{"id":27853,"date":"2025-03-07T10:28:07","date_gmt":"2025-03-07T02:28:07","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=27853"},"modified":"2025-03-07T10:28:07","modified_gmt":"2025-03-07T02:28:07","slug":"light-r1","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/de\/light-r1\/","title":{"rendered":"Light-R1\uff1a360\u5f00\u6e90\u7684\u6570\u5b66\u9886\u57df\u8d85\u5f3a\u63a8\u7406\u6a21\u578b"},"content":{"rendered":"<p>Light-R1 \u662f\u7531\u5947\u864e360\uff08Qihoo360\uff09\u56e2\u961f\u5f00\u53d1\u7684\u4e00\u6b3e\u5f00\u6e90\u4eba\u5de5\u667a\u80fd\u6a21\u578b\uff0c\u4e13\u6ce8\u4e8e\u6570\u5b66\u9886\u57df\u7684\u957f\u94fe\u63a8\u7406\uff08Chain-of-Thought, COT\uff09\u3002\u5b83\u57fa\u4e8e Qwen2.5-32B-Instruct \u6a21\u578b\uff0c\u901a\u8fc7\u72ec\u7279\u7684\u8bfe\u7a0b\u5f0f\u76d1\u7763\u5fae\u8c03\uff08SFT\uff09\u548c\u76f4\u63a5\u504f\u597d\u4f18\u5316\uff08DPO\uff09\u8bad\u7ec3\u65b9\u6cd5\uff0c\u4ec5\u7528\u7ea61000\u7f8e\u5143\u7684\u6210\u672c\uff0812\u53f0H800\u673a\u56686\u5c0f\u65f6\u8bad\u7ec3\uff09\uff0c\u5728\u6570\u5b66\u7ade\u8d5b AIME24 \u548c AIME25 \u4e0a\u5206\u522b\u53d6\u5f97\u4e8676.6\u548c64.6\u7684\u9ad8\u5206\uff0c\u8d85\u8d8a\u4e86\u6b64\u524d\u8868\u73b0\u4f18\u5f02\u7684 DeepSeek-R1-Distill-Qwen-32B\uff0872.6\u548c54.9\uff09\u3002Light-R1 \u7684\u4eae\u70b9\u5728\u4e8e\u4ece\u65e0\u957f\u94fe\u63a8\u7406\u80fd\u529b\u7684\u6a21\u578b\u8d77\u6b65\uff0c\u7ed3\u5408\u53bb\u6c61\u67d3\u6570\u636e\u548c\u6a21\u578b\u878d\u5408\u6280\u672f\uff0c\u5b9e\u73b0\u4e86\u9ad8\u6548\u4e14\u7ecf\u6d4e\u7684\u6027\u80fd\u7a81\u7834\u3002\u9879\u76ee\u4e0d\u4ec5\u53d1\u5e03\u4e86\u6a21\u578b\uff0c\u8fd8\u5f00\u653e\u4e86\u6240\u6709\u8bad\u7ec3\u6570\u636e\u96c6\u548c\u4ee3\u7801\uff0c\u65e8\u5728\u63a8\u52a8\u957f\u94fe\u63a8\u7406\u6a21\u578b\u7684\u666e\u53ca\u4e0e\u53d1\u5c55\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"aligncenter  wp-image-27854\" title=\"Light-R1\uff1a360\u5f00\u6e90\u7684\u6570\u5b66\u9886\u57df\u8d85\u5f3a\u63a8\u7406\u6a21\u578b-1\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/29e9e1f154add57.png\" alt=\"Light-R1\uff1a360\u5f00\u6e90\u7684\u6570\u5b66\u9886\u57df\u8d85\u5f3a\u63a8\u7406\u6a21\u578b-1\" width=\"665\" height=\"242\" srcset=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/29e9e1f154add57.png 909w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/29e9e1f154add57-768x280.png 768w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/29e9e1f154add57-18x7.png 