{"id":32121,"date":"2025-07-05T02:04:26","date_gmt":"2025-07-04T18:04:26","guid":{"rendered":"https:\/\/www.kdjingpai.com\/?p=32121"},"modified":"2025-07-05T02:04:26","modified_gmt":"2025-07-04T18:04:26","slug":"deepseek-tng-r1t2-chimera","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/en\/deepseek-tng-r1t2-chimera\/","title":{"rendered":"DeepSeek-TNG-R1T2-Chimera\uff1a\u5fb7\u56fd TNG \u53d1\u5e03\u7684 DeepSeek \u589e\u5f3a\u7248"},"content":{"rendered":"<p>DeepSeek-TNG-R1T2-Chimera \u662f\u7531 TNG Technology Consulting GmbH \u5f00\u53d1\u7684\u4e00\u6b3e\u5f00\u6e90\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff0c\u6258\u7ba1\u5728 Hugging Face \u5e73\u53f0\u4e0a\u3002\u8be5\u6a21\u578b\u4e8e 2025 \u5e74 7 \u6708 2 \u65e5\u53d1\u5e03\uff0c\u662f DeepSeek-R1T-Chimera \u7684\u5347\u7ea7\u7248\uff0c\u878d\u5408\u4e86 R1\u3001V3-0324 \u548c R1-0528 \u4e09\u4e2a\u6bcd\u6a21\u578b\uff0c\u901a\u8fc7 Assembly of Experts\uff08AoE\uff09\u65b9\u6cd5\u8fdb\u884c\u7cbe\u7ec6\u5316\u6784\u5efa\u3002R1T2 \u5728\u901f\u5ea6\u548c\u667a\u80fd\u6027\u4e0a\u627e\u5230\u5e73\u8861\u70b9\uff0c\u76f8\u6bd4 R1 \u5feb\u7ea6 20%\uff0c\u6bd4 R1-0528 \u5feb\u4e24\u500d\u4ee5\u4e0a\uff0c\u540c\u65f6\u5728 GPQA \u548c AIME-24\/25 \u7b49\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u8868\u73b0\u51fa\u66f4\u9ad8\u7684\u667a\u80fd\u6027\u3002\u5b83\u4fee\u590d\u4e86\u524d\u4ee3\u6a21\u578b\u7684\u00a0\u6807\u8bb0\u4e00\u81f4\u6027\u95ee\u9898\uff0c\u9002\u5408\u9700\u8981\u9ad8\u6548\u63a8\u7406\u548c\u5feb\u901f\u54cd\u5e94\u7684\u573a\u666f\u3002\u6a21\u578b\u91c7\u7528 MIT \u8bb8\u53ef\u8bc1\uff0c\u5f00\u653e\u6743\u91cd\uff0c\u4f9b\u5f00\u53d1\u8005\u514d\u8d39\u4f7f\u7528\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"aligncenter size-full wp-image-32122\" title=\"DeepSeek-TNG-R1T2-Chimera\uff1a\u5fb7\u56fd TNG \u53d1\u5e03\u7684 DeepSeek \u589e\u5f3a\u7248-1\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/07\/a560d053bb70d37.jpg\" alt=\"DeepSeek-TNG-R1T2-Chimera\uff1a\u5fb7\u56fd TNG \u53d1\u5e03\u7684 DeepSeek \u589e\u5f3a\u7248-1\" width=\"1337\" height=\"999\" srcset=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/07\/a560d053bb70d37.jpg 1337w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/07\/a560d053bb70d37-16x12.jpg 16w\" sizes=\"auto, (max-width: 1337px) 100vw, 