{"id":29364,"date":"2025-03-25T17:16:29","date_gmt":"2025-03-25T09:16:29","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=29364"},"modified":"2025-03-25T17:16:29","modified_gmt":"2025-03-25T09:16:29","slug":"deepseek-v3-moxingbai","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/de\/deepseek-v3-moxingbai\/","title":{"rendered":"DeepSeek-V3 \u6a21\u578b\u4f4e\u8c03\u66f4\u65b0\uff0c\u4ee3\u7801\u80fd\u529b\u8dc3\u5347\uff0c\u5251\u6307 Claude-3.7"},"content":{"rendered":"<p>\u79d1\u6280\u9886\u57df\u7684\u7ade\u4e89\u603b\u662f\u6697\u6d41\u6d8c\u52a8\u3002\u8fd1\u65e5\uff0c\u4e2d\u56fd AI \u521b\u4f01 DeepSeek\uff08\u6df1\u5ea6\u6c42\u7d22\uff09\u56e2\u961f\u5728\u672a\u8fdb\u884c\u5927\u89c4\u6a21\u5ba3\u4f20\u7684\u60c5\u51b5\u4e0b\uff0c\u4f4e\u8c03\u5730\u66f4\u65b0\u4e86\u5176 V3 \u57fa\u5ea7\u6a21\u578b\uff0c\u65b0\u7248\u672c\u00a0<strong>DeepSeek-V3-0324<\/strong>\u00a0\u5df2\u6084\u7136\u4e0a\u7ebf Hugging Face \u5e73\u53f0\uff0c\u4f9b\u5f00\u53d1\u8005\u4e0b\u8f7d\u548c\u90e8\u7f72\u3002\u5c3d\u7ba1\u6b64\u6b21\u66f4\u65b0\u975e\u5e38\u4f4e\u8c03\uff0c\u4f46\u65b0\u6a21\u578b\u5728\u4ee3\u7801\u80fd\u529b\u65b9\u9762\u7684\u663e\u8457\u63d0\u5347\uff0c\u8fc5\u901f\u5f15\u53d1\u4e86\u6280\u672f\u793e\u533a\u7684\u9ad8\u5ea6\u5173\u6ce8\u548c\u70ed\u70c8\u8ba8\u8bba\u3002<\/p>\n<p>\u51e0\u4e2a\u5c0f\u65f6\u524d\uff0cDeepSeek-AI \u5f00\u6e90\u4e86 DeepSeekV3 \u7684\u66f4\u65b0\u7248\u672c\uff0c\u7248\u672c\u53f7 0324\uff0c\u4e8e 2025 \u5e74 3 \u6708 24 \u65e5\u4e0a\u4f20\u81f3 HuggingFace\uff0c\u5e76\u91c7\u7528 MIT \u534f\u8bae\u5f00\u6e90\u3002<\/p>\n<p>\u6a21\u578b\u914d\u7f6e\u4fe1\u606f\u663e\u793a\uff0cDeepSeekV3-0324 \u4f9d\u7136\u662f MoE \u5927\u6a21\u578b\uff0c\u5305\u542b 256 \u4e2a\u8def\u7531\u4e13\u5bb6\u548c 1 \u4e2a\u5171\u4eab\u4e13\u5bb6\uff0c\u6bcf\u4e2a <a href=\"https:\/\/www.kdjingpai.com\/en\/tokenization\/\">token<\/a> \u4f7f\u7528 8 \u4e2a\u4e13\u5bb6\u8fdb\u884c\u63a8\u7406\u3002 DeepSeekV3-0324 \u901a\u8fc7 RoPE \u53ef\u6269\u5c55\u81f3\u6700\u9ad8 163840 \u4e0a\u4e0b\u6587\u957f\u5ea6 (160K)\u3002 \u6a21\u578b\u8bcd\u6c47\u8868\u5927\u5c0f\u4e3a 129280 \u4e2a\uff0c\u5e76\u96c6\u6210\u4e86 LoRA \u673a\u5236\uff0c\u652f\u6301\u8f7b\u91cf\u7ea7\u5fae\u8c03\u3002<\/p>\n<p>\u4e0e 2024 \u5e74 12 \u6708 26 \u65e5\u53d1\u5e03\u7684 DeepSeekV3 \u76f8\u6bd4\uff0c\u8fd9\u4e9b\u53c2\u6570\u5747\u672a\u53d1\u751f\u53d8\u5316\uff0c\u8fd9\u610f\u5473\u7740\u6b64\u6b21\u66f4\u65b0\u5927\u6982\u7387\u662f\u539f\u6709\u6a21\u578b\u6301\u7eed\u8bad\u7ec3\u6216\u540e\u8bad\u7ec3\u7684\u7ed3\u679c\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/559e6761efde63d.jpeg\" alt=\"DeepSeek-V3-0324 \u4ee3\u7801\u80fd\u529b\u63d0\u5347\" width=\"575\" height=\"719\" \/><\/p>\n<p>&nbsp;<\/p>\n<h3>\u4ee3\u7801\u80fd\u529b\u7a81\u98de\u731b\u8fdb\uff0c\u76f4\u903c\u95ed\u6e90\u6807\u6746<\/h3>\n<p>\u7528\u6237\u7684\u7b2c\u4e00\u5370\u8c61\u548c\u591a\u65b9\u6d4b\u8bd5\u8868\u660e\uff0cDeepSeek-V3-0324 \u6700\u5f15\u4eba\u6ce8\u76ee\u7684\u6539\u8fdb\u5728\u4e8e\u5176\u4ee3\u7801\u751f\u6210\u548c\u7406\u89e3\u80fd\u529b\u3002\u8bb8\u591a\u6d4b\u8bd5\u8005\u6307\u51fa\uff0c\u5728\u6570\u5b66\u63a8\u7406\u548c\u524d\u7aef\u5f00\u53d1\u7b49\u9886\u57df\uff0c\u65b0\u7248\u672c\u7684\u6027\u80fd\u751a\u81f3\u8d85\u8d8a\u4e86 <a href=\"https:\/\/www.kdjingpai.com\/en\/claudeanquanfubai\/\">Claude<\/a> 3.5 \u548c Claude 3.7 Sonnet\u3002\u793e\u4ea4\u5a92\u4f53 X \u5e73\u53f0\u4e0a\u7684\u535a\u4e3b @KuittinenPetri \u66f4\u662f\u76f4\u8a00\uff0cDeepSeek-V3-0324 \u53ef\u4ee5\u8f7b\u677e\u4e14\u514d\u8d39\u5730\u521b\u5efa\u7cbe\u7f8e\u7684 HTML5, CSS \u548c\u524d\u7aef\u4ee3\u7801\uff0c\u8fd9\u5bf9 <a href=\"https:\/\/www.kdjingpai.com\/en\/claudeanquanfubai\/\">Anthropic<\/a> \u548c OpenAI \u6784\u6210\u4e86\u65b0\u7684\u6311\u6218\u3002<\/p>\n<p>\u4e3e\u4f8b\u6765\u8bf4\uff0c\u4ec5\u9700\u4e00\u4e2a\u7b80\u5355\u7684\u6307\u4ee4\uff0cDeepSeek-V3-0324 \u5c31\u80fd\u4e3a\u4e00\u4e2a\u540d\u4e3a NexusAI \u7684 AI \u516c\u53f8\u751f\u6210\u4e00\u4e2a\u5916\u89c2\u7cbe\u7f8e\u7684\u54cd\u5e94\u5f0f\u9996\u9875\uff0c\u5e76\u5c06\u6240\u6709\u5143\u7d20\u6574\u5408\u5230\u4e00\u4e2a HTML5 \u6587\u4ef6\u4e2d\u3002 \u751f\u6210\u7684\u4ee3\u7801\u957f\u8fbe 958 \u884c\uff0c\u5b9e\u73b0\u4e86\u4e00\u4e2a\u4ea4\u4e92\u5f0f\u4e14\u79fb\u52a8\u8bbe\u5907\u53cb\u597d\u7684\u7f51\u7ad9\uff0c\u751a\u81f3\u5305\u542b\u4e86\u6240\u9700\u7684\u56fe\u50cf\u8d44\u6e90\u3002 @KuittinenPetri \u8ba4\u4e3a\uff0cDeepSeek-V3-0324 \u662f <a href=\"https:\/\/www.kdjingpai.com\/en\/deepseek-chatshena\/\">DeepSeek<\/a> \u76ee\u524d\u6700\u597d\u7684\u975e\u63a8\u7406\u6a21\u578b\uff0c\u4e0d\u4ec5\u5728\u521b\u610f\u5199\u4f5c\u65b9\u9762\u8868\u73b0\u51fa\u8272\uff0c\u73b0\u5728\u751a\u81f3\u6bd4 