18w\" sizes=\"auto, (max-width: 665px) 100vw, 665px\" \/><\/p>\n<p>&nbsp;<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li><strong>\u6570\u5b66\u63a8\u7406\u80fd\u529b<\/strong>\uff1a\u9488\u5bf9 AIME \u7b49\u9ad8\u96be\u5ea6\u6570\u5b66\u7ade\u8d5b\uff0c\u63d0\u4f9b\u7cbe\u51c6\u7684\u95ee\u9898\u6c42\u89e3\u80fd\u529b\u3002<\/li>\n<li><strong>\u957f\u94fe\u63a8\u7406\u652f\u6301<\/strong>\uff1a\u901a\u8fc7\u786c\u7f16\u7801\u00a0<code>&lt;think&gt;<\/code>\u00a0\u6807\u7b7e\uff0c\u5f3a\u5236\u6a21\u578b\u9010\u6b65\u63a8\u7406\u590d\u6742\u95ee\u9898\u3002<\/li>\n<li><strong>\u5f00\u6e90\u8d44\u6e90\u5171\u4eab<\/strong>\uff1a\u63d0\u4f9b\u5b8c\u6574\u7684 SFT \u548c DPO \u8bad\u7ec3\u6570\u636e\u96c6\u53ca\u57fa\u4e8e 360-LLaMA-Factory \u7684\u8bad\u7ec3\u811a\u672c\u3002<\/li>\n<li><strong>\u9ad8\u6548\u63a8\u7406\u90e8\u7f72<\/strong>\uff1a\u652f\u6301 <a href=\"https:\/\/www.kdjingpai.com\/de\/vllm\/\">vLLM<\/a> \u548c SGLang \u6846\u67b6\uff0c\u4f18\u5316\u6a21\u578b\u63a8\u7406\u901f\u5ea6\u4e0e\u8d44\u6e90\u5360\u7528\u3002<\/li>\n<li><strong>\u6a21\u578b\u8bc4\u4f30\u5de5\u5177<\/strong>\uff1a\u96c6\u6210 DeepScaleR \u8bc4\u4f30\u4ee3\u7801\uff0c\u63d0\u4f9b AIME24 \u7b49\u57fa\u51c6\u6d4b\u8bd5\u7ed3\u679c\u3002<\/li>\n<li><strong>\u6570\u636e\u53bb\u6c61\u67d3<\/strong>\uff1a\u786e\u4fdd\u8bad\u7ec3\u6570\u636e\u9488\u5bf9 MATH-500\u3001AIME24\/25 \u7b49\u57fa\u51c6\u65e0\u6c61\u67d3\uff0c\u63d0\u5347\u516c\u5e73\u6027\u3002<\/li>\n<\/ul>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>\u83b7\u53d6\u4e0e\u5b89\u88c5\u6d41\u7a0b<\/h3>\n<p>Light-R1 \u662f\u4e00\u4e2a\u5f00\u6e90\u9879\u76ee\uff0c\u6258\u7ba1\u5728 GitHub \u4e0a\uff0c\u7528\u6237\u53ef\u4ee5\u901a\u8fc7\u4ee5\u4e0b\u6b65\u9aa4\u83b7\u53d6\u5e76\u90e8\u7f72\u6a21\u578b\uff1a<\/p>\n<ol>\n<li><strong>\u8bbf\u95ee GitHub \u4ed3\u5e93<\/strong><br \/>\n\u6253\u5f00\u6d4f\u89c8\u5668\uff0c\u8f93\u5165\u7f51\u5740\u00a0<code>https:\/\/github.com\/Qihoo360\/Light-R1<\/code>\uff0c\u8fdb\u5165\u9879\u76ee\u4e3b\u9875\u3002\u9875\u9762\u5305\u542b\u6a21\u578b\u4ecb\u7ecd\u3001\u6570\u636e\u96c6\u94fe\u63a5\u548c\u4ee3\u7801\u8bf4\u660e\u3002<\/li>\n<li><strong>\u514b\u9686\u4ed3\u5e93<\/strong><br \/>\n\u5728\u7ec8\u7aef\u6216\u547d\u4ee4\u884c\u4e2d\u8f93\u5165\u4ee5\u4e0b\u547d\u4ee4\uff0c\u5c06\u9879\u76ee\u514b\u9686\u5230\u672c\u5730\uff1a<\/li>\n<\/ol>\n<pre><code>git clone