\u6625\u8282\u8d34\u6625\u8054，\u662f\u523b\u5728\u6211\u4eec\u6587\u5316\u57fa\u56e0\u91cc\u7684\u4eea\u5f0f\u611f\u3002\u4f46\u6bcf\u5e74\u60f3\u4e00\u526f\u65e2\u5e94\u666f\u53c8\u6709\u65b0\u610f\u7684\u5bf9\u8054，\u5bf9\u4e0d\u5c11\u4eba\u6765\u8bf4\u662f\u4e2a\u5c0f\u96be\u9898\u3002\u73b0\u5728，\u8fbe\u6469\u9662AliceMind\u56e2\u961f\u628a\u8fd9\u4e8b\u513f\u4ea4\u7ed9\u4e86AI\u3002<\/p>\n
\u4ed6\u4eec\u57fa\u4e8e\u5f3a\u5927\u7684PALM\u5927\u6a21\u578b，\u4e13\u95e8\u8bad\u7ec3\u4e86\u4e00\u4e2a\u201c\u6625\u8054\u751f\u6210\u6a21\u578b-\u4e2d\u6587-base\u201d\u3002\u4f60\u53ea\u9700\u8981\u8f93\u5165\u4e24\u4e2a\u5b57\u7684\u795d\u798f\u8bcd，\u6bd4\u5982\u201c\u4e94\u798f\u201d\u3001\u201c\u5b89\u5eb7\u201d\u3001\u201c\u5154\u5e74\u201d，\u5b83\u5c31\u80fd\u5e2e\u4f60\u751f\u6210\u4e00\u526f\u4e3b\u9898\u5951\u5408\u3001\u5bf9\u4ed7\u5de5\u6574\u7684\u6625\u8054\u3002\u8fd9\u542c\u8d77\u6765\u5f88\u9177，\u4f46\u4e00\u4e2a\u73b0\u5b9e\u95ee\u9898\u6446\u5728\u9762\u524d：\u8fd9\u79cd\u5927\u6a21\u578b，\u662f\u4e0d\u662f\u975e\u5f97\u6709\u6602\u8d35\u7684GPU\u670d\u52a1\u5668\u624d\u80fd\u8dd1\u8d77\u6765？<\/p>\n
\u672c\u6587\u5c06\u5e26\u4f60\u8fdb\u884c\u4e00\u6b21\u5f7b\u5e95\u7684\u201c\u538b\u529b\u6d4b\u8bd5\u201d：\u5728\u4e00\u53f0\u666e\u901a\u7684\u3001\u6ca1\u6709\u72ec\u7acb\u663e\u5361（GPU）\u7684\u4f4e\u914dCPU\u670d\u52a1\u5668\u4e0a，\u5b8c\u6574\u90e8\u7f72\u5e76\u8fd0\u884c\u8fd9\u4e2a\u8fbe\u6469\u9662PALM\u6625\u8054\u6a21\u578b\u3002\u6211\u4eec\u5c06\u9a8c\u8bc1\u5176\u63a8\u7406\u7684\u53ef\u884c\u6027\u3001\u901f\u5ea6\u4ee5\u53ca\u5b9e\u9645\u751f\u6210\u6548\u679c，\u4e3a\u4f60\u63d0\u4f9b\u4e00\u4e2a\u4f4e\u6210\u672c\u4f53\u9a8cAI\u521b\u4f5c\u6625\u8054\u7684\u5b9e\u6218\u6307\u5357\u3002<\/p>\n
\u5728\u5f00\u59cb\u4e4b\u524d，\u6211\u4eec\u5148\u660e\u786e\u8fd9\u6b21\u5b9e\u9a8c\u7684\u201c\u6218\u573a\u201d\u914d\u7f6e，\u8fd9\u6709\u52a9\u4e8e\u4f60\u5bf9\u6bd4\u81ea\u5df1\u7684\u73af\u5883：<\/p>\n