R1 \u66f4\u64c5\u957f\u751f\u6210 HTML5 + CSS + \u524d\u7aef\u4ee3\u7801\u3002 \u53e6\u4e00\u4f4d\u7528\u6237\u4e5f\u6210\u529f\u5730\u8ba9 DeepSeek-V3-0324 \u521b\u5efa\u4e86\u4e00\u4e2a\u7f51\u7ad9\uff0c\u6a21\u578b\u4e00\u53e3\u6c14\u751f\u6210\u4e86 800 \u591a\u884c\u4ee3\u7801\uff0c\u4e14\u7f51\u7ad9\u5e03\u5c40\u76f8\u5f53\u6210\u529f\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/e7cdc1d88836a8f.jpeg\" alt=\"DeepSeek-V3-0324 \u4ee3\u7801\u751f\u6210\u793a\u4f8b\" width=\"745\" height=\"418\" \/><\/p>\n<p>&nbsp;<\/p>\n<h3>\u591a\u573a\u666f\u5b9e\u6d4b\uff0c\u5c55\u73b0\u5f3a\u5927\u7f16\u7a0b\u6f5c\u529b<\/h3>\n<p>\u4e3a\u4e86\u66f4\u5168\u9762\u5730\u8bc4\u4f30 DeepSeek-V3-0324 \u7684\u7f16\u7a0b\u80fd\u529b\uff0c\u8bb8\u591a\u7814\u7a76\u8005\u8fdb\u884c\u4e86\u4e0d\u540c\u573a\u666f\u7684\u6d4b\u8bd5\uff0c\u5e76\u5c06\u5176\u4e0e\u65e7\u7248 V3\u3001Claude 3.7 \u548c o1 pro \u7b49\u6a21\u578b\u8fdb\u884c\u4e86\u5bf9\u6bd4\u3002 \u6d4b\u8bd5\u7ed3\u679c\u8bc1\u660e\u4e86\u65b0\u7248 V3 \u5728\u7f16\u7a0b\u65b9\u9762\u7684\u663e\u8457\u8fdb\u6b65\u3002<\/p>\n<ul>\n<li><strong>\u6587\u5b57\u8f6c\u53ef\u89c6\u5316\u7f51\u9875\uff1a<\/strong>\u00a0\u5728\u5c06\u6587\u5b57\u63cf\u8ff0\u8f6c\u6362\u4e3a\u4ea4\u4e92\u5f0f\u7f51\u9875\u7684\u6d4b\u8bd5\u4e2d\uff0cDeepSeek-V3-0324 \u76f8\u6bd4\u65e7\u7248 V3 \u8868\u73b0\u51fa\u8d28\u7684\u98de\u8dc3\u3002 \u65b0\u7248\u672c\u751f\u6210\u7684\u7f51\u9875\u4e0d\u4ec5\u5185\u5bb9\u66f4\u52a0\u4e30\u5bcc\uff0c\u7528\u6237\u754c\u9762\u8bbe\u8ba1\u548c\u6392\u7248\u4e5f\u5f97\u5230\u663e\u8457\u6539\u8fdb\uff0c\u5176\u63d0\u5347\u5e45\u5ea6\u751a\u81f3\u8d85\u8d8a\u4e86 Claude 3.5 \u5230 3.7 \u7684\u5347\u7ea7\u3002 \u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0cDeepSeek-V3-0324 \u80fd\u591f\u6839\u636e\u8be6\u7ec6\u7684\u63d0\u793a\uff0c\u5c06 PDF \u6587\u4ef6\u5185\u5bb9\u8f6c\u6362\u4e3a\u7f8e\u89c2\u7684\u4e2d\u6587\u53ef\u89c6\u5316\u7f51\u9875\uff0c\u800c\u8fd9\u901a\u5e38\u88ab\u8ba4\u4e3a\u662f Claude 3.7 \u7684\u4f18\u52bf\u9886\u57df\u3002<\/li>\n<li><strong>\u751f\u6210 3D \u52a8\u753b\uff1a<\/strong>\u00a0\u5728\u901a\u8fc7 JS \u4ee3\u7801\u5e93\u751f\u6210\u53ef\u4ea4\u4e92\u7684 3D \u6f14\u793a\u6548\u679c\u7684\u6d4b\u8bd5\u4e2d\uff0c\u65b0\u7248 V3 \u80fd\u591f\u5c06\u5de7\u514b\u529b\u5236\u4f5c\u7684\u6bcf\u4e2a\u6b65\u9aa4\u5efa\u6a21\u51fa\u6765\uff0c\u5e76\u652f\u6301\u6807\u7b7e\u4e92\u52a8\u548c\u4fa7\u8fb9\u680f\u3002 \u867d\u7136\u4e0e Claude \u76f8\u6bd4\u4ecd\u6709\u63d0\u5347\u7a7a\u95f4\uff0c\u4f46\u5df2\u7ecf\u663e\u8457\u8d85\u8d8a\u65e7\u7248 V3 \u7684\u80fd\u529b\u3002<\/li>\n<li><strong>UI \u7ec4\u4ef6\u8bbe\u8ba1\uff1a<\/strong>\u00a0\u5728\u5929\u6c14\u9884\u62a5 UI \u7ec4\u4ef6\u8bbe\u8ba1\u6d4b\u8bd5\u4e2d\uff0cV3-0324 \u5728\u52a8\u753b\u8868\u73b0\u548c\u5929\u6c14\u6587\u5b57\u6807\u6ce8\u7684\u51c6\u786e\u6027\u4e0a\u5747\u6709\u63d0\u5347\uff0c\u8868\u660e\u5176\u5728\u751f\u6210\u5b9e\u7528\u6027\u7528\u6237\u754c\u9762\u65b9\u9762\u5177\u5907\u66f4\u5f3a\u7684\u5b9e\u529b\u3002<\/li>\n<li><strong>\u7269\u7406\u4e16\u754c\u6a21\u62df\uff1a<\/strong>\u00a0\u5728\u6a21\u62df\u5c0f\u7403\u5728\u4e00\u4e2a\u65cb\u8f6c\u7684\u516d\u8fb9\u5f62\u5185\u5f39\u8df3\u7684\u6d4b\u8bd5\u4e2d\uff0cDeepSeek-V3-0324 \u7cbe\u786e\u5730\u5b9e\u73b0\u4e86\u5c0f\u7403\u7684\u78b0\u649e\u6548\u679c\u3002 \u867d\u7136\u4ecd\u5b58\u5728\u4e00\u4e9b\u7455\u75b5\uff0c\u4f46\u6574\u4f53\u8868\u73b0\u4f18\u4e8e\u65e7\u7248 V3\uff0c\u5e76\u80fd\u4e0e o1 pro \u76f8\u63d0\u5e76\u8bba\u3002<\/li>\n<li><strong>AI \u6e38\u620f\u751f\u6210\uff1a<\/strong>\u00a0\u6700\u5f15\u4eba\u6ce8\u76ee\u7684\u662f\uff0cDeepSeek-V3-0324 \u4ec5\u51ed\u4e00\u53e5\u8bdd\u6307\u4ee4\u5c31\u751f\u6210\u4e86\u53ef\u73a9\u7684\u50cf\u7d20\u8d2a\u5403\u86c7\u6e38\u620f\uff0c\u5e76\u5e26\u6709\u97f3\u6548\u548c AI \u8f85\u52a9\u6a21\u5f0f\u3002 \u5c3d\u7ba1\u5728\u590d\u6742\u5ea6\u548c\u5b8c\u7f8e\u5ea6\u4e0a\u53ef\u80fd\u4e0e Claude 3.7 \u7684 extend thinking \u6a21\u5f0f\u5b58\u5728\u5dee\u8ddd\uff0c\u4f46\u80fd\u591f\u5b8c\u6210\u4e00\u4e2a\u529f\u80fd\u5b8c\u6574\u7684\u6e38\u620f\u5df2\u7ecf\u5145\u5206\u5c55\u73b0\u4e86\u5176\u5f3a\u5927\u7684\u7f16\u7a0b\u80fd\u529b\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h3>\u6280\u672f\u7279\u6027\u4e0e\u6210\u672c\u4f18\u52bf<\/h3>\n<p>DeepSeek-V3-0324 \u5c1a\u672a\u516c\u5e03\u8be6\u7ec6\u7684\u6a21\u578b\u5361\uff0c\u4f46\u5df2\u77e5\u5176\u53c2\u6570\u89c4\u6a21\u4e3a 6850 \u4ebf\u3002 \u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0cDeepSeek V3 \u91c7\u7528\u7684\u662f\u00a0<strong>\u6df7\u5408\u4e13\u5bb6\u6a21\u578b (MoE)<\/strong>\u00a0\u67b6\u6784\uff0c\u62e5\u6709 6710 \u4ebf\u53c2\u6570\uff0c\u5176\u4e2d\u6bcf\u6b21\u63a8\u7406\u4ec5\u6fc0\u6d3b 