https:\/\/github.com\/Qihoo360\/Light-R1.git\r\n<\/code><\/pre>\n<p>\u514b\u9686\u5b8c\u6210\u540e\uff0c\u8fdb\u5165\u9879\u76ee\u76ee\u5f55\uff1a<\/p>\n<pre><code>cd Light-R1\r\n<\/code><\/pre>\n<ol start=\"3\">\n<li><strong>\u4e0b\u8f7d\u6a21\u578b\u6587\u4ef6<\/strong><br \/>\nLight-R1-32B \u6a21\u578b\u6258\u7ba1\u5728 Hugging Face \u4e0a\u3002\u8bbf\u95ee\u00a0<code>https:\/\/huggingface.co\/Qihoo360\/Light-R1-32B<\/code>\uff0c\u6839\u636e\u9875\u9762\u63d0\u793a\u4e0b\u8f7d\u6a21\u578b\u6743\u91cd\u6587\u4ef6\u3002\u4e0b\u8f7d\u540e\uff0c\u5c06\u6587\u4ef6\u653e\u7f6e\u5728\u672c\u5730\u4ed3\u5e93\u7684\u5408\u9002\u76ee\u5f55\uff08\u5982\u00a0<code>models\/<\/code>\uff09\uff0c\u5177\u4f53\u8def\u5f84\u53ef\u53c2\u8003\u9879\u76ee\u6587\u6863\u3002<\/li>\n<li><strong>\u5b89\u88c5\u4f9d\u8d56\u73af\u5883<\/strong><br \/>\nLight-R1 \u63a8\u8350\u4f7f\u7528 vLLM \u6216 SGLang \u8fdb\u884c\u63a8\u7406\uff0c\u9700\u5b89\u88c5\u76f8\u5173\u4f9d\u8d56\u3002\u4ee5 vLLM \u4e3a\u4f8b\uff1a<\/li>\n<\/ol>\n<ul>\n<li>\u786e\u4fdd\u5df2\u5b89\u88c5 Python 3.8 \u6216\u4ee5\u4e0a\u7248\u672c\u3002<\/li>\n<li>\u5b89\u88c5 vLLM\uff1a\n<pre><code>pip install vllm\r\n<\/code><\/pre>\n<\/li>\n<li>\u5982\u679c\u9700\u8981 GPU \u652f\u6301\uff0c\u786e\u4fdd CUDA \u5df2\u914d\u7f6e\u597d\uff08\u63a8\u8350 12 \u53f0 H800 \u6216\u540c\u7b49\u7b97\u529b\u8bbe\u5907\uff09\u3002<\/li>\n<\/ul>\n<ol start=\"5\">\n<li><strong>\u51c6\u5907\u6570\u636e\u96c6\uff08\u53ef\u9009\uff09<\/strong><br \/>\n\u5982\u679c\u9700\u8981\u590d\u73b0\u8bad\u7ec3\u6216\u81ea\u5b9a\u4e49\u5fae\u8c03\uff0c\u53ef\u4ece GitHub \u9875\u9762\u4e0b\u8f7d SFT \u548c DPO \u6570\u636e\u96c6\uff08\u94fe\u63a5\u5728\u00a0<code>Curriculum SFT &amp; DPO datasets<\/code>\u00a0\u90e8\u5206\uff09\u3002\u89e3\u538b\u540e\u653e\u7f6e\u5728\u00a0<code>data\/<\/code>\u00a0\u76ee\u5f55\u3002<\/li>\n<\/ol>\n<h3>\u4e3b\u8981\u529f\u80fd\u64cd\u4f5c\u6d41\u7a0b<\/h3>\n<h4>1. \u4f7f\u7528 Light-R1 \u8fdb\u884c\u6570\u5b66\u63a8\u7406<\/h4>\n<p>Light-R1 \u7684\u6838\u5fc3\u529f\u80fd\u662f\u89e3\u51b3\u6570\u5b66\u95ee\u9898\uff0c\u5c24\u5176\u662f\u9700\u8981\u957f\u94fe\u63a8\u7406\u7684\u590d\u6742\u9898\u76ee\u3002\u4ee5\u4e0b\u662f\u64cd\u4f5c\u6b65\u9aa4\uff1a<\/p>\n<ul>\n<li><strong>\u542f\u52a8\u63a8\u7406\u670d\u52a1<\/strong><br \/>\n\u5728\u7ec8\u7aef\u4e2d\u8fdb\u5165\u9879\u76ee\u76ee\u5f55\uff0c\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u542f\u52a8 vLLM \u63a8\u7406\u670d\u52a1\uff1a<\/li>\n<\/ul>\n<pre><code>python -m vllm.entrypoints.api_server --model path\/to\/Light-R1-32B\r\n<\/code><\/pre>\n<p>\u5176\u4e2d\u00a0<code>path\/to\/Light-R1-32B<\/code>\u00a0\u66ff\u6362\u4e3a\u5b9e\u9645\u6a21\u578b\u6587\u4ef6\u8def\u5f84\u3002\u542f\u52a8\u540e\uff0c\u670d\u52a1\u9ed8\u8ba4\u76d1\u542c\u672c\u5730\u7aef\u53e3\uff08\u901a\u5e38\u4e3a 8000\uff09\u3002<\/p>\n<ul>\n<li><strong>\u53d1\u9001\u63a8\u7406\u8bf7\u6c42<\/strong><br \/>\n\u4f7f\u7528 Python \u811a\u672c\u6216 curl \u547d\u4ee4\u5411\u6a21\u578b\u53d1\u9001\u6570\u5b66\u95ee\u9898\u3002\u4ee5 curl \u4e3a\u4f8b\uff1a<\/li>\n<\/ul>\n<pre><code>curl http:\/\/localhost:8000\/v1\/completions \r\n-H \"Content-Type: application\/json\" \r\n-d '{\r\n\"model\": \"Light-R1-32B\",\r\n\"prompt\": \"&lt;think&gt;Solve the equation: 2x + 3 = 7&lt;\/think&gt;\",\r\n\"max_tokens\": 200\r\n}'\r\n<\/code><\/pre>\n<p>\u6a21\u578b\u4f1a\u8fd4\u56de\u9010\u6b65\u63a8\u7406\u8fc7\u7a0b\u548c\u7b54\u6848\uff0c\u4f8b\u5982\uff1a<\/p>\n<pre><code>{\r\n\"choices\": [{\r\n\"text\": \"&lt;think&gt;First, subtract 3 from both sides: 2x + 3 - 3 = 7 - 3, so 2x = 4. Then, divide both sides by 2: 2x \/ 2 = 4 \/ 2, so x = 2.&lt;\/think&gt; The solution is x = 2.\"\r\n}]\r\n}\r\n<\/code><\/pre>\n<ul>\n<li><strong>\u6ce8\u610f\u4e8b\u9879<\/strong><\/li>\n<li><code>&lt;think&gt;<\/code>\u00a0\u6807\u7b7e\u662f\u786c\u7f16\u7801\u7684\uff0c\u5fc5\u987b\u5305\u542b\u5728\u8f93\u5165\u4e2d\u4ee5\u89e6\u53d1\u957f\u94fe\u63a8\u7406\u3002<\/li>\n<li>\u5bf9\u4e8e AIME \u7ea7\u522b\u96be\u9898\uff0c\u5efa\u8bae\u589e\u52a0\u00a0<code>max_tokens<\/code>\uff08\u5982 500\uff09\uff0c\u786e\u4fdd\u63a8\u7406\u5b8c\u6574\u3002<\/li>\n<\/ul>\n<h4>2. \u590d\u73b0\u6a21\u578b\u8bad\u7ec3<\/h4>\n<p>\u5982\u679c\u60f3\u590d\u73b0 Light-R1 \u7684\u8bad\u7ec3\u8fc7\u7a0b\u6216\u57fa\u4e8e\u5176\u8fdb\u884c\u4e8c\u6b21\u5f00\u53d1\uff0c\u53ef\u6309\u4ee5\u4e0b\u6b65\u9aa4\u64cd\u4f5c\uff1a<\/p>\n<ul>\n<li><strong>\u51c6\u5907\u8bad\u7ec3\u73af\u5883<\/strong><br \/>\n\u4f7f\u7528 360-LLaMA-Factory \u6846\u67b6\uff0c\u5b89\u88c5\u4f9d\u8d56\uff1a<\/li>\n<\/ul>\n<pre><code>pip install -r train-scripts\/requirements.txt\r\n<\/code><\/pre>\n<ul>\n<li><strong>\u8fd0\u884c SFT Stage 1<\/strong><br \/>\n\u7f16\u8f91\u00a0<code>train-scripts\/sft_stage1.sh<\/code>\uff0c\u786e\u4fdd\u6a21\u578b\u8def\u5f84\u548c\u6570\u636e\u96c6\u8def\u5f84\u6b63\u786e\uff0c\u7136\u540e\u6267\u884c\uff1a<\/li>\n<\/ul>\n<pre><code>bash