\u53ef\u4ee5\u770b\u5230，\u8fd9\u662f\u4e00\u53f0\u975e\u5e38\u666e\u901a\u751a\u81f3\u6709\u4e9b\u8fc7\u65f6\u7684\u670d\u52a1\u5668，\u5b8c\u5168\u6ca1\u6709GPU\u3002\u6211\u4eec\u7684\u76ee\u6807\u5c31\u662f\u5728\u8fd9\u53f0\u673a\u5668\u4e0a，\u8ba9AI\u6625\u8054\u6a21\u578b\u201c\u8dd1\u8d77\u6765\u201d\u3002<\/p>\n
\u9996\u5148，\u6211\u4eec\u901a\u8fc7SSH\u8fde\u63a5\u5230\u670d\u52a1\u5668，\u8fdb\u884c\u57fa\u7840\u68c0\u67e5\u3002<\/p>\n
# 1. \u68c0\u67e5\u7cfb\u7edf\u7248\u672c\u548c\u5185\u6838
\nlsb_release -a
\nuname -a<\/p>\n
# 2. \u68c0\u67e5Python\u7248\u672c（\u9700\u89813.8+，\u63a8\u83503.10）
\npython3 –version<\/p>\n
# 3. \u68c0\u67e5\u5185\u5b58\u548c\u4ea4\u6362\u7a7a\u95f4（Swap），\u4f4e\u5185\u5b58\u73af\u5883\u4e0bSwap\u5f88\u91cd\u8981
\nfree -h<\/p>\n
\u5982\u679c\u4f60\u7684Python\u7248\u672c\u4f4e\u4e8e3.8，\u9700\u8981\u5148\u5347\u7ea7\u3002\u5728Ubuntu\u4e0a\u53ef\u4ee5\u8fd9\u6837\u5b89\u88c5Python 3.10：<\/p>\n
sudo apt update
\nsudo apt install software-properties-common -y
\nsudo add-apt-repository ppa:deadsnakes\/ppa -y
\nsudo apt update
\nsudo apt install python3.10 python3.10-venv python3.10-dev -y<\/p>\n
# \u5c06python3\u6307\u5411\u65b0\u7248\u672c（\u53ef\u9009，\u8c28\u614e\u64cd\u4f5c）
\nsudo update-alternatives –install \/usr\/bin\/python3 python3 \/usr\/bin\/python3.10 1<\/p>\n
\u4e3a\u4e86\u907f\u514d\u5305\u51b2\u7a81，\u5f3a\u70c8\u5efa\u8bae\u4e3a\u8fd9\u4e2a\u9879\u76ee\u521b\u5efa\u72ec\u7acb\u7684\u865a\u62df\u73af\u5883\u3002<\/p>\n
# \u5728\u7528\u6237\u76ee\u5f55\u4e0b\u521b\u5efa\u9879\u76ee\u6587\u4ef6\u5939
\ncd ~
\nmkdir -p ai_spring_couplet
\ncd ai_spring_couplet<\/p>\n
# \u521b\u5efaPython\u865a\u62df\u73af\u5883
\npython3 -m venv venv<\/p>\n
# \u6fc0\u6d3b\u865a\u62df\u73af\u5883
\nsource venv\/bin\/activate<\/p>\n
# \u6fc0\u6d3b\u540e，\u547d\u4ee4\u884c\u63d0\u793a\u7b26\u524d\u4f1a\u51fa\u73b0 (venv) \u6807\u8bc6<\/p>\n
\u63a5\u4e0b\u6765\u662f\u6838\u5fc3\u73af\u8282：\u5728\u65e0GPU\u7684\u670d\u52a1\u5668\u4e0a\u90e8\u7f72\u6625\u8054\u6a21\u578b\u3002\u6574\u4e2a\u8fc7\u7a0b\u53ef\u4ee5\u5206\u4e3a\u6a21\u578b\u4e0b\u8f7d\u3001\u4f9d\u8d56\u5b89\u88c5\u3001\u670d\u52a1\u542f\u52a8\u4e09\u6b65\u3002<\/p>\n
