{"id":34965,"date":"2025-08-06T12:38:10","date_gmt":"2025-08-06T04:38:10","guid":{"rendered":"https:\/\/www.kdjingpai.com\/?p=34965"},"modified":"2025-08-06T12:38:10","modified_gmt":"2025-08-06T04:38:10","slug":"openai-gpt-oss-recipes","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/en\/openai-gpt-oss-recipes\/","title":{"rendered":"\u5fae\u8c03OpenAI GPT OSS\u6a21\u578b\u7684\u811a\u672c\u548c\u6559\u7a0b\u96c6\u5408"},"content":{"rendered":"<p><code>gpt-oss-recipes<\/code>\u00a0\u662f\u7531 Hugging Face \u7ef4\u62a4\u7684\u4e00\u4e2a GitHub \u4ed3\u5e93\uff0c\u4e13\u6ce8\u4e8e\u63d0\u4f9b\u4f7f\u7528 OpenAI GPT OSS \u6a21\u578b\u7684\u811a\u672c\u548c Jupyter Notebook \u6559\u7a0b\u3002\u4ed3\u5e93\u5305\u542b\u4e86\u9488\u5bf9 OpenAI \u6700\u65b0\u5f00\u6e90\u6a21\u578b\u00a0<code>gpt-oss-120b<\/code>\u00a0\u548c\u00a0<code>gpt-oss-20b<\/code> \u7684\u914d\u7f6e\u548c\u4f7f\u7528\u793a\u4f8b\u3002\u8fd9\u4e9b\u6a21\u578b\u4ee5\u5f3a\u5927\u7684\u63a8\u7406\u80fd\u529b\u548c\u9ad8\u6548\u7684\u8d44\u6e90\u5360\u7528\u8457\u79f0\uff0c\u9002\u5408\u5f00\u53d1\u8005\u5728\u751f\u4ea7\u73af\u5883\u6216\u4e2a\u4eba\u8bbe\u5907\u4e0a\u8fd0\u884c\u3002\u4ed3\u5e93\u4e2d\u7684\u4ee3\u7801\u548c\u6587\u6863\u5e2e\u52a9\u7528\u6237\u5feb\u901f\u4e0a\u624b\u6a21\u578b\u63a8\u7406\u3001\u5fae\u8c03\u548c\u90e8\u7f72\uff0c\u8986\u76d6\u4ece\u73af\u5883\u8bbe\u7f6e\u5230\u590d\u6742\u4efb\u52a1\u7684\u5b9e\u73b0\u3002\u6240\u6709\u5185\u5bb9\u57fa\u4e8e Apache 2.0 \u8bb8\u53ef\u8bc1\uff0c\u5141\u8bb8\u81ea\u7531\u4f7f\u7528\u548c\u4fee\u6539\u3002<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li>\u63d0\u4f9b\u00a0<code>gpt-oss-120b<\/code>\u00a0\u548c\u00a0<code>gpt-oss-20b<\/code>\u00a0\u6a21\u578b\u7684\u914d\u7f6e\u811a\u672c\uff0c\u652f\u6301\u5feb\u901f\u5207\u6362\u6a21\u578b\u5927\u5c0f\u3002<\/li>\n<li>\u5305\u542b\u73af\u5883\u8bbe\u7f6e\u4ee3\u7801\uff0c\u652f\u6301 Python \u865a\u62df\u73af\u5883\u548c\u4f9d\u8d56\u5b89\u88c5\u3002<\/li>\n<li>\u63d0\u4f9b\u63a8\u7406\u793a\u4f8b\uff0c\u5c55\u793a\u5982\u4f55\u4f7f\u7528\u6a21\u578b\u751f\u6210\u6587\u672c\u6216\u6267\u884c\u5de5\u5177\u8c03\u7528\u3002<\/li>\n<li>\u652f\u6301\u6a21\u578b\u5fae\u8c03\uff0c\u5305\u542b\u591a\u8bed\u8a00\u63a8\u7406\u6570\u636e\u96c6\u7684\u5904\u7406\u793a\u4f8b\u3002<\/li>\n<li>\u63d0\u4f9b\u4e0e Transformers\u3001vLLM \u548c <a