train-scripts\/sft_stage1.sh\r\n<\/code><\/pre>\n<p>\u8be5\u9636\u6bb5\u4f7f\u7528 76k \u6570\u636e\u96c6\uff0c\u7ea6\u9700 3 \u5c0f\u65f6\uff0812 \u53f0 H800\uff09\u3002<\/p>\n<ul>\n<li><strong>\u8fd0\u884c SFT Stage 2<\/strong><br \/>\n\u7c7b\u4f3c\u5730\uff0c\u8fd0\u884c\uff1a<\/li>\n<\/ul>\n<pre><code>bash train-scripts\/sft_stage2.sh\r\n<\/code><\/pre>\n<p>\u4f7f\u7528 3k \u66f4\u96be\u6570\u636e\u96c6\uff0c\u63d0\u5347\u6a21\u578b\u6027\u80fd\u3002<\/p>\n<ul>\n<li><strong>\u8fd0\u884c DPO<\/strong><br \/>\n\u6267\u884c\uff1a<\/li>\n<\/ul>\n<pre><code>bash train-scripts\/dpo.sh\r\n<\/code><\/pre>\n<p>DPO \u57fa\u4e8e SFT Stage 2 \u7ed3\u679c\uff0c\u8fdb\u4e00\u6b65\u4f18\u5316\u63a8\u7406\u80fd\u529b\u3002<\/p>\n<ul>\n<li><strong>\u6a21\u578b\u5408\u5e76<\/strong><br \/>\n\u4f7f\u7528\u63d0\u4f9b\u7684\u5408\u5e76\u811a\u672c\uff08\u5982\u00a0<code>merge_models.py<\/code>\uff09\uff0c\u5c06 SFT \u548c DPO \u6a21\u578b\u878d\u5408\uff1a<\/li>\n<\/ul>\n<pre><code>python merge_models.py --sft-model sft_stage2 --dpo-model dpo --output Light-R1-32B\r\n<\/code><\/pre>\n<h4>3. \u8bc4\u4f30\u6a21\u578b\u6027\u80fd<\/h4>\n<p>Light-R1 \u63d0\u4f9b\u8bc4\u4f30\u5de5\u5177\uff0c\u53ef\u6d4b\u8bd5 AIME24 \u7b49\u57fa\u51c6\uff1a<\/p>\n<ul>\n<li><strong>\u8fd0\u884c\u8bc4\u4f30\u811a\u672c<\/strong><br \/>\n\u5728\u00a0<code>deepscaler-release\/<\/code>\u00a0\u76ee\u5f55\u4e0b\uff0c\u6267\u884c\uff1a<\/li>\n<\/ul>\n<pre><code>python evaluate.py --model Light-R1-32B --benchmark AIME24\r\n<\/code><\/pre>\n<p>\u7ed3\u679c\u5c06\u8bb0\u5f55\u5728\u65e5\u5fd7\u4e2d\uff0c64 \u6b21\u5e73\u5747\u5f97\u5206\u5e94\u63a5\u8fd1 76.6\u3002<\/p>\n<h3>\u7279\u8272\u529f\u80fd\u8be6\u89e3<\/h3>\n<h4>\u957f\u94fe\u63a8\u7406\u4f18\u5316<\/h4>\n<p>Light-R1 \u901a\u8fc7\u00a0<code>&lt;think&gt;<\/code>\u00a0\u548c\u00a0<code>&lt;\/think&gt;<\/code>\u00a0\u7279\u6b8a\u6807\u8bb0\uff0c\u786e\u4fdd\u6a21\u578b\u5728\u6570\u5b66\u95ee\u9898\u4e0a\u9010\u6b65\u63a8\u7406\u3002\u4f8b\u5982\uff0c\u8f93\u5165\uff1a<\/p>\n<pre><code>&lt;think&gt;Find the number of positive integers less than 100 that are divisible by 3 or 5.&lt;\/think&gt;\r\n<\/code><\/pre>\n<p>\u6a21\u578b\u8f93\u51fa\uff1a<\/p>\n<pre><code>&lt;think&gt;Let\u2019s use inclusion-exclusion. Numbers divisible by 3: 99 \u00f7 3 = 33. Numbers divisible by 5: 99 \u00f7 5 = 19. Numbers divisible by 15 (LCM of 3 and 5): 99 \u00f7 15 = 6. Total = 33 + 19 - 6 = 46.&lt;\/think&gt; Answer: 46.