\u6839\u636e\u63d0\u4f9b\u7684\u8d44\u6599，\u6a21\u578b\u9700\u8981\u9884\u7f6e\u5728\u7279\u5b9a\u8def\u5f84\u3002\u6211\u4eec\u5148\u521b\u5efa\u8fd9\u4e2a\u76ee\u5f55\u7ed3\u6784\u3002<\/p>\n
# \u521b\u5efa\u6a21\u578b\u5b58\u50a8\u76ee\u5f55
\nsudo mkdir -p \/root\/ai-models\/iic\/spring_couplet_generation
\n# \u5982\u679c\u65e0\u6cd5\u4f7f\u7528sudo，\u6216\u60f3\u5728\u7528\u6237\u76ee\u5f55\u4e0b\u64cd\u4f5c，\u4e5f\u53ef\u4ee5\u8c03\u6574\u8def\u5f84，\u540e\u7eed\u9700\u8981\u4fee\u6539\u4ee3\u7801\u4e2d\u7684\u8def\u5f84\u6307\u5411
\n# \u8fd9\u91cc\u6211\u4eec\u4ee5\u7528\u6237\u76ee\u5f55\u4e3a\u4f8b，\u66f4\u901a\u7528
\nmkdir -p ~\/ai-models\/spring_couplet<\/p>\n
# \u5047\u8bbe\u4f60\u5df2\u7ecf\u901a\u8fc7\u5176\u4ed6\u65b9\u5f0f（\u5982scp\u3001wget\u4eceModelScope\u4e0b\u8f7d）\u83b7\u5f97\u4e86\u6a21\u578b\u6587\u4ef6
\n# \u6a21\u578b\u901a\u5e38\u5305\u542b\u591a\u4e2a\u6587\u4ef6，\u5982config.json, pytorch_model.bin, vocab.txt\u7b49
\n# \u8fd9\u91cc\u6f14\u793a\u5982\u4f55\u4eceModelScope\u5e93\u4e0b\u8f7d（\u9700\u5b89\u88c5modelscope\u5e93）
\npip install modelscope<\/p>\n
# \u5728Python\u4ea4\u4e92\u73af\u5883\u4e2d\u4e0b\u8f7d\u6a21\u578b
\npython3 -c "
\nfrom modelscope import snapshot_download
\nmodel_dir = snapshot_download('damo\/nlp_palm2.0_text-generation_spring-couplet')
\nprint(f'\u6a21\u578b\u5df2\u4e0b\u8f7d\u81f3: {model_dir}')
\n"
\n# \u4e0b\u8f7d\u5b8c\u6210\u540e，\u5c06\u6a21\u578b\u6587\u4ef6\u590d\u5236\u5230\u6211\u4eec\u521b\u5efa\u7684\u76ee\u5f55
\n# \u5047\u8bbe\u4e0b\u8f7d\u76ee\u5f55\u662f \/root\/.cache\/modelscope\/hub\/damo\/nlp_palm2.0_text-generation_spring-couplet
\n# cp -r \/root\/.cache\/modelscope\/hub\/damo\/nlp_palm2.0_text-generation_spring-couplet\/* ~\/ai-models\/spring_couplet\/<\/p>\n
\u5173\u952e\u70b9：\u5bf9\u4e8eCPU\u73af\u5883，\u786e\u4fdd\u4e0b\u8f7d\u7684\u6a21\u578b\u662fPyTorch\u683c\u5f0f（.bin\u6216.pth），\u5e76\u4e14\u517c\u5bb9CPU\u63a8\u7406\u3002PALM\u6a21\u578b\u901a\u5e38\u9ed8\u8ba4\u652f\u6301\u3002<\/p>\n
\u5728\u9879\u76ee\u6839\u76ee\u5f55（~\/ai_spring_couplet）\u4e0b，\u521b\u5efarequirements.txt\u6587\u4ef6，\u5185\u5bb9\u5982\u4e0b：<\/p>\n