1337px\" \/><\/p>\n<p>&nbsp;<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li>\u9ad8\u6548\u6587\u672c\u751f\u6210\uff1a\u5feb\u901f\u751f\u6210\u6d41\u7545\u3001\u51c6\u786e\u7684\u6587\u672c\uff0c\u9002\u5408\u5bf9\u8bdd\u3001\u5185\u5bb9\u521b\u4f5c\u7b49\u4efb\u52a1\u3002<\/li>\n<li>\u9ad8\u7ea7\u63a8\u7406\u80fd\u529b\uff1a\u652f\u6301\u590d\u6742\u95ee\u9898\u5206\u6790\u548c\u903b\u8f91\u63a8\u7406\uff0c\u9002\u7528\u4e8e\u5b66\u672f\u7814\u7a76\u548c\u6280\u672f\u6587\u6863\u5904\u7406\u3002<\/li>\n<li>\u591a\u8bed\u8a00\u652f\u6301\uff1a\u5904\u7406\u591a\u79cd\u8bed\u8a00\u8f93\u5165\uff0c\u9002\u5408\u56fd\u9645\u5316\u5e94\u7528\u573a\u666f\u3002<\/li>\n<li>\u4f18\u5316 <a href=\"https:\/\/www.kdjingpai.com\/en\/tokenization\/\">token<\/a> \u6548\u7387\uff1a\u76f8\u6bd4 R1-0528\uff0c\u8f93\u51fa token \u66f4\u5c11\uff0c\u964d\u4f4e\u8ba1\u7b97\u6210\u672c\u3002<\/li>\n<li>\u4fee\u590d\u00a0\u6807\u8bb0\u95ee\u9898\uff1a\u786e\u4fdd\u63a8\u7406\u8fc7\u7a0b\u4e2d\u7684\u4e00\u81f4\u6027\uff0c\u63d0\u5347\u6a21\u578b\u53ef\u9760\u6027\u3002<\/li>\n<li>\u5f00\u6e90\u6a21\u578b\u6743\u91cd\uff1a\u57fa\u4e8e MIT \u8bb8\u53ef\u8bc1\uff0c\u5141\u8bb8\u7528\u6237\u81ea\u7531\u4e0b\u8f7d\u3001\u4fee\u6539\u548c\u90e8\u7f72\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>\u5b89\u88c5\u6d41\u7a0b<\/h3>\n<p>DeepSeek-TNG-R1T2-Chimera \u662f\u4e00\u4e2a\u6258\u7ba1\u5728 Hugging Face \u7684\u6a21\u578b\uff0c\u9700\u901a\u8fc7 Python \u73af\u5883\u7ed3\u5408 Hugging Face \u7684 Transformers \u5e93\u4f7f\u7528\u3002\u4ee5\u4e0b\u662f\u8be6\u7ec6\u5b89\u88c5\u548c\u4f7f\u7528\u6b65\u9aa4\uff1a<\/p>\n<h4>1. \u5b89\u88c5\u73af\u5883<\/h4>\n<p>\u786e\u4fdd\u672c\u5730\u6216\u4e91\u7aef\u73af\u5883\u5df2\u5b89\u88c5 Python 3.8 \u6216\u66f4\u9ad8\u7248\u672c\uff0c\u5e76\u914d\u7f6e\u597d pip \u5305\u7ba1\u7406\u5668\u3002\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u5b89\u88c5\u5fc5\u8981\u4f9d\u8d56\uff1a<\/p>\n<pre><code>pip install transformers torch\r\n<\/code><\/pre>\n<ul>\n<li><code>transformers<\/code>\u00a0\u662f Hugging Face \u63d0\u4f9b\u7684\u5e93\uff0c\u7528\u4e8e\u52a0\u8f7d\u548c\u8fd0\u884c\u6a21\u578b\u3002<\/li>\n<li><code>torch<\/code>\u00a0\u662f PyTorch \u6846\u67b6\uff0c\u786e\u4fdd\u6a21\u578b\u63a8\u7406\u6b63\u5e38\u8fd0\u884c\u3002<\/li>\n<\/ul>\n<p>\u5982\u679c\u4f7f\u7528 GPU \u52a0\u901f\uff0c\u9700\u5b89\u88c5\u652f\u6301 CUDA \u7684 PyTorch \u7248\u672c\u3002\u8bf7\u53c2\u8003 PyTorch \u5b98\u7f51\uff0c\u6839\u636e\u786c\u4ef6\u914d\u7f6e\u9009\u62e9\u5408\u9002\u7684\u7248\u672c\uff0c\u4f8b\u5982\uff1a<\/p>\n<pre><code>pip install torch --index-url https:\/\/download.pytorch.org\/whl\/cu118\r\n<\/code><\/pre>\n<h4>2. \u4e0b\u8f7d\u6a21\u578b<\/h4>\n<p>DeepSeek-TNG-R1T2-Chimera \u7684\u6a21\u578b\u6743\u91cd\u53ef\u76f4\u63a5\u4ece Hugging Face \u4e0b\u8f7d\u3002\u4f7f\u7528\u4ee5\u4e0b Python \u4ee3\u7801\u52a0\u8f7d\u6a21\u578b\uff1a<\/p>\n<pre><code>from transformers import AutoModelForCausalLM, AutoTokenizer\r\nmodel_name = \"tngtech\/DeepSeek-TNG-R1T2-Chimera\"\r\ntokenizer = AutoTokenizer.from_pretrained(model_name)\r\nmodel = AutoModelForCausalLM.from_pretrained(model_name)\r\n<\/code><\/pre>\n<ul>\n<li>\u786e\u4fdd\u7f51\u7edc\u8fde\u63a5\u7a33\u5b9a\uff0c\u6a21\u578b\u6587\u4ef6\u8f83\u5927\uff0c\u4e0b\u8f7d\u53ef\u80fd\u9700\u8981\u65f6\u95f4\u3002<\/li>\n<li>\u5982\u679c\u672c\u5730\u5b58\u50a8\u7a7a\u95f4\u6709\u9650\uff0c\u53ef\u4f7f\u7528 Hugging Face \u7684\u00a0<code>cache_dir<\/code>\u00a0\u53c2\u6570\u6307\u5b9a\u7f13\u5b58\u8def\u5f84\uff1a<\/li>\n<\/ul>\n<pre><code>model = AutoModelForCausalLM.from_pretrained(model_name, cache_dir=\"\/path\/to\/cache\")\r\n<\/code><\/pre>\n<h4>3. \u914d\u7f6e\u8fd0\u884c\u73af\u5883<\/h4>\n<p>\u6a21\u578b\u652f\u6301 CPU \u548c GPU \u8fd0\u884c\u3002GPU \u73af\u5883\u53ef\u663e\u8457\u63d0\u5347\u63a8\u7406\u901f\u5ea6\u3002\u786e\u4fdd GPU \u9a71\u52a8\u548c CUDA \u7248\u672c\u4e0e PyTorch \u517c\u5bb9\u3002\u5982\u679c\u4f7f\u7528\u591a GPU\uff0c\u53ef\u542f\u7528\u00a0<code>device_map=\"auto\"<\/code>\u00a0\u81ea\u52a8\u5206\u914d\uff1a<\/p>\n<pre><code>model = AutoModelForCausalLM.from_pretrained(model_name, device_map=\"auto\")\r\n<\/code><\/pre>\n<h4>4. \u4f7f\u7528\u6a21\u578b<\/h4>\n<p>\u52a0\u8f7d\u6a21\u578b\u540e\uff0c\u53ef\u4ee5\u901a\u8fc7\u4ee5\u4e0b\u4ee3\u7801\u8fdb\u884c\u6587\u672c\u751f\u6210\u6216\u63a8\u7406\uff1a<\/p>\n<pre><code>input_text = \"\u8bf7\u89e3\u91ca\u91cf\u5b50\u8ba1\u7b97\u7684\u57fa\u672c\u539f\u7406\"\r\ninputs = tokenizer(input_text, return_tensors=\"pt\").to(\"cuda\")  # \u5982\u679c\u4f7f\u7528 GPU\r\noutputs = model.generate(**inputs, max_length=200)\r\nresponse = tokenizer.decode(outputs[0], skip_special_tokens=True)\r\nprint(response)\r\n<\/code><\/pre>\n<ul>\n<li><code>max_length<\/code>\u00a0\u53c2\u6570\u63a7\u5236\u751f\u6210\u6587\u672c\u7684\u6700\u5927\u957f\u5ea6\uff0c\u53ef\u6839\u636e\u9700\u6c42\u8c03\u6574\u3002<\/li>\n<li>\u82e5\u9700\u66f4\u9ad8\u8d28\u91cf\u8f93\u51fa\uff0c\u53ef\u8bbe\u7f6e\u00a0<code>temperature=0.7<\/code>\u00a0\u548c\u00a0<code>top_p=0.9<\/code>\u00a0\u8c03\u6574\u751f\u6210\u968f\u673a\u6027\uff1a<\/li>\n<\/ul>\n<pre><code>outputs = model.generate(**inputs, max_length=200, temperature=0.7, top_p=0.9)\r\n<\/code><\/pre>\n<h4>5. \u4e3b\u8981\u529f\u80fd\u64cd\u4f5c<\/h4>\n<ul>\n<li><strong>\u6587\u672c\u751f\u6210<\/strong>\uff1a\u8f93\u5165\u4efb\u610f\u6587\u672c\u63d0\u793a\uff0c\u6a21\u578b\u53ef\u751f\u6210\u8fde\u8d2f\u7684\u56de\u7b54\u3002\u4f8b\u5982\uff0c\u8f93\u5165\u201c\u5199\u4e00\u7bc7\u5173\u4e8eAI\u4f26\u7406\u7684\u77ed\u6587\u201d\uff0c\u6a21\u578b\u4f1a\u751f\u6210\u7ed3\u6784\u6e05\u6670\u7684\u6587\u7ae0\u3002<\/li>\n<li><strong>\u903b\u8f91\u63a8\u7406<\/strong>\uff1a\u8f93\u5165\u590d\u6742\u95ee\u9898\uff0c\u5982\u201c\u89e3\u51b3\u4ee5\u4e0b\u6570\u5b66\u95ee\u9898\uff1ax^2 + 2x &#8211; 8 = 0\u201d\uff0c\u6a21\u578b\u4f1a\u9010\u6b65\u63a8\u7406\u5e76\u7ed9\u51fa\u7b54\u6848\u3002<\/li>\n<li><strong>\u591a\u8bed\u8a00\u4efb\u52a1<\/strong>\uff1a\u8f93\u5165\u975e\u82f1\u8bed\u63d0\u793a\uff0c\u5982\u201c\u7528\u897f\u73ed\u7259\u8bed\u4ecb\u7ecd\u5df4\u9ece\u201d\uff0c\u6a21\u578b\u4f1a\u751f\u6210\u76f8\u5e94\u8bed\u8a00\u7684\u56de\u7b54\u3002<\/li>\n<li><strong>\u4f18\u5316\u63a8\u7406<\/strong>\uff1a\u901a\u8fc7\u8bbe\u7f6e\u00a0<code>max_length<\/code>\u00a0\u548c\u00a0<code>num_beams<\/code>\uff08\u5982\u00a0<code>num_beams=4<\/code>\uff09\u542f\u7528\u675f\u641c\u7d22\uff0c\u63d0\u5347\u751f\u6210\u8d28\u91cf\uff1a<\/li>\n<\/ul>\n<pre><code>outputs = model.generate(**inputs, max_length=200, num_beams=4)\r\n<\/code><\/pre>\n<h4>6. \u90e8\u7f72\u5230\u751f\u4ea7\u73af\u5883<\/h4>\n<p>\u82e5\u9700\u5c06\u6a21\u578b\u90e8\u7f72\u5230\u670d\u52a1\u5668\uff0c\u63a8\u8350\u4f7f\u7528 Hugging Face \u7684 Inference API \u6216\u7b2c\u4e09\u65b9\u63a8\u7406\u670d\u52a1\uff08\u5982 vLLM\uff09\u3002\u672c\u5730\u90e8\u7f72\u65f6\uff0c\u9700\u786e\u4fdd\u670d\u52a1\u5668\u6709\u8db3\u591f\u5185\u5b58\uff08\u5efa\u8bae 32GB \u4ee5\u4e0a\uff09\u548c GPU \u8d44\u6e90\uff08\u81f3\u5c11 16GB \u663e\u5b58\uff09\u3002\u53ef\u53c2\u8003 Hugging Face \u5b98\u65b9\u6587\u6863\uff1a<\/p>\n<pre><code>https:\/\/huggingface.co\/docs\/transformers\/main\/en\/main_classes\/pipelines\r\n<\/code><\/pre>\n<h4>7. \u6ce8\u610f\u4e8b\u9879<\/h4>\n<ul>\n<li>\u6a21\u578b\u672a\u90e8\u7f72\u5728\u4efb\u4f55\u63a8\u7406\u63d0\u4f9b\u5546\uff0c\u9700\u81ea\u884c\u4e0b\u8f7d\u548c\u914d\u7f6e\u3002<\/li>\n<li>\u8fd0\u884c\u524d\u68c0\u67e5\u786c\u4ef6\u8d44\u6e90\uff0c671B \u53c2\u6570\u91cf\u9700\u8981\u8f83\u9ad8\u7b97\u529b\u3002<\/li>\n<li>\u82e5\u9700\u5fae\u8c03\uff0c\u53ef\u4f7f\u7528 Hugging Face \u7684\u00a0<code>Trainer<\/code>\u00a0\u7c7b\uff0c\u53c2\u8003\u5b98\u65b9\u6587\u6863\uff1a<\/li>\n<\/ul>\n<pre><code>https:\/\/huggingface.co\/docs\/transformers\/main\/en\/training\r\n<\/code><\/pre>\n<h3>\u7279\u8272\u529f\u80fd\u64cd\u4f5c<\/h3>\n<ul>\n<li><strong>\u9ad8\u6548\u63a8\u7406<\/strong>\uff1a\u76f8\u6bd4 R1-0528\uff0cR1T2 \u7684 token \u6548\u7387\u66f4\u9ad8\uff0c\u9002\u5408\u9ad8\u9891\u63a8\u7406\u4efb\u52a1\u3002\u8bbe\u7f6e\u00a0<code>max_length=100<\/code>\u00a0\u53ef\u5feb\u901f\u751f\u6210\u77ed\u6587\u672c\u3002<\/li>\n<li><strong>\u6807\u8bb0\u4fee\u590d<\/strong>\uff1a\u6a21\u578b\u5728\u63a8\u7406\u65f6\u81ea\u52a8\u5904\u7406\u00a0\u6807\u8bb0\uff0c\u786e\u4fdd\u8f93\u51fa\u4e00\u81f4\u3002\u65e0\u9700\u624b\u52a8\u5e72\u9884\u3002<\/li>\n<li><strong>\u5f00\u6e90\u7075\u6d3b\u6027<\/strong>\uff1a\u5f00\u53d1\u8005\u53ef\u4fee\u6539\u6a21\u578b\u6743\u91cd\uff0c\u9002\u914d\u7279\u5b9a\u4efb\u52a1\u3002\u4f8b\u5982\uff0c\u5fae\u8c03\u540e\u53ef\u7528\u4e8e\u5b9a\u5236\u5316\u5bf9\u8bdd\u7cfb\u7edf\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>\u5e94\u7528\u573a\u666f<\/h2>\n<ol>\n<li><strong>\u5b66\u672f\u7814\u7a76<\/strong><br \/>\n\u7814\u7a76\u4eba\u5458\u53ef\u4f7f\u7528 R1T2 \u5206\u6790\u5b66\u672f\u6587\u732e\u3001\u751f\u6210\u7814\u7a76\u62a5\u544a\u6216\u89e3\u7b54\u590d\u6742\u95ee\u9898\u3002\u4f8b\u5982\uff0c\u8f93\u5165\u201c\u603b\u7ed3\u91cf\u5b50\u529b\u5b66\u6700\u65b0\u8fdb\u5c55\u201d\uff0c\u6a21\u578b\u4f1a\u63d0\u53d6\u5173\u952e\u4fe1\u606f\u5e76\u751f\u6210\u7b80\u6d01\u62a5\u544a\u3002<\/li>\n<li><strong>\u5185\u5bb9\u521b\u4f5c<\/strong><br \/>\n\u5185\u5bb9\u521b\u4f5c\u8005\u53ef\u5229\u7528\u6a21\u578b\u751f\u6210\u6587\u7ae0\u3001\u793e\u4ea4\u5a92\u4f53\u5e16\u5b50\u6216\u8425\u9500\u6587\u6848\u3002\u8f93\u5165\u201c\u5199\u4e00\u7bc7\u5173\u4e8e\u73af\u4fdd\u7684\u535a\u5ba2\u201d\uff0c\u5373\u53ef\u83b7\u5f97\u7ed3\u6784\u6e05\u6670\u7684\u6587\u7ae0\u3002<\/li>\n<li><strong>\u6280\u672f\u5f00\u53d1<\/strong><br \/>\n\u5f00\u53d1\u8005\u53ef\u5c06\u6a21\u578b\u96c6\u6210\u5230\u804a\u5929\u673a\u5668\u4eba\u6216\u667a\u80fd\u52a9\u624b\u4e2d\uff0c\u652f\u6301\u591a\u8bed\u8a00\u4ea4\u4e92\u548c\u590d\u6742\u4efb\u52a1\u5904\u7406\u3002\u4f8b\u5982\uff0c\u6784\u5efa\u5ba2\u670d\u673a\u5668\u4eba\u5904\u7406\u7528\u6237\u67e5\u8be2\u3002<\/li>\n<li><strong>\u6559\u80b2\u8f85\u52a9<\/strong><br \/>\n\u5b66\u751f\u548c\u6559\u5e08\u53ef\u4f7f\u7528\u6a21\u578b\u89e3\u7b54\u6570\u5b66\u3001\u7269\u7406\u7b49\u95ee\u9898\uff0c\u6216\u751f\u6210\u5b66\u4e60\u6750\u6599\u3002\u4f8b\u5982\uff0c\u8f93\u5165\u201c\u89e3\u91ca\u725b\u987f\u7b2c\u4e8c\u5b9a\u5f8b\u201d\uff0c\u6a21\u578b\u4f1a\u63d0\u4f9b\u8be6\u7ec6\u8bb2\u89e3\u3002<\/li>\n<\/ol>\n<p>&nbsp;<\/p>\n<h2>QA<\/h2>\n<ol>\n<li><strong>DeepSeek-TNG-R1T2-Chimera \u9002\u5408\u54ea\u4e9b\u7528\u6237\uff1f<\/strong><br \/>\n\u9002\u5408\u9700\u8981\u9ad8\u6548\u6587\u672c\u751f\u6210\u548c\u63a8\u7406\u7684\u5f00\u53d1\u8005\u3001\u7814\u7a76\u4eba\u5458\u548c\u5185\u5bb9\u521b\u4f5c\u8005\u3002\u6a21\u578b\u5f00\u6e90\uff0c\u9002\u5408\u6709\u4e00\u5b9a\u7f16\u7a0b\u80fd\u529b\u7684\u7528\u6237\u3002<\/li>\n<li><strong>\u4e0e DeepSeek-R1T \u76f8\u6bd4\uff0cR1T2 \u6709\u4f55\u6539\u8fdb\uff1f<\/strong><br \/>\nR1T2 \u878d\u5408\u4e09\u4e2a\u6bcd\u6a21\u578b\uff0c\u901f\u5ea6\u63d0\u9ad8 20%\uff0c\u4fee\u590d\u00a0\u6807\u8bb0\u95ee\u9898\uff0c\u5e76\u5728 GPQA \u7b49\u6d4b\u8bd5\u4e2d\u8868\u73b0\u66f4\u4f18\u3002<\/li>\n<li><strong>\u5982\u4f55\u964d\u4f4e\u6a21\u578b\u8fd0\u884c\u7684\u786c\u4ef6\u9700\u6c42\uff1f<\/strong><br \/>\n\u53ef\u4f7f\u7528\u6a21\u578b\u91cf\u5316\u6280\u672f\uff08\u5982 4-bit \u91cf\u5316\uff09\u6216\u4e91\u7aef GPU \u90e8\u7f72\uff0c\u53c2\u8003 Hugging Face \u6587\u6863\u3002<\/li>\n<li><strong>\u6a21\u578b\u652f\u6301\u54ea\u4e9b\u8bed\u8a00\uff1f<\/strong><br \/>\n\u652f\u6301\u591a\u8bed\u8a00\uff0c\u5305\u62ec\u82f1\u8bed\u3001\u4e2d\u6587\u3001\u897f\u73ed\u7259\u8bed\u7b49\uff0c\u5177\u4f53\u652f\u6301\u8303\u56f4\u9700\u6d4b\u8bd5\u9a8c\u8bc1\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>DeepSeek-TNG-R1T2-Chimera \u662f\u7531 TNG Technology Consulting GmbH \u5f00\u53d1\u7684\u4e00\u6b3e\u5f00\u6e90\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff0c\u6258\u7ba1\u5728 Hugging Face \u5e73\u53f0\u4e0a\u3002\u8be5\u6a21\u578b\u4e8e 2025 \u5e74 7 \u6708 2 \u65e5\u53d1\u5e03\uff0c\u662f D&#8230;<\/p>\n","protected":false},"author":1,"featured_media":31260,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20,392,397],"tags":[230],"class_list":["post-32121","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tool","category-models","category-text-model","tag-aikaiyuanxiangmu"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/32121","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=32121"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/32121\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media\/31260"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media?parent=32121"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/categories?post=32121"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/tags?post=32121"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}