\r\n<\/code><\/pre>\n<h4>\u6570\u636e\u53bb\u6c61\u67d3\u4fdd\u969c<\/h4>\n<p>\u8bad\u7ec3\u6570\u636e\u7ecf\u8fc7\u4e25\u683c\u53bb\u6c61\u67d3\uff0c\u786e\u4fdd AIME24\/25 \u7b49\u57fa\u51c6\u516c\u5e73\u6027\u3002\u7528\u6237\u53ef\u901a\u8fc7\u68c0\u67e5\u6570\u636e\u96c6\uff08<code>data\/<\/code>\u00a0\u76ee\u5f55\uff09\u9a8c\u8bc1\u65e0\u91cd\u590d\u9898\u76ee\u3002<\/p>\n<h4>\u4f4e\u6210\u672c\u8bad\u7ec3\u8303\u4f8b<\/h4>\n<p>Light-R1 \u8bc1\u660e\u4e86\u9ad8\u6548\u8bad\u7ec3\u7684\u53ef\u884c\u6027\uff0c\u7528\u6237\u53ef\u53c2\u8003\u8bad\u7ec3\u811a\u672c\uff0c\u9488\u5bf9\u5176\u4ed6\u9886\u57df\uff08\u5982\u7269\u7406\uff09\u5b9a\u5236\u6a21\u578b\u3002<\/p>\n<h3>\u4f7f\u7528\u6280\u5de7<\/h3>\n<ul>\n<li><strong>\u63d0\u9ad8\u63a8\u7406\u51c6\u786e\u6027<\/strong>\uff1a\u589e\u52a0\u00a0<code>max_tokens<\/code>\u00a0\u6216\u591a\u6b21\u8fd0\u884c\u53d6\u5e73\u5747\u503c\u3002<\/li>\n<li><strong>\u8c03\u8bd5\u95ee\u9898<\/strong>\uff1a\u67e5\u770b\u8bc4\u4f30\u65e5\u5fd7\uff0c\u5206\u6790\u6a21\u578b\u5728\u7279\u5b9a\u9898\u76ee\u4e0a\u7684\u63a8\u7406\u8fc7\u7a0b\u3002<\/li>\n<li><strong>\u793e\u533a\u652f\u6301<\/strong>\uff1a\u52a0\u5165 GitHub \u9875\u9762\u63d0\u4f9b\u7684 WeChat \u7fa4\uff0c\u4e0e\u5f00\u53d1\u8005\u4ea4\u6d41\u3002<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Light-R1 \u662f\u7531\u5947\u864e360\uff08Qihoo360\uff09\u56e2\u961f\u5f00\u53d1\u7684\u4e00\u6b3e\u5f00\u6e90\u4eba\u5de5\u667a\u80fd\u6a21\u578b\uff0c\u4e13\u6ce8\u4e8e\u6570\u5b66\u9886\u57df\u7684\u957f\u94fe\u63a8\u7406\uff08Chain-of-Thought, COT\uff09\u3002\u5b83\u57fa\u4e8e Qwen2.5-32B-Instruct \u6a21\u578b\uff0c\u901a\u8fc7\u72ec\u7279\u7684\u8bfe\u7a0b\u5f0f\u76d1\u7763\u5fae\u8c03\uff08S&#8230;<\/p>\n","protected":false},"author":1,"featured_media":61985,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[230],"class_list":["post-27853","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tool","tag-aikaiyuanxiangmu"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/27853","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/comments?post=27853"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/27853\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media\/61985"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media?parent=27853"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/categories?post=27853"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/tags?post=27853"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}