torch>=1.12.0
\ntransformers>=4.25.0
\nmodelscope>=1.4.0
\ngradio>=3.36.0
\nsentencepiece
\nprotobuf<\/p>\n
\u7136\u540e\u5b89\u88c5：<\/p>\n
# \u786e\u4fdd\u5728\u865a\u62df\u73af\u5883\u4e2d
\npip install -r requirements.txt -i https:\/\/pypi.tuna.tsinghua.edu.cn\/simple<\/p>\n
\u7279\u522b\u6ce8\u610f：\u5b89\u88c5PyTorch\u65f6，\u52a1\u5fc5\u9009\u62e9CPU\u7248\u672c\u3002\u5982\u679c\u4f60\u4e0a\u9762\u7684\u547d\u4ee4\u5b89\u88c5\u7684\u662fGPU\u7248，\u8bf7\u4f7f\u7528\u4ee5\u4e0b\u547d\u4ee4\u5b89\u88c5\u7eafCPU\u7248，\u8fd9\u80fd\u663e\u8457\u51cf\u5c11\u4f9d\u8d56\u4f53\u79ef\u548c\u6f5c\u5728\u51b2\u7a81：<\/p>\n
pip install torch torchvision torchaudio –index-url https:\/\/download.pytorch.org\/whl\/cpu<\/p>\n
\u5728\u9879\u76ee\u6839\u76ee\u5f55\u521b\u5efaapp.py，\u8fd9\u662f\u57fa\u4e8eGradio\u7684Web\u5e94\u7528\u4e3b\u7a0b\u5e8f\u3002\u6211\u4eec\u9700\u8981\u6839\u636eCPU\u73af\u5883\u8fdb\u884c\u4e00\u4e9b\u4f18\u5316\u3002<\/p>\n
# app.py
\nimport gradio as gr
\nfrom transformers import AutoTokenizer, AutoModelForCausalLM
\nimport torch
\nimport os
\nimport time<\/p>\n
# \u8bbe\u7f6e\u8bbe\u5907\u4e3aCPU
\ndevice = torch.device("cpu")
\nprint(f"Using device: {device}")<\/p>\n
# 1. \u52a0\u8f7d\u6a21\u578b\u548c\u5206\u8bcd\u5668
\nmodel_dir = os.path.expanduser("~\/ai-models\/spring_couplet") # \u4fee\u6539\u4e3a\u4f60\u7684\u6a21\u578b\u8def\u5f84
\nprint(f"Loading model from: {model_dir}")<\/p>\n
try:
\n tokenizer = AutoTokenizer.from_pretrained(model_dir, trust_remote_code=True)
\n # \u5173\u952e：\u52a0\u8f7d\u6a21\u578b\u65f6\u6307\u5b9adevice_map='cpu'，\u5e76\u8bbe\u7f6elow_cpu_mem_usage\u6709\u52a9\u4e8e\u51cf\u5c11\u5185\u5b58\u5cf0\u503c
\n model = AutoModelForCausalLM.from_pretrained(
\n model_dir,
\n trust_remote_code=True,
\n device_map='cpu',
\n low_cpu_mem_usage=True,
\n torch_dtype=torch.float32 # CPU\u4e0a\u4f7f\u7528float32
\n )
\n model.to(device)
\n model.eval() # \u8bbe\u7f6e\u4e3a\u8bc4\u4f30\u6a21\u5f0f
\n print("Model and tokenizer loaded successfully!")