href=\"https:\/\/www.kdjingpai.com\/de\/ollama\/\">Ollama<\/a> \u7b49\u6846\u67b6\u7684\u96c6\u6210\u6559\u7a0b\u3002<\/li>\n<li>\u652f\u6301\u5728\u4e0d\u540c\u786c\u4ef6\uff08H100 GPU\u3001\u6d88\u8d39\u7ea7\u8bbe\u5907\uff09\u4e0a\u8fd0\u884c\u6a21\u578b\u7684\u4f18\u5316\u914d\u7f6e\u3002<\/li>\n<\/ul>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>\u5b89\u88c5\u6d41\u7a0b<\/h3>\n<p>\u8981\u4f7f\u7528\u00a0<code>gpt-oss-recipes<\/code>\u00a0\u4ed3\u5e93\u4e2d\u7684\u811a\u672c\uff0c\u9996\u5148\u9700\u8981\u514b\u9686\u4ed3\u5e93\u5e76\u8bbe\u7f6e Python \u73af\u5883\u3002\u4ee5\u4e0b\u662f\u8be6\u7ec6\u6b65\u9aa4\uff1a<\/p>\n<ol>\n<li><strong>\u514b\u9686\u4ed3\u5e93<\/strong><br \/>\n\u6253\u5f00\u7ec8\u7aef\uff0c\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u514b\u9686\u4ed3\u5e93\u5230\u672c\u5730\uff1a<\/p>\n<pre><code>git clone https:\/\/github.com\/huggingface\/gpt-oss-recipes.git\r\ncd gpt-oss-recipes\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u521b\u5efa\u865a\u62df\u73af\u5883<\/strong><br \/>\n\u5efa\u8bae\u4f7f\u7528 Python 3.11 \u521b\u5efa\u865a\u62df\u73af\u5883\u4ee5\u786e\u4fdd\u517c\u5bb9\u6027\u3002\u63a8\u8350\u4f7f\u7528\u00a0<code>uv<\/code>\u00a0\u5de5\u5177\uff1a<\/p>\n<pre><code>uv venv <a href=\"https:\/\/www.kdjingpai.com\/de\/gpt-oss\/\">gpt-oss<\/a> --python 3.11\r\nsource gpt-oss\/bin\/activate\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u5b89\u88c5\u4f9d\u8d56<\/strong><br \/>\n\u5b89\u88c5\u5fc5\u8981\u7684 Python \u5305\uff0c\u5305\u62ec PyTorch \u548c Transformers\u3002\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\uff1a<\/p>\n<pre><code>uv pip install --upgrade pip\r\nuv pip install torch==2.8.0 torchvision torchaudio --index-url https:\/\/download.pytorch.org\/whl\/test\/cu128\r\nuv pip install -U transformers accelerate\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u5b89\u88c5 Triton \u5185\u6838\uff08\u53ef\u9009\uff09<\/strong><br \/>\n\u5982\u679c\u786c\u4ef6\u652f\u6301 MXFP4 \u91cf\u5316\uff08\u5982 H100 \u6216 RTX 50xx\uff09\uff0c\u53ef\u5b89\u88c5 Triton \u5185\u6838\u4ee5\u4f18\u5316\u6027\u80fd\uff1a<\/p>\n<pre><code>uv pip install git+https:\/\/github.com\/triton-lang\/triton.git@main#subdirectory=python\/triton_kernels\r\n<\/code><\/pre>\n<\/li>\n<\/ol>\n<h3>\u914d\u7f6e\u6a21\u578b<\/h3>\n<p>\u4ed3\u5e93\u63d0\u4f9b\u4e24\u4e2a\u6a21\u578b\uff1a<code>gpt-oss-120b<\/code>\uff08117B \u53c2\u6570\uff0c\u9002\u5408\u9ad8\u6027\u80fd GPU\uff09\u548c\u00a0<code>gpt-oss-20b<\/code>\uff0821B \u53c2\u6570\uff0c\u9002\u5408\u6d88\u8d39\u7ea7\u786c\u4ef6\uff09\u3002\u5728\u811a\u672c\u4e2d\u4fee\u6539\u00a0<code>model_path<\/code>\u00a0\u53d8\u91cf\u9009\u62e9\u6a21\u578b\u3002\u4f8b\u5982\uff1a<\/p>\n<pre><code>model_path = \"openai\/gpt-oss-20b\"  # \u9009\u62e9 20B \u6a21\u578b\r\n# model_path = \"openai\/gpt-oss-120b\"  # \u9009\u62e9 120B \u6a21\u578b\r\n<\/code><\/pre>\n<p>\u811a\u672c\u4f1a\u6839\u636e\u6a21\u578b\u5927\u5c0f\u81ea\u52a8\u914d\u7f6e\u8bbe\u5907\u6620\u5c04\u548c\u4f18\u5316\u8bbe\u7f6e\u3002<\/p>\n<h3>\u8fd0\u884c\u63a8\u7406<\/h3>\n<p>\u4ed3\u5e93\u5305\u542b\u7b80\u5355\u7684\u63a8\u7406\u793a\u4f8b\uff0c\u7528\u4e8e\u751f\u6210\u6587\u672c\u6216\u6267\u884c\u7279\u5b9a\u4efb\u52a1\u3002\u4ee5\u4e0b\u662f\u4e00\u4e2a\u4f7f\u7528\u00a0<code>gpt-oss-20b<\/code>\u00a0\u6a21\u578b\u751f\u6210\u6587\u672c\u7684\u793a\u4f8b\uff1a<\/p>\n<ol>\n<li>\u6253\u5f00\u00a0<code>inference.py<\/code>\u00a0\u6587\u4ef6\uff08\u6216\u7c7b\u4f3c\u811a\u672c\uff09\u3002<\/li>\n<li>\u786e\u4fdd\u5df2\u52a0\u8f7d\u6a21\u578b\u548c\u5206\u8bcd\u5668\uff1a\n<pre><code>from transformers import AutoModelForCausalLM, AutoTokenizer\r\nmodel_id = \"openai\/gpt-oss-20b\"\r\ntokenizer = AutoTokenizer.from_pretrained(model_id)\r\nmodel = AutoModelForCausalLM.from_pretrained(model_id, device_map=\"auto\", torch_dtype=\"auto\")\r\n<\/code><\/pre>\n<\/li>\n<li>\u8f93\u5165\u63d0\u793a\u5e76\u751f\u6210\u7ed3\u679c\uff1a\n<pre><code>messages = [{\"role\": \"user\", \"content\": \"\u5982\u4f55\u7528 Python \u5199\u4e00\u4e2a\u6392\u5e8f\u7b97\u6cd5\uff1f\"}]\r\ninputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors=\"pt\").to(model.device)\r\ngenerated = model.generate(**inputs, max_new_tokens=200)\r\nprint(tokenizer.decode(generated[0]))\r\n<\/code><\/pre>\n<\/li>\n<li>\u8fd0\u884c\u811a\u672c\uff0c\u6a21\u578b\u4f1a\u8fd4\u56de\u6392\u5e8f\u7b97\u6cd5\u7684 Python \u4ee3\u7801\u793a\u4f8b\u3002<\/li>\n<\/ol>\n<h3>\u8c03\u6574\u63a8\u7406\u53c2\u6570<\/h3>\n<p>\u53ef\u4ee5\u901a\u8fc7\u7cfb\u7edf\u63d0\u793a\u8c03\u6574\u63a8\u7406\u7684\u8be6\u7ec6\u7a0b\u5ea6\u3002\u4f8b\u5982\uff0c\u8bbe\u7f6e\u9ad8\u63a8\u7406\u7ea7\u522b\uff1a<\/p>\n<pre><code>messages = [\r\n{\"role\": \"system\", \"content\": \"Reasoning: