{"id":32935,"date":"2025-07-21T06:14:45","date_gmt":"2025-07-20T22:14:45","guid":{"rendered":"https:\/\/www.kdjingpai.com\/?p=32935"},"modified":"2025-07-21T06:14:45","modified_gmt":"2025-07-20T22:14:45","slug":"nextcoder-32b","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/de\/nextcoder-32b\/","title":{"rendered":"NextCoder-32B\uff1a\u652f\u6301\u4ee3\u7801\u7f16\u8f91\u4e0e\u4f18\u5316\u7684\u5f00\u6e90\u5927\u6a21\u578b"},"content":{"rendered":"<p>NextCoder-32B \u662f\u7531\u5fae\u8f6f\u5f00\u53d1\u5e76\u5728 Hugging Face \u5e73\u53f0\u4e0a\u53d1\u5e03\u7684\u5f00\u6e90\u4ee3\u7801\u7f16\u8f91\u5927\u6a21\u578b\u3002\u5b83\u57fa\u4e8e Qwen2.5 \u6a21\u578b\uff0c\u901a\u8fc7 Selective Knowledge Transfer\uff08SeleKT\uff09\u6280\u672f\u8fdb\u884c\u4f18\u5316\uff0c\u4e13\u4e3a\u4ee3\u7801\u751f\u6210\u3001\u4fee\u590d\u548c\u4f18\u5316\u8bbe\u8ba1\u3002\u6a21\u578b\u652f\u6301\u9ad8\u8fbe 32K \u4ee4\u724c\u7684\u8d85\u957f\u4e0a\u4e0b\u6587\uff0c\u9002\u7528\u4e8e\u590d\u6742\u7f16\u7a0b\u4efb\u52a1\u3002NextCoder-32B \u5728\u4ee3\u7801\u7f16\u8f91\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u8868\u73b0\u51fa\u8272\uff0c\u4e0e GPT-4o \u6027\u80fd\u76f8\u5f53\uff0c\u76f8\u6bd4\u57fa\u7ebf\u6a21\u578b\u63d0\u5347\u4e86\u7ea6 44%\u3002\u5b83\u63d0\u4f9b\u7b80\u5355\u6613\u7528\u7684\u63a5\u53e3\uff0c\u5f00\u53d1\u8005\u53ef\u901a\u8fc7 Hugging Face \u7684 Transformers \u5e93\u5feb\u901f\u52a0\u8f7d\u548c\u4f7f\u7528\u3002\u8be5\u6a21\u578b\u9002\u7528\u4e8e\u9700\u8981\u9ad8\u6548\u5904\u7406\u4ee3\u7801\u76f8\u5173\u4efb\u52a1\u7684\u5f00\u53d1\u8005\uff0c\u4f46\u9700\u6ce8\u610f\u5176\u5bf9\u6076\u610f\u63d0\u793a\u7684\u6f5c\u5728\u98ce\u9669\uff0c\u5efa\u8bae\u5728\u4f7f\u7528\u524d\u8fdb\u884c\u4ee3\u7801\u5ba1\u67e5\u3002<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li>\u4ee3\u7801\u751f\u6210\uff1a\u6839\u636e\u7528\u6237\u8f93\u5165\u7684\u81ea\u7136\u8bed\u8a00\u63d0\u793a\uff0c\u751f\u6210\u51c6\u786e\u7684\u4ee3\u7801\u7247\u6bb5\u3002<\/li>\n<li>\u4ee3\u7801\u4fee\u590d\uff1a\u81ea\u52a8\u68c0\u6d4b\u5e76\u4fee\u590d\u4ee3\u7801\u4e2d\u7684\u8bed\u6cd5\u9519\u8bef\u3001\u903b\u8f91\u95ee\u9898\u548c\u6f5c\u5728\u6f0f\u6d1e\u3002<\/li>\n<li>\u4ee3\u7801\u4f18\u5316\uff1a\u6539\u8fdb\u4ee3\u7801\u7ed3\u6784\uff0c\u63d0\u5347\u4ee3\u7801\u8fd0\u884c\u6548\u7387\u548c\u53ef\u8bfb\u6027\u3002<\/li>\n<li>\u957f\u4e0a\u4e0b\u6587\u652f\u6301\uff1a\u652f\u6301\u9ad8\u8fbe 32K \u4ee4\u724c\u7684\u8f93\u5165\uff0c\u9002\u5408\u5904\u7406\u5927\u578b\u4ee3\u7801\u5e93\u6216\u590d\u6742\u9879\u76ee\u3002<\/li>\n<li>\u591a\u8bed\u8a00\u652f\u6301\uff1a\u517c\u5bb9 Python\u3001JavaScript\u3001C++ \u7b49\u591a\u79cd\u7f16\u7a0b\u8bed\u8a00\u3002<\/li>\n<li>\u96c6\u6210 