370 \u4ebf\u53c2\u6570\u3002\u00a0<em>\uff08\u7f16\u8005\u6ce8\uff1aMoE \u6a21\u578b\u901a\u8fc7\u5c06\u5927\u578b\u6a21\u578b\u5206\u89e3\u4e3a\u591a\u4e2a\u201c\u4e13\u5bb6\u201d\u5b50\u7f51\u7edc\uff0c\u4ece\u800c\u5728\u4fdd\u8bc1\u6a21\u578b\u6027\u80fd\u7684\u540c\u65f6\uff0c\u663e\u8457\u964d\u4f4e\u8ba1\u7b97\u6210\u672c\u548c\u5ef6\u8fdf\u3002\uff09<\/em>\u00a0\u4e3a\u4e86\u89e3\u51b3\u4f20\u7edf MoE \u6a21\u578b\u4e2d\u4e13\u5bb6\u8d1f\u8f7d\u4e0d\u5747\u8861\u7684\u95ee\u9898\uff0cDeepSeek \u5728 V3 \u4e2d\u521b\u65b0\u6027\u5730\u63d0\u51fa\u4e86\u00a0<strong>\u8f85\u52a9\u635f\u5931\u514d\u8d39\u7684\u8d1f\u8f7d\u5747\u8861\u7b56\u7565<\/strong>\u00a0\uff0c\u901a\u8fc7\u5f15\u5165\u201c\u504f\u5dee\u9879\u201d\u52a8\u6001\u8c03\u6574\u4e13\u5bb6\u8d1f\u8f7d\uff0c\u4ece\u800c\u63d0\u5347\u6a21\u578b\u6027\u80fd\u548c\u8bad\u7ec3\u6548\u7387\u3002 \u6b64\u5916\uff0cV3 \u8fd8\u91c7\u7528\u4e86\u00a0<strong>\u8282\u70b9\u53d7\u9650\u7684\u8def\u7531\u673a\u5236<\/strong>\u00a0\uff0c\u4ee5\u51cf\u5c11\u5927\u89c4\u6a21\u5206\u5e03\u5f0f\u8bad\u7ec3\u4e2d\u7684\u901a\u4fe1\u6210\u672c\u3002<\/p>\n<p>\u9664\u4e86\u5f3a\u5927\u7684\u6027\u80fd\uff0cDeepSeek-V3-0324 \u5ef6\u7eed\u4e86\u5bbd\u677e\u7684 MIT \u5f00\u6e90\u534f\u8bae\u3002 \u66f4\u4e3a\u5173\u952e\u7684\u662f\uff0c\u5176 API \u4ef7\u683c\u6781\u5177\u7ade\u4e89\u529b\uff0c\u76f8\u8f83\u4e8e OpenAI \u7684\u00a0<code>o1-pro<\/code>\u00a0\u4fbf\u5b9c\u81f3\u5c11 50 \u500d\u3002 \u4e0e Claude 3.7 \u76f8\u6bd4\uff0cDeepSeek v3 \u7684\u8f93\u5165\u4ef7\u683c\u7ea6\u4e3a\u5176\u5341\u5206\u4e4b\u4e00\uff0c\u800c\u8f93\u51fa\u4ef7\u683c\u5728\u6807\u51c6\u65f6\u6bb5\u7ea6\u4e3a\u5176\u5341\u4e09\u5206\u4e4b\u4e00\uff0c\u4f18\u60e0\u65f6\u6bb5\u751a\u81f3\u8fbe\u5230\u4e8c\u5341\u4e03\u5206\u4e4b\u4e00\u3002 \u8fd9\u79cd\u6781\u5177\u5438\u5f15\u529b\u7684\u4ef7\u683c\u4f18\u52bf\uff0c\u7ed3\u5408\u5176\u5f00\u6e90\u7279\u6027\uff0c\u65e0\u7591\u5c06\u4e3a AI \u7f16\u7a0b\u7684\u666e\u53ca\u548c\u53d1\u5c55\u63d0\u4f9b\u5f3a\u5927\u52a8\u529b\u3002<\/p>\n<p>&nbsp;<\/p>\n<h3>DeepSeek-V3-0324 \u6a21\u578b\u7279\u6027<\/h3>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"aligncenter  wp-image-29365\" title=\"\u6df1\u5ea6\u6c42\u7d22 V3 \u6a21\u578b\u4f4e\u8c03\u66f4\u65b0\uff0c\u4ee3\u7801\u80fd\u529b\u8dc3\u5347\uff0c\u5251\u6307 OpenAI \u4e0e