high\"},\r\n{\"role\": \"user\", \"content\": \"\u89e3\u91ca\u91cf\u5b50\u8ba1\u7b97\u7684\u57fa\u672c\u539f\u7406\"}\r\n]\r\n<\/code><\/pre>\n<p>\u9ad8\u63a8\u7406\u7ea7\u522b\u4f1a\u751f\u6210\u66f4\u8be6\u7ec6\u7684\u63a8\u7406\u8fc7\u7a0b\uff0c\u9002\u5408\u590d\u6742\u95ee\u9898\u3002<\/p>\n<h3>\u5fae\u8c03\u6a21\u578b<\/h3>\n<p>\u4ed3\u5e93\u63d0\u4f9b\u5fae\u8c03\u793a\u4f8b\uff0c\u57fa\u4e8e Hugging Face \u7684 TRL \u5e93\u548c LoRA \u6280\u672f\u3002\u4ee5\u4e0b\u662f\u5fae\u8c03\u00a0<code>gpt-oss-20b<\/code>\u00a0\u7684\u6b65\u9aa4\uff1a<\/p>\n<ol>\n<li>\u4e0b\u8f7d\u591a\u8bed\u8a00\u63a8\u7406\u6570\u636e\u96c6\uff1a\n<pre><code>from datasets import load_dataset\r\ndataset = load_dataset(\"HuggingFaceH4\/Multilingual-Thinking\", split=\"train\")\r\n<\/code><\/pre>\n<\/li>\n<li>\u914d\u7f6e LoRA \u53c2\u6570\u5e76\u52a0\u8f7d\u6a21\u578b\uff1a\n<pre><code>from transformers import AutoModelForCausalLM\r\nfrom peft import PeftModel, LoraConfig\r\nmodel_name = \"openai\/gpt-oss-20b\"\r\nlora_config = LoraConfig(r=8, lora_alpha=32, target_modules=[\"q_proj\", \"v_proj\"])\r\nmodel = AutoModelForCausalLM.from_pretrained(model_name, device_map=\"auto\")\r\nmodel = PeftModel(model, lora_config)\r\n<\/code><\/pre>\n<\/li>\n<li>\u4f7f\u7528 TRL \u5e93\u8fdb\u884c\u5fae\u8c03\uff08\u53c2\u8003\u4ed3\u5e93\u4e2d\u7684\u00a0<code>finetune.ipynb<\/code>\uff09\u3002<\/li>\n<li>\u4fdd\u5b58\u5fae\u8c03\u540e\u7684\u6a21\u578b\uff0c\u7528\u4e8e\u7279\u5b9a\u4efb\u52a1\u5982\u591a\u8bed\u8a00\u63a8\u7406\u3002<\/li>\n<\/ol>\n<h3>\u4f7f\u7528 vLLM \u6216 Ollama<\/h3>\n<p>\u5982\u679c\u9700\u8981\u5feb\u901f\u90e8\u7f72\uff0c\u4ed3\u5e93\u652f\u6301 <a href=\"https:\/\/www.kdjingpai.com\/de\/vllm\/\">vLLM<\/a> \u548c Ollama\uff1a<\/p>\n<ul>\n<li><strong>vLLM<\/strong>\uff1a\u542f\u52a8 OpenAI \u517c\u5bb9\u7684\u670d\u52a1\u5668\uff1a\n<pre><code>uv pip install --pre vllm==0.10.1+gptoss --extra-index-url https:\/\/wheels.vllm.ai\/gpt-oss\/\r\nvllm serve openai\/gpt-oss-20b\r\n<\/code><\/pre>\n<\/li>\n<li><strong>Ollama<\/strong>\uff1a\u5728\u6d88\u8d39\u7ea7\u786c\u4ef6\u4e0a\u8fd0\u884c\uff1a\n<pre><code>ollama pull gpt-oss:20b\r\nollama run gpt-oss:20b\r\n<\/code><\/pre>\n<\/li>\n<\/ul>\n<h3>\u7279\u8272\u529f\u80fd\u64cd\u4f5c<\/h3>\n<ul>\n<li><strong>\u5de5\u5177\u8c03\u7528<\/strong>\uff1a\u6a21\u578b\u652f\u6301\u51fd\u6570\u8c03\u7528\u548c Web \u641c\u7d22\u3002\u4f8b\u5982\uff0c\u8c03\u7528\u5929\u6c14\u51fd\u6570\uff1a\n<pre><code>tools = [{\"type\": \"function\", \"function\": {\"name\": \"get_current_weather\", \"description\": \"\u83b7\u53d6\u6307\u5b9a\u5730\u70b9\u7684\u5929\u6c14\", \"parameters\": {\"type\": \"object\", \"properties\": {\"location\": {\"type\": \"string\"}}}}}]\r\nmessages = [{\"role\": \"user\", \"content\": \"\u5df4\u9ece\u7684\u5929\u6c14\u5982\u4f55\uff1f\"}]\r\nresponse = client.chat.completions.create(model=\"openai\/gpt-oss-120b:cerebras\", messages=messages, tools=tools)\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u591a\u8bed\u8a00\u63a8\u7406<\/strong>\uff1a\u901a\u8fc7\u5fae\u8c03\uff0c\u6a21\u578b\u53ef\u751f\u6210\u82f1\u8bed\u3001\u897f\u73ed\u7259\u8bed\u3001\u6cd5\u8bed\u7b49\u8bed\u8a00\u7684\u63a8\u7406\u8fc7\u7a0b\u3002\u7528\u6237\u53ef\u6307\u5b9a\u63a8\u7406\u8bed\u8a00\uff0c\u4f8b\u5982\uff1a\n<pre><code>messages = [{\"role\": \"system\", \"content\": \"Reasoning language: Spanish\"}, {\"role\": \"user\", \"content\": \"\u00bfCu\u00e1l es la capital de Australia?\"}]\r\n<\/code><\/pre>\n<\/li>\n<\/ul>\n<h2>\u5e94\u7528\u573a\u666f<\/h2>\n<ol>\n<li><strong>AI \u5f00\u53d1\u5b9e\u9a8c<\/strong><br \/>\n\u5f00\u53d1\u8005\u53ef\u4f7f\u7528\u4ed3\u5e93\u4e2d\u7684\u811a\u672c\u6d4b\u8bd5 GPT OSS \u6a21\u578b\u5728\u4e0d\u540c\u4efb\u52a1\u4e2d\u7684\u8868\u73b0\uff0c\u5982\u6587\u672c\u751f\u6210\u3001\u4ee3\u7801\u751f\u6210\u6216\u95ee\u7b54\u7cfb\u7edf\u3002\u9002\u5408\u5feb\u901f\u539f\u578b\u5f00\u53d1\u3002<\/li>\n<li><strong>\u672c\u5730\u6a21\u578b\u90e8\u7f72<\/strong><br \/>\n\u4f01\u4e1a\u6216\u4e2a\u4eba\u53ef\u5728\u672c\u5730\u8bbe\u5907\u4e0a\u90e8\u7f72\u00a0<code>gpt-oss-20b<\/code>\uff0c\u7528\u4e8e\u9690\u79c1\u654f\u611f\u7684\u573a\u666f\uff0c\u5982\u5185\u90e8\u6587\u6863\u5904\u7406\u6216\u5ba2\u6237\u652f\u6301\u3002<\/li>\n<li><strong>\u6559\u80b2\u4e0e\u7814\u7a76<\/strong><br \/>\n\u7814\u7a76\u4eba\u5458\u53ef\u5229\u7528\u5fae\u8c03\u6559\u7a0b\uff0c\u57fa\u4e8e\u7279\u5b9a\u6570\u636e\u96c6\uff08\u5982\u591a\u8bed\u8a00\u63a8\u7406\uff09\u4f18\u5316\u6a21\u578b\uff0c\u63a2\u7d22\u5927\u6a21\u578b\u5728\u5b66\u672f\u9886\u57df\u7684\u5e94\u7528\u3002<\/li>\n<li><strong>\u751f\u4ea7\u73af\u5883\u96c6\u6210<\/strong><br \/>\n\u4ed3\u5e93\u652f\u6301\u901a\u8fc7 vLLM \u90e8\u7f72 API \u670d\u52a1\u5668\uff0c\u9002\u5408\u5c06\u6a21\u578b\u96c6\u6210\u5230\u751f\u4ea7\u73af\u5883\u4e2d\uff0c\u5982\u804a\u5929\u673a\u5668\u4eba\u6216\u81ea\u52a8\u5316\u5de5\u4f5c\u6d41\u3002<\/li>\n<\/ol>\n<h2>QA<\/h2>\n<ol>\n<li><strong>\u4ed3\u5e93\u652f\u6301\u54ea\u4e9b\u6a21\u578b\uff1f<\/strong><br \/>\n\u4ed3\u5e93\u652f\u6301\u00a0<code>gpt-oss-120b<\/code>\uff08117B \u53c2\u6570\uff09\u548c\u00a0<code>gpt-oss-20b<\/code>\uff0821B \u53c2\u6570\uff09\uff0c\u5206\u522b\u9002\u5408\u9ad8\u6027\u80fd GPU \u548c\u6d88\u8d39\u7ea7\u786c\u4ef6\u3002<\/li>\n<li><strong>\u5982\u4f55\u9009\u62e9\u9002\u5408\u7684\u6a21\u578b\uff1f<\/strong><br \/>\n\u5982\u679c\u6709 H100 GPU\uff0c\u63a8\u8350\u4f7f\u7528\u00a0<code>gpt-oss-120b<\/code>\uff1b\u5982\u679c\u4f7f\u7528\u666e\u901a\u8bbe\u5907\uff0816GB \u5185\u5b58\uff09\uff0c\u9009\u62e9\u00a0<code>gpt-oss-20b<\/code>\u3002<\/li>\n<li><strong>\u9700\u8981\u54ea\u4e9b\u786c\u4ef6\uff1f<\/strong><br \/>\n<code>gpt-oss-20b<\/code>\u00a0\u9700\u8981 16GB \u5185\u5b58\uff0c<code>gpt-oss-120b<\/code>\u00a0\u9700\u8981 80GB GPU\uff08\u5982 H100\uff09\u3002MXFP4 \u91cf\u5316\u53ef\u964d\u4f4e\u8d44\u6e90\u9700\u6c42\u3002<\/li>\n<li><strong>\u5982\u4f55\u5904\u7406\u6a21\u578b\u63a8\u7406\u4e2d\u7684\u9519\u8bef\uff1f<\/strong><br \/>\n\u786e\u4fdd\u4f7f\u7528 harmony \u683c\u5f0f\u5904\u7406\u8f93\u5165\u8f93\u51fa\u3002\u68c0\u67e5\u786c\u4ef6\u517c\u5bb9\u6027\u5e76\u66f4\u65b0\u4f9d\u8d56\u9879\uff0c\u5982 PyTorch \u548c Triton \u5185\u6838\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>gpt-oss-recipes\u00a0\u662f\u7531 Hugging Face \u7ef4\u62a4\u7684\u4e00\u4e2a GitHub \u4ed3\u5e93\uff0c\u4e13\u6ce8\u4e8e\u63d0\u4f9b\u4f7f\u7528 OpenAI GPT OSS \u6a21\u578b\u7684\u811a\u672c\u548c Jupyter Notebook \u6559\u7a0b\u3002\u4ed3\u5e93\u5305\u542b\u4e86\u9488\u5bf9 OpenAI \u6700\u65b0\u5f00\u6e90\u6a21\u578b\u00a0&#8230;<\/p>\n","protected":false},"author":1,"featured_media":32782,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[425,20,483],"tags":[230],"class_list":["post-34965","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-professional","category-tool","category-fine-tuning","tag-aikaiyuanxiangmu"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/34965","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/comments?post=34965"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/34965\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media\/32782"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media?parent=34965"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/categories?post=34965"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/tags?post=34965"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}