Transformers \u5e93\uff1a\u901a\u8fc7 Hugging Face \u7684\u6700\u65b0 Transformers \u5e93\u52a0\u8f7d\u6a21\u578b\uff0c\u63d0\u4f9b\u4fbf\u6377\u7684\u5f00\u53d1\u4f53\u9a8c\u3002<\/li>\n<\/ul>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>\u5b89\u88c5\u6d41\u7a0b<\/h3>\n<p>\u8981\u4f7f\u7528 NextCoder-32B\uff0c\u9700\u5148\u5b89\u88c5 Hugging Face \u7684 Transformers \u5e93\uff0c\u63a8\u8350\u4f7f\u7528\u6700\u65b0\u7248\u672c\uff08\u5efa\u8bae \u22654.37.0\uff09\uff0c\u4ee5\u907f\u514d\u517c\u5bb9\u6027\u95ee\u9898\u3002\u4ee5\u4e0b\u662f\u8be6\u7ec6\u5b89\u88c5\u548c\u4f7f\u7528\u6b65\u9aa4\uff1a<\/p>\n<ol>\n<li><strong>\u5b89\u88c5 Python \u73af\u5883<\/strong><br \/>\n\u786e\u4fdd\u672c\u5730\u5df2\u5b89\u88c5 Python 3.8 \u6216\u4ee5\u4e0a\u7248\u672c\u3002\u53ef\u4ee5\u901a\u8fc7\u4ee5\u4e0b\u547d\u4ee4\u68c0\u67e5 Python \u7248\u672c\uff1a<\/p>\n<pre><code>python --version\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u5b89\u88c5 Transformers \u5e93<\/strong><br \/>\n\u4f7f\u7528 pip \u5b89\u88c5\u6700\u65b0\u7248\u672c\u7684 Transformers \u5e93\uff1a<\/p>\n<pre><code>pip install transformers&gt;=4.37.0\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u5b89\u88c5 PyTorch<\/strong><br \/>\nNextCoder-32B \u4f9d\u8d56 PyTorch \u8fd0\u884c\uff0c\u63a8\u8350\u5b89\u88c5\u652f\u6301 GPU \u7684\u7248\u672c\u4ee5\u63d0\u9ad8\u6027\u80fd\uff1a<\/p>\n<pre><code>pip install torch\r\n<\/code><\/pre>\n<p>\u5982\u679c\u4f7f\u7528 GPU\uff0c\u8bf7\u6839\u636e\u786c\u4ef6\u9009\u62e9\u5408\u9002\u7684 CUDA \u7248\u672c\uff0c\u53c2\u8003\u00a0PyTorch \u5b98\u7f51\u3002<\/li>\n<li><strong>\u4e0b\u8f7d\u6a21\u578b\u548c\u5206\u8bcd\u5668<\/strong><br \/>\nNextCoder-32B \u6a21\u578b\u548c\u5206\u8bcd\u5668\u53ef\u4ee5\u76f4\u63a5\u4ece Hugging Face \u52a0\u8f7d\u3002\u786e\u4fdd\u6709\u8db3\u591f\u7684\u5b58\u50a8\u7a7a\u95f4\uff08\u6a21\u578b\u7ea6\u5360 60GB\uff09\u3002\u4ee5\u4e0b\u662f\u52a0\u8f7d\u4ee3\u7801\u793a\u4f8b\uff1a<\/p>\n<pre><code>from transformers import AutoModelForCausalLM, AutoTokenizer\r\nmodel_name = \"microsoft\/NextCoder-32B\"\r\nmodel = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=\"auto\", device_map=\"auto\")\r\ntokenizer = AutoTokenizer.from_pretrained(model_name)\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u9a8c\u8bc1\u5b89\u88c5<\/strong><br \/>\n\u8fd0\u884c\u4e0a\u8ff0\u4ee3\u7801\u540e\uff0c\u82e5\u65e0\u62a5\u9519\uff0c\u8bf4\u660e\u6a21\u578b\u548c\u5206\u8bcd\u5668\u52a0\u8f7d\u6210\u529f\u3002\u5efa\u8bae\u4f7f\u7528 GPU \u73af\u5883\u4ee5\u52a0\u901f\u63a8\u7406\u3002<\/li>\n<\/ol>\n<h3>\u4f7f\u7528\u65b9\u6cd5<\/h3>\n<p>NextCoder-32B \u7684\u6838\u5fc3\u529f\u80fd\u662f\u4ee3\u7801\u751f\u6210\u548c\u4fee\u590d\u3002\u4ee5\u4e0b\u662f\u5177\u4f53\u64cd\u4f5c\u6d41\u7a0b\uff1a<\/p>\n<h4>1. \u4ee3\u7801\u751f\u6210<\/h4>\n<p>\u7528\u6237\u53ef\u4ee5\u901a\u8fc7\u81ea\u7136\u8bed\u8a00\u63d0\u793a\u751f\u6210\u4ee3\u7801\u3002\u4f8b\u5982\uff0c\u751f\u6210\u4e00\u4e2a\u9664\u6cd5\u51fd\u6570\u5e76\u5904\u7406\u8fb9\u7f18\u60c5\u51b5\uff1a<\/p>\n<pre><code>prompt = \"\"\"Fix the following function that divides two numbers to handle all the edge cases:\r\ndef divide(a, b):\r\nreturn a\/b\r\n\"\"\"\r\nmessages = [{\"role\": \"user\", \"content\": prompt}]\r\ntext = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)\r\nmodel_inputs = tokenizer([text], return_tensors=\"pt\").to(model.device)\r\ngenerated_ids = model.generate(**model_inputs, max_new_tokens=1024)\r\nresponse = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]\r\nprint(response)\r\n<\/code><\/pre>\n<p>\u8f93\u51fa\u793a\u4f8b\u53ef\u80fd\u4e3a\uff1a<\/p>\n<pre><code>def divide(a, b):\r\nif b == 0:\r\nraise ValueError(\"Division by zero is not allowed\")\r\nif not isinstance(a, (int, float)) or not isinstance(b, (int, float)):\r\nraise TypeError(\"Inputs must be numbers\")\r\nreturn a \/ b\r\n<\/code><\/pre>\n<h4>2. \u4ee3\u7801\u4fee\u590d<\/h4>\n<p>NextCoder-32B \u53ef\u81ea\u52a8\u4fee\u590d\u4ee3\u7801\u4e2d\u7684\u9519\u8bef\u3002\u4f8b\u5982\uff0c\u8f93\u5165\u6709\u62fc\u5199\u9519\u8bef\u7684\u4ee3\u7801\uff08\u5982\u00a0<code>returm<\/code>\uff09\uff0c\u6a21\u578b\u4f1a\u7ea0\u6b63\u4e3a\u00a0<code>return<\/code>\u00a0\u5e76\u4f18\u5316\u903b\u8f91\u3002\u64cd\u4f5c\u65b9\u5f0f\u4e0e\u4ee3\u7801\u751f\u6210\u7c7b\u4f3c\uff0c\u53ea\u9700\u5c06\u95ee\u9898\u4ee3\u7801\u4f5c\u4e3a\u63d0\u793a\u8f93\u5165\u3002<\/p>\n<h4>3. \u4ee3\u7801\u4f18\u5316<\/h4>\n<p>\u82e5\u9700\u4f18\u5316\u5df2\u6709\u4ee3\u7801\uff0c\u8f93\u5165\u539f\u59cb\u4ee3\u7801\u5e76\u8bf4\u660e\u4f18\u5316\u76ee\u6807\u3002\u4f8b\u5982\uff1a<\/p>\n<pre><code>prompt = \"\"\"Optimize this function for better performance and readability:\r\ndef factorial(n):\r\nif n == 0:\r\nreturn 1\r\nelse:\r\nreturn n * factorial(n-1)\r\n\"\"\"\r\n<\/code><\/pre>\n<p>\u6a21\u578b\u53ef\u80fd\u8fd4\u56de\u66f4\u9ad8\u6548\u7684\u8fed\u4ee3\u5b9e\u73b0\uff1a<\/p>\n<pre><code>def factorial(n):\r\nif not isinstance(n, int) or n &lt; 0:\r\nraise ValueError(\"Input must be a non-negative integer\")\r\nresult = 1\r\nfor i in range(1, n + 1):\r\nresult *= i\r\nreturn result\r\n<\/code><\/pre>\n<h4>4. \u957f\u4e0a\u4e0b\u6587\u5904\u7406<\/h4>\n<p>NextCoder-32B \u652f\u6301\u9ad8\u8fbe 32K \u4ee4\u724c\u7684\u8f93\u5165\uff0c\u9002\u5408\u5904\u7406\u5927\u578b\u4ee3\u7801\u5e93\u3002\u7528\u6237\u53ef\u4ee5\u5c06\u6574\u4e2a\u6587\u4ef6\u6216\u591a\u4e2a\u51fd\u6570\u4f5c\u4e3a\u8f93\u5165\uff0c\u6a21\u578b\u4f1a\u5206\u6790\u4e0a\u4e0b\u6587\u5e76\u63d0\u4f9b\u51c6\u786e\u7684\u4fee\u6539\u5efa\u8bae\u3002\u4f8b\u5982\uff0c\u8f93\u5165\u4e00\u4e2a\u5b8c\u6574\u7684 