Anthropic-1\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/1383396337066c6.png\" alt=\"\u6df1\u5ea6\u6c42\u7d22 V3 \u6a21\u578b\u4f4e\u8c03\u66f4\u65b0\uff0c\u4ee3\u7801\u80fd\u529b\u8dc3\u5347\uff0c\u5251\u6307 OpenAI \u4e0e Anthropic-1\" width=\"726\" height=\"343\" srcset=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/1383396337066c6.png 4154w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/1383396337066c6-768x363.png 768w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/1383396337066c6-1536x727.png 1536w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/1383396337066c6-2048x969.png 2048w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/1383396337066c6-18x9.png 18w\" sizes=\"auto, (max-width: 726px) 100vw, 726px\" \/><\/p>\n<p>DeepSeek-V3-0324 \u5728\u591a\u4e2a\u5173\u952e\u65b9\u9762\u5c55\u73b0\u4e86\u76f8\u6bd4\u524d\u4ee3 DeepSeek-V3 \u7684\u663e\u8457\u6539\u8fdb\u3002<\/p>\n<ul>\n<li><strong>\u63a8\u7406\u80fd\u529b\u63d0\u5347\uff1a<\/strong>\n<ul>\n<li>MMLU-Pro: 75.9 \u2192 81.2 (+5.3)<\/li>\n<li>GPQA: 59.1 \u2192 68.4 (+9.3)<\/li>\n<li>AIME: 39.6 \u2192 59.4 (+19.8)<\/li>\n<li>LiveCodeBench: 39.2 \u2192 49.2 (+10.0)<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u524d\u7aef Web \u5f00\u53d1\u80fd\u529b\u589e\u5f3a\uff1a<\/strong>\n<ul>\n<li>\u4ee3\u7801\u53ef\u6267\u884c\u6027\u63d0\u5347<\/li>\n<li>\u7f51\u9875\u548c\u6e38\u620f\u524d\u7aef\u5728\u7f8e\u5b66\u4e0a\u66f4\u4ee4\u4eba\u6109\u60a6<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u4e2d\u6587\u5199\u4f5c\u80fd\u529b\u63d0\u5347\uff1a<\/strong>\n<ul>\n<li>\u98ce\u683c\u548c\u5185\u5bb9\u8d28\u91cf\u63d0\u5347\uff1a\n<ul>\n<li>\u66f4\u63a5\u8fd1 R1 \u5199\u4f5c\u98ce\u683c<\/li>\n<li>\u4e2d\u957f\u6587\u5199\u4f5c\u8d28\u91cf\u66f4\u9ad8<\/li>\n<\/ul>\n<\/li>\n<li>\u529f\u80fd\u589e\u5f3a\n<ul>\n<li>\u591a\u8f6e\u4ea4\u4e92\u5f0f\u6539\u5199\u80fd\u529b\u5f97\u5230\u6539\u8fdb<\/li>\n<li>\u7ffb\u8bd1\u8d28\u91cf\u548c\u4e66\u4fe1\u8d28\u91cf\u5f97\u5230\u4f18\u5316<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u4e2d\u6587\u641c\u7d22\u80fd\u529b\u63d0\u5347\uff1a<\/strong>\n<ul>\n<li>\u62a5\u544a\u5206\u6790\u8bf7\u6c42\u8f93\u51fa\u66f4\u52a0\u8be6\u7ec6<\/li>\n<\/ul>\n<\/li>\n<li><strong>Function Calling \u529f\u80fd\u6539\u8fdb\uff1a<\/strong>\n<ul>\n<li><a href=\"https:\/\/www.kdjingpai.com\/en\/hanshudiaoyongfunct\/\">Function