\nexcept Exception as e:
\n print(f"Error loading model: {e}")
\n raise<\/p>\n
# 2. \u6625\u8054\u751f\u6210\u51fd\u6570
\ndef generate_couplet(keyword):
\n """
\n \u6839\u636e\u4e24\u4e2a\u5b57\u7684\u795d\u798f\u8bcd\u751f\u6210\u6625\u8054
\n """
\n if not keyword or len(keyword.strip()) != 2:
\n return "\u8bf7\u8f93\u5165\u4e24\u4e2a\u5b57\u7684\u795d\u798f\u8bcd，\u4f8b\u5982：\u4e94\u798f\u3001\u5e78\u798f\u3001\u5154\u5e74\u3002", ""<\/p>\n
start_time = time.time()<\/p>\n
# \u6784\u5efa\u8f93\u5165\u6587\u672c，\u6a21\u578b\u8bad\u7ec3\u65f6\u7684\u7279\u5b9a\u683c\u5f0f
\n input_text = f"{keyword}："<\/p>\n
try:
\n # \u7f16\u7801\u8f93\u5165
\n inputs = tokenizer(input_text, return_tensors="pt")
\n inputs = {k: v.to(device) for k, v in inputs.items()}<\/p>\n
# \u751f\u6210\u53c2\u6570\u8bbe\u7f6e（\u9488\u5bf9CPU\u4f18\u5316，\u51cf\u5c11\u751f\u6210\u957f\u5ea6\u4ee5\u52a0\u5feb\u901f\u5ea6）
\n with torch.no_grad(): # \u7981\u7528\u68af\u5ea6\u8ba1\u7b97，\u8282\u7701\u5185\u5b58
\n outputs = model.generate(
\n **inputs,
\n max_new_tokens=64, # \u6700\u5927\u751f\u6210\u957f\u5ea6，\u63a7\u5236\u5bf9\u8054\u5b57\u6570
\n do_sample=True, # \u4f7f\u7528\u91c7\u6837，\u4f7f\u8f93\u51fa\u66f4\u591a\u6837
\n top_p=0.9, # \u6838\u91c7\u6837\u53c2\u6570
\n temperature=0.8, # \u6e29\u5ea6\u53c2\u6570，\u63a7\u5236\u968f\u673a\u6027
\n repetition_penalty=1.2, # \u91cd\u590d\u60e9\u7f5a，\u907f\u514d\u91cd\u590d\u7528\u8bcd
\n pad_token_id=tokenizer.eos_token_id,
\n num_return_sequences=1 # \u53ea\u751f\u6210\u4e00\u4e2a\u7ed3\u679c
\n )<\/p>\n
# \u89e3\u7801\u8f93\u51fa
\n generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)<\/p>\n
# \u540e\u5904\u7406：\u4ece\u751f\u6210\u6587\u672c\u4e2d\u63d0\u53d6\u4e0a\u4e0b\u8054
\n # \u6a21\u578b\u8f93\u51fa\u683c\u5f0f\u901a\u5e38\u662f\u201c\u5173\u952e\u8bcd：\u4e0a\u8054，\u4e0b\u8054\u201d
\n if "：" in generated_text:
\n _, couplet_part = generated_text.split("：", 1)
\n # \u6e05\u7406\u53ef\u80fd\u7684\u989d\u5916\u6587\u672c
\n couplet_part = couplet_part.strip().replace("\\\\n", " ")
\n # \u5c1d\u8bd5\u6309\u9017\u53f7\u5206\u5272\u4e0a\u4e0b\u8054
\n if "，" in couplet_part:
\n parts = couplet_part.split("，", 1)
\n upper_line = parts[0].strip()
\n lower_line = parts[1].strip() if len(parts) > 1 else ""
