9\uff0c\u8bf7\u8bbf\u95ee\uff1a<\/p>\n
51CTO AI.x\u793e\u533a<\/p>\n
https:\/\/www.51cto.com\/aigc\/<\/p>\n