Calling<\/a> \u51c6\u786e\u7387\u63d0\u9ad8\uff0c\u4fee\u590d\u4e86 V3 \u7248\u672c\u7684\u9057\u7559\u95ee\u9898<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h3>\u4f7f\u7528\u5efa\u8bae<\/h3>\n<p><strong>\u7cfb\u7edf\u63d0\u793a (System Prompt)<\/strong><\/p>\n<p>DeepSeek \u5b98\u65b9 web\/app \u4e2d\u4f7f\u7528\u4e86\u76f8\u540c\u7684\u7cfb\u7edf\u63d0\u793a\uff0c\u5e76\u5e26\u6709\u5177\u4f53\u65e5\u671f\u3002<\/p>\n<pre><code>\u8be5\u52a9\u624b\u4e3aDeepSeek Chat\uff0c\u7531\u6df1\u5ea6\u6c42\u7d22\u516c\u53f8\u521b\u9020\u3002\r\n\u4eca\u5929\u662f{current date}\u3002\r\n<\/code><\/pre>\n<p>\u4f8b\u5982\uff1a<\/p>\n<pre><code>\u8be5\u52a9\u624b\u4e3aDeepSeek Chat\uff0c\u7531\u6df1\u5ea6\u6c42\u7d22\u516c\u53f8\u521b\u9020\u3002\r\n\u4eca\u5929\u662f3\u670824\u65e5\uff0c\u661f\u671f\u4e00\u3002\r\n<\/code><\/pre>\n<p><strong>Temperature \u53c2\u6570\u8bbe\u7f6e<\/strong><\/p>\n<p>\u5728 DeepSeek web \u548c\u5e94\u7528\u73af\u5883\u4e2d\uff0ctemperature \u53c2\u6570 (<i>T<sub>model<\/sub><\/i>) \u88ab\u8bbe\u7f6e\u4e3a 0.3\u3002 \u8003\u8651\u5230\u8bb8\u591a\u7528\u6237\u5728 API \u8c03\u7528\u4e2d\u4f7f\u7528\u9ed8\u8ba4 temperature 1.0\uff0cDeepSeek \u5b9e\u65bd\u4e86 API temperature (<i>T<sub>api<\/sub><\/i>) \u6620\u5c04\u673a\u5236\uff0c\u5c06 API \u8f93\u5165 temperature \u503c 1.0 \u8c03\u6574\u4e3a\u6700\u5408\u9002\u7684\u6a21\u578b temperature \u8bbe\u7f6e 0.3\u3002<\/p>\n<p><i>T<sub>model<\/sub><\/i>\u00a0=\u00a0<i>T<sub>api<\/sub><\/i>\u00a0\u00d7 0.3 (0 \u2264\u00a0<i>T<sub>api<\/sub><\/i>\u00a0\u2264 1)<\/p>\n<p><i>T<sub>model<\/sub><\/i>\u00a0=\u00a0<i>T<sub>api<\/sub><\/i>\u00a0&#8211; 0.7 (1 &lt;\u00a0<i>T<sub>api<\/sub><\/i>\u00a0\u2264 2)<\/p>\n<p>\u56e0\u6b64\uff0c\u5982\u679c\u60a8\u901a\u8fc7 API \u8c03\u7528 V3\uff0ctemperature 1.0 \u76f8\u5f53\u4e8e\u6a21\u578b temperature 0.3\u3002<\/p>\n<p><strong>\u6587\u4ef6\u4e0a\u4f20\u548c Web \u641c\u7d22\u7684 Prompts<\/strong><\/p>\n<p>\u5bf9\u4e8e\u6587\u4ef6\u4e0a\u4f20\uff0c\u8bf7\u9075\u5faa\u4ee5\u4e0b\u6a21\u677f\u521b\u5efa prompts\uff0c\u5176\u4e2d\u00a0<code>{file_name}<\/code>\u3001<code>{file_content}<\/code>\u00a0\u548c\u00a0<code>{question}<\/code>\u00a0\u4e3a\u53c2\u6570\u3002<\/p>\n<pre><code>file_template = \\\r\n\"\"\"[file name]: {file_name}\r\n[file content begin]\r\n{file_content}\r\n[file content end]\r\n{question}\"\"\"\r\n<\/code><\/pre>\n<p>\u5bf9\u4e8e