\n else:
\n # \u5982\u679c\u6ca1\u6709\u660e\u663e\u5206\u9694\u7b26，\u5c1d\u8bd5\u5176\u4ed6\u65b9\u5f0f\u6216\u8fd4\u56de\u6574\u4f53
\n upper_line = couplet_part
\n lower_line = ""
\n else:
\n upper_line = generated_text
\n lower_line = ""<\/p>\n
inference_time = time.time() – start_time<\/p>\n
# \u8fd4\u56de\u7ed3\u679c
\n result_text = f"\u4e0a\u8054：{upper_line}\\\\n\u4e0b\u8054：{lower_line}\\\\n\u6a2a\u6279：{keyword}"
\n info_text = f"\u751f\u6210\u8017\u65f6：{inference_time:.2f}\u79d2 | \u8bbe\u5907：CPU"<\/p>\n
return result_text, info_text<\/p>\n
except Exception as e:
\n return f"\u751f\u6210\u8fc7\u7a0b\u4e2d\u51fa\u73b0\u9519\u8bef：{str(e)}", ""<\/p>\n
# 3. \u521b\u5efaGradio\u754c\u9762
\ndef create_interface():
\n with gr.Blocks(title="AI\u6625\u8054\u751f\u6210\u5668 (CPU\u7248)", theme=gr.themes.Soft()) as demo:
\n gr.Markdown("""
\n # 🏮 AI \u6625\u8054\u751f\u6210\u5668 (CPU\u90e8\u7f72\u7248)
\n \u57fa\u4e8e\u8fbe\u6469\u9662PALM\u5927\u6a21\u578b\u7684\u667a\u80fd\u6625\u8054\u751f\u6210\u7cfb\u7edf\u3002\u672c\u7248\u672c\u4e13\u4e3a**\u65e0GPU\u7684CPU\u670d\u52a1\u5668**\u4f18\u5316\u3002
\n """)<\/p>\n
with gr.Row():
\n with gr.Column(scale=1):
\n keyword_input = gr.Textbox(
\n label="\u8f93\u5165\u4e24\u4e2a\u5b57\u7684\u795d\u798f\u8bcd",
\n placeholder="\u4f8b\u5982：\u4e94\u798f\u3001\u5e78\u798f\u3001\u5154\u5e74\u3001\u5409\u7965…",
\n max_lines=1,
\n value="\u65b0\u6625"
\n )
\n submit_btn = gr.Button("\u751f\u6210\u6625\u8054", variant="primary")
\n gr.Markdown("**\u5c0f\u63d0\u793a**：\u8f93\u5165\u50cf`\u5b89\u5eb7`\u3001`\u56e2\u5706`\u3001`\u5bcc\u8d35`\u8fd9\u7c7b\u5bd3\u610f\u7f8e\u597d\u7684\u4e24\u5b57\u8bcd，\u6548\u679c\u66f4\u4f73\u3002")<\/p>\n
with gr.Column(scale=2):
\n output_text = gr.Textbox(
\n label="\u751f\u6210\u7684\u6625\u8054",
\n lines=5,
\n interactive=False
\n )
\n info_text = gr.Textbox(
\n label="\u751f\u6210\u4fe1\u606f",
\n lines=1,
\n interactive=False
\n )
\n copy_btn = gr.Button("\u590d\u5236\u6625\u8054")<\/p>\n
# \u793a\u4f8b\u8bcd\u6309\u94ae
\n gr.Markdown("### \u8bd5\u8bd5\u8fd9\u4e9b\u795d\u798f\u8bcd：")
\n example_words = ["\u4e94\u798f", "\u5e78\u798f", "\u5154\u5e74", "\u5409\u7965", "\u5b89\u5eb7", "\u56e2\u5706", "\u5bcc\u8d35", "\u5982\u610f"]