Python \u6587\u4ef6\uff0c\u6a21\u578b\u53ef\u4e00\u6b21\u6027\u4fee\u590d\u591a\u4e2a\u51fd\u6570\u7684\u9519\u8bef\u3002<\/p>\n<h4>5. \u6ce8\u610f\u4e8b\u9879<\/h4>\n<ul>\n<li><strong>\u5b89\u5168\u6027<\/strong>\uff1aNextCoder-32B \u53ef\u80fd\u5bf9\u6076\u610f\u63d0\u793a\u751f\u6210\u4e0d\u5b89\u5168\u4ee3\u7801\u3002\u5efa\u8bae\u5728\u6c99\u76d2\u73af\u5883\u4e2d\u8fd0\u884c\u751f\u6210\u7684\u4ee3\u7801\uff0c\u5e76\u8fdb\u884c\u4eba\u5de5\u5ba1\u67e5\u3002<\/li>\n<li><strong>\u786c\u4ef6\u8981\u6c42<\/strong>\uff1a\u6a21\u578b\u8fd0\u884c\u9700\u8981\u9ad8\u6027\u80fd\u786c\u4ef6\uff0c\u63a8\u8350\u81f3\u5c11 16GB \u663e\u5b58\u7684 GPU \u6216 64GB \u5185\u5b58\u7684 CPU\u3002<\/li>\n<li><strong>\u4f9d\u8d56\u7248\u672c<\/strong>\uff1a\u4f7f\u7528 Transformers &lt;4.37.0 \u53ef\u80fd\u5bfc\u81f4\u52a0\u8f7d\u9519\u8bef\uff0c\u5efa\u8bae\u59cb\u7ec8\u4fdd\u6301\u6700\u65b0\u7248\u672c\u3002<\/li>\n<\/ul>\n<h3>\u8fdb\u9636\u4f7f\u7528<\/h3>\n<ul>\n<li><strong>\u6279\u91cf\u5904\u7406<\/strong>\uff1a\u901a\u8fc7\u6279\u91cf\u8f93\u5165\u591a\u4e2a\u63d0\u793a\uff0c\u6a21\u578b\u53ef\u4e00\u6b21\u6027\u751f\u6210\u6216\u4fee\u590d\u591a\u6bb5\u4ee3\u7801\uff0c\u63d0\u9ad8\u6548\u7387\u3002<\/li>\n<li><strong>\u81ea\u5b9a\u4e49\u63d0\u793a<\/strong>\uff1a\u7528\u6237\u53ef\u901a\u8fc7\u8c03\u6574\u63d0\u793a\u6a21\u677f\uff0c\u63a7\u5236\u8f93\u51fa\u98ce\u683c\uff0c\u5982\u751f\u6210\u6ce8\u91ca\u4e30\u5bcc\u7684\u4ee3\u7801\u6216\u7279\u5b9a\u8bed\u8a00\u7684\u4ee3\u7801\u3002<\/li>\n<li><strong>\u4e0e IDE \u96c6\u6210<\/strong>\uff1a\u901a\u8fc7 Hugging Face \u7684 API \u6216\u672c\u5730\u90e8\u7f72\uff0c\u5c06\u6a21\u578b\u96c6\u6210\u5230 VS Code \u6216 PyCharm \u7b49 IDE \u4e2d\uff0c\u5b9e\u73b0\u5b9e\u65f6\u4ee3\u7801\u8865\u5168\u3002<\/li>\n<\/ul>\n<h2>\u5e94\u7528\u573a\u666f<\/h2>\n<ol>\n<li><strong>\u8f6f\u4ef6\u5f00\u53d1<\/strong><br \/>\n\u5f00\u53d1\u8005\u53ef\u4f7f\u7528 NextCoder-32B \u5feb\u901f\u751f\u6210\u6837\u677f\u4ee3\u7801\u3001\u4fee\u590d bug \u6216\u4f18\u5316\u73b0\u6709\u4ee3\u7801\uff0c\u7f29\u77ed\u5f00\u53d1\u5468\u671f\u3002<\/li>\n<li><strong>\u6559\u80b2\u548c\u5b66\u4e60<\/strong><br \/>\n\u7f16\u7a0b\u521d\u5b66\u8005\u53ef\u901a\u8fc7\u6a21\u578b\u751f\u6210\u793a\u4f8b\u4ee3\u7801\uff0c\u7406\u89e3\u7f16\u7a0b\u903b\u8f91\uff0c\u6216\u68c0\u67e5\u7ec3\u4e60\u4ee3\u7801\u4e2d\u7684\u9519\u8bef\u3002<\/li>\n<li><strong>\u4ee3\u7801\u5ba1\u67e5<\/strong><br \/>\n\u56e2\u961f\u53ef\u5728\u4ee3\u7801\u63d0\u4ea4\u524d\u4f7f\u7528\u6a21\u578b\u626b\u63cf\u6f5c\u5728\u95ee\u9898\uff0c\u63d0\u9ad8\u4ee3\u7801\u8d28\u91cf\uff0c\u51cf\u5c11\u4eba\u5de5\u5ba1\u67e5\u65f6\u95f4\u3002<\/li>\n<li><strong>\u5f00\u6e90\u9879\u76ee\u7ef4\u62a4<\/strong><br \/>\n\u5f00\u6e90\u9879\u76ee\u7ef4\u62a4\u8005\u53ef\u5229\u7528\u6a21\u578b\u6279\u91cf\u5904\u7406\u4ee3\u7801\u8d21\u732e\uff0c\u81ea\u52a8\u4fee\u590d\u683c\u5f0f\u9519\u8bef\u6216\u903b\u8f91\u95ee\u9898\u3002<\/li>\n<\/ol>\n<h2>QA<\/h2>\n<ol>\n<li><strong>NextCoder-32B \u652f\u6301\u54ea\u4e9b\u7f16\u7a0b\u8bed\u8a00\uff1f<\/strong><br \/>\n\u6a21\u578b\u652f\u6301 Python\u3001JavaScript\u3001C++\u3001Java \u7b49\u591a\u79cd\u4e3b\u6d41\u7f16\u7a0b\u8bed\u8a00\uff0c\u9002\u7528\u4e8e\u5927\u591a\u6570\u5f00\u53d1\u573a\u666f\u3002<\/li>\n<li><strong>\u5982\u4f55\u907f\u514d\u751f\u6210\u4e0d\u5b89\u5168\u7684\u4ee3\u7801\uff1f<\/strong><br \/>\n\u5728\u6c99\u76d2\u73af\u5883\u4e2d\u8fd0\u884c\u751f\u6210\u7684\u4ee3\u7801\uff0c\u5e76\u8fdb\u884c\u4eba\u5de5\u5ba1\u67e5\u3002\u907f\u514d\u76f4\u63a5\u8f93\u5165\u53ef\u80fd\u89e6\u53d1\u6076\u610f\u884c\u4e3a\u7684\u63d0\u793a\u3002<\/li>\n<li><strong>\u6a21\u578b\u8fd0\u884c\u9700\u8981\u4ec0\u4e48\u786c\u4ef6\uff1f<\/strong><br \/>\n\u63a8\u8350\u4f7f\u7528\u81f3\u5c11 16GB \u663e\u5b58\u7684 GPU \u6216 64GB \u5185\u5b58\u7684 CPU\uff0c\u4ee5\u786e\u4fdd\u6d41\u7545\u8fd0\u884c\u3002<\/li>\n<li><strong>\u662f\u5426\u9700\u8981\u8054\u7f51\u4f7f\u7528\uff1f<\/strong><br \/>\n\u6a21\u578b\u53ef\u79bb\u7ebf\u8fd0\u884c\uff0c\u4f46\u9700\u63d0\u524d\u4e0b\u8f7d\u6a21\u578b\u6587\u4ef6\u548c\u4f9d\u8d56\u5e93\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>NextCoder-32B \u662f\u7531\u5fae\u8f6f\u5f00\u53d1\u5e76\u5728 Hugging Face \u5e73\u53f0\u4e0a\u53d1\u5e03\u7684\u5f00\u6e90\u4ee3\u7801\u7f16\u8f91\u5927\u6a21\u578b\u3002\u5b83\u57fa\u4e8e Qwen2.5 \u6a21\u578b\uff0c\u901a\u8fc7 Selective Knowledge Transfer\uff08SeleKT\uff09\u6280\u672f\u8fdb\u884c\u4f18\u5316\uff0c\u4e13\u4e3a\u4ee3\u7801\u751f\u6210\u3001&#8230;<\/p>\n","protected":false},"author":1,"featured_media":31260,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[392,397],"tags":[],"class_list":["post-32935","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-models","category-text-model"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/32935","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=32935"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/32935\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media\/31260"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media?parent=32935"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/categories?post=32935"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/tags?post=32935"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}