\n example_row = gr.Row()
\n with example_row:
\n for word in example_words:
\n gr.Button(word, size="sm").click(
\n fn=lambda w=word: w,
\n inputs=[],
\n outputs=[keyword_input]
\n )<\/p>\n
# \u7ed1\u5b9a\u4e8b\u4ef6
\n submit_btn.click(
\n fn=generate_couplet,
\n inputs=[keyword_input],
\n outputs=[output_text, info_text]
\n )<\/p>\n
# \u590d\u5236\u6309\u94ae\u529f\u80fd（\u901a\u8fc7JavaScript\u5b9e\u73b0）
\n copy_btn.click(
\n fn=None,
\n inputs=[output_text],
\n js="(text) => {navigator.clipboard.writeText(text); return '\u5df2\u590d\u5236\u5230\u526a\u8d34\u677f！';}"
\n )<\/p>\n
# \u9875\u9762\u52a0\u8f7d\u65f6\u751f\u6210\u4e00\u4e2a\u793a\u4f8b
\n demo.load(
\n fn=lambda: generate_couplet("\u65b0\u6625"),
\n inputs=[],
\n outputs=[output_text, info_text]
\n )<\/p>\n
return demo<\/p>\n
if __name__ == "__main__":
\n # \u521b\u5efa\u754c\u9762\u5e76\u542f\u52a8，\u7ed1\u5b9a\u5230\u6240\u6709\u7f51\u7edc\u63a5\u53e3，\u65b9\u4fbf\u8fdc\u7a0b\u8bbf\u95ee
\n demo = create_interface()
\n demo.launch(
\n server_name="0.0.0.0", # \u5141\u8bb8\u8fdc\u7a0b\u8bbf\u95ee
\n server_port=7860,
\n share=False, # CPU\u670d\u52a1\u5668\u6027\u80fd\u6709\u9650，\u4e0d\u5efa\u8bae\u751f\u6210\u516c\u7f51\u94fe\u63a5
\n debug=False # \u751f\u4ea7\u73af\u5883\u5173\u95eddebug\u6a21\u5f0f
\n )<\/p>\n
\u4e3a\u4e86\u65b9\u4fbf，\u521b\u5efa\u4e00\u4e2a\u542f\u52a8\u811a\u672cstart.sh：<\/p>\n
#!\/bin\/bash
\n# start.sh<\/p>\n
# \u6fc0\u6d3b\u865a\u62df\u73af\u5883
\nsource ~\/ai_spring_couplet\/venv\/bin\/activate<\/p>\n
# \u68c0\u67e5\u662f\u5426\u5728\u865a\u62df\u73af\u5883\u4e2d
\nif [ -z "$VIRTUAL_ENV" ]; then
\n echo "\u865a\u62df\u73af\u5883\u6fc0\u6d3b\u5931\u8d25，\u8bf7\u68c0\u67e5\u8def\u5f84\u3002"
\n exit 1
\nfi<\/p>\n
echo "\u6b63\u5728\u542f\u52a8AI\u6625\u8054\u751f\u6210\u670d\u52a1（CPU\u6a21\u5f0f）…"
\necho "\u670d\u52a1\u5c06\u5728 http:\/\/<\u4f60\u7684\u670d\u52a1\u5668IP>:7860 \u53ef\u7528"
\necho "\u6309 Ctrl+C \u505c\u6b62\u670d\u52a1"<\/p>\n
# \u542f\u52a8\u5e94\u7528
\npython3 ~\/ai_spring_couplet\/app.py<\/p>\n
\u7ed9\u811a\u672c\u6267\u884c\u6743\u9650\u5e76\u8fd0\u884c：<\/p>\n
chmod +x start.sh
\n.\/start.sh<\/p>\n
\u670d\u52a1\u542f\u52a8\u540e，\u6253\u5f00\u6d4f\u89c8\u5668\u8bbf\u95ee http:\/\/\u4f60\u7684\u670d\u52a1\u5668IP:7860，\u5c31\u80fd\u770b\u5230Web\u754c\u9762\u4e86\u3002\u4e0b\u9762\u662f\u6211\u4eec\u5728\u8fd9\u53f0\u4f4e\u914dCPU\u670d\u52a1\u5668\u4e0a\u7684\u5b9e\u6d4b\u7ed3\u679c\u3002<\/p>\n
\u6211\u4eec\u6d4b\u8bd5\u4e86\u591a\u4e2a\u795d\u798f\u8bcd，\u6a21\u578b\u5747\u80fd\u751f\u6210\u7b26\u5408\u4e3b\u9898\u3001\u5bf9\u4ed7\u57fa\u672c\u5de5\u6574\u7684\u6625\u8054：<\/p>\n
| \u65b0\u6625<\/td>\n | \u6625\u56de\u5927\u5730\u98ce\u5149\u597d<\/td>\n | \u798f\u6ee1\u4eba\u95f4\u559c\u4e8b\u591a<\/td>\n | \u5bf9\u4ed7\u5de5\u6574，\u5e74\u5473\u8db3<\/td>\n<\/tr>\n | ||||||||||||||||||
| \u4e94\u798f<\/td>\n | \u798f\u4f34\u5409\u7965\u4e07\u4e8b\u5174<\/td>\n | \u559c\u968f\u5982\u610f\u5343\u5bb6\u4e50<\/td>\n | \u7d27\u6263\u201c\u798f\u201d\u4e3b\u9898<\/td>\n<\/tr>\n | ||||||||||||||||||
| \u5b89\u5eb7<\/td>\n | \u5b89\u5c45\u4e50\u4e1a\u5bb6\u5bb6\u798f<\/td>\n | \u5eb7\u4e50\u5e74\u534e\u5904\u5904\u6625<\/td>\n | \u201c\u5b89\u5eb7\u201d\u4e8c\u5b57\u62c6\u5206\u5de7\u5999<\/td>\n<\/tr>\n | ||||||||||||||||||
| \u56e2\u5706<\/td>\n | \u56e2\u5706\u7f8e\u6ee1\u5e78\u798f\u5bb6<\/td>\n | \u548c\u7766\u6e29\u99a8\u5feb\u4e50\u5e74<\/td>\n | \u6e29\u99a8\u559c\u5e86<\/td>\n<\/tr>\n | ||||||||||||||||||
| \u5154\u5e74<\/td>\n | \u5154\u8dc3\u5343\u5c71\u4f20\u559c\u8baf<\/td>\n | \u6885\u5f00\u4e07\u91cc\u62a5\u65b0\u6625<\/td>\n | \u7ed3\u5408\u751f\u8096\u4e0e\u6625\u8282<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n \u4ece\u751f\u6210\u6548\u679c\u770b，\u6a21\u578b\u786e\u5b9e\u7406\u89e3\u4e86\u4e24\u5b57\u795d\u798f\u8bcd\u7684\u542b\u4e49，\u5e76\u80fd\u56f4\u7ed5\u5176\u6269\u5c55\u51fa\u610f\u5883\u76f8\u7b26\u7684\u4e0a\u4e0b\u8054\u3002\u867d\u7136\u4e2a\u522b\u5bf9\u8054\u5728\u5e73\u4ec4\u6216\u6587\u5b66\u6027\u4e0a\u53ef\u80fd\u4e0d\u53ca\u4e13\u4e1a\u521b\u4f5c，\u4f46\u5bf9\u4e8e\u65e5\u5e38\u4f7f\u7528\u3001\u589e\u6dfb\u8282\u65e5\u6c14\u6c1b\u6765\u8bf4，\u5b8c\u5168\u5408\u683c\u4e14\u9887\u5177\u65b0\u610f\u3002<\/p>\n 4.2 \u6027\u80fd\u6d4b\u8bd5：CPU\u63a8\u7406\u901f\u5ea6\u4e0e\u8d44\u6e90\u5360\u7528<\/h4>\n\u8fd9\u662f\u672c\u6b21\u9a8c\u8bc1\u7684\u6838\u5fc3\u3002\u6211\u4eec\u5728\u670d\u52a1\u5668\u4e0a\u8fdb\u884c\u4e86\u591a\u8f6e\u6d4b\u8bd5：<\/p>\n \u5355\u6b21\u751f\u6210\u8017\u65f6：<\/p>\n
|