Web \u641c\u7d22\uff0c<code>{search_results}<\/code>\u3001<code>{cur_date}<\/code>\u00a0\u548c\u00a0<code>{question}<\/code>\u00a0\u4e3a\u53c2\u6570\u3002<\/p>\n<p>\u4e2d\u6587\u67e5\u8be2 Prompt\uff1a<\/p>\n<pre><code>search_answer_zh_template = \\\r\n'''# \u4ee5\u4e0b\u5185\u5bb9\u662f\u57fa\u4e8e\u7528\u6237\u53d1\u9001\u7684\u6d88\u606f\u7684\u641c\u7d22\u7ed3\u679c:\r\n{search_results}\r\n\u5728\u6211\u7ed9\u4f60\u7684\u641c\u7d22\u7ed3\u679c\u4e2d\uff0c\u6bcf\u4e2a\u7ed3\u679c\u90fd\u662f[webpage X begin]...[webpage X end]\u683c\u5f0f\u7684\uff0cX\u4ee3\u8868\u6bcf\u7bc7\u6587\u7ae0\u7684\u6570\u5b57\u7d22\u5f15\u3002\u8bf7\u5728\u9002\u5f53\u7684\u60c5\u51b5\u4e0b\u5728\u53e5\u5b50\u672b\u5c3e\u5f15\u7528\u4e0a\u4e0b\u6587\u3002\u8bf7\u6309\u7167\u5f15\u7528\u7f16\u53f7[citation:X]\u7684\u683c\u5f0f\u5728\u7b54\u6848\u4e2d\u5bf9\u5e94\u90e8\u5206\u5f15\u7528\u4e0a\u4e0b\u6587\u3002\u5982\u679c\u4e00\u53e5\u8bdd\u6e90\u81ea\u591a\u4e2a\u4e0a\u4e0b\u6587\uff0c\u8bf7\u5217\u51fa\u6240\u6709\u76f8\u5173\u7684\u5f15\u7528\u7f16\u53f7\uff0c\u4f8b\u5982[citation:3][citation:5]\uff0c\u5207\u8bb0\u4e0d\u8981\u5c06\u5f15\u7528\u96c6\u4e2d\u5728\u6700\u540e\u8fd4\u56de\u5f15\u7528\u7f16\u53f7\uff0c\u800c\u662f\u5728\u7b54\u6848\u5bf9\u5e94\u90e8\u5206\u5217\u51fa\u3002\r\n\u5728\u56de\u7b54\u65f6\uff0c\u8bf7\u6ce8\u610f\u4ee5\u4e0b\u51e0\u70b9\uff1a\r\n- \u4eca\u5929\u662f{cur_date}\u3002\r\n- \u5e76\u975e\u641c\u7d22\u7ed3\u679c\u7684\u6240\u6709\u5185\u5bb9\u90fd\u4e0e\u7528\u6237\u7684\u95ee\u9898\u5bc6\u5207\u76f8\u5173\uff0c\u4f60\u9700\u8981\u7ed3\u5408\u95ee\u9898\uff0c\u5bf9\u641c\u7d22\u7ed3\u679c\u8fdb\u884c\u7504\u522b\u3001\u7b5b\u9009\u3002\r\n- \u5bf9\u4e8e\u5217\u4e3e\u7c7b\u7684\u95ee\u9898\uff08\u5982\u5217\u4e3e\u6240\u6709\u822a\u73ed\u4fe1\u606f\uff09\uff0c\u5c3d\u91cf\u5c06\u7b54\u6848\u63a7\u5236\u572810\u4e2a\u8981\u70b9\u4ee5\u5185\uff0c\u5e76\u544a\u8bc9\u7528\u6237\u53ef\u4ee5\u67e5\u770b\u641c\u7d22\u6765\u6e90\u3001\u83b7\u5f97\u5b8c\u6574\u4fe1\u606f\u3002\u4f18\u5148\u63d0\u4f9b\u4fe1\u606f\u5b8c\u6574\u3001\u6700\u76f8\u5173\u7684\u5217\u4e3e\u9879\uff1b\u5982\u975e\u5fc5\u8981\uff0c\u4e0d\u8981\u4e3b\u52a8\u544a\u8bc9\u7528\u6237\u641c\u7d22\u7ed3\u679c\u672a\u63d0\u4f9b\u7684\u5185\u5bb9\u3002\r\n- \u5bf9\u4e8e\u521b\u4f5c\u7c7b\u7684\u95ee\u9898\uff08\u5982\u5199\u8bba\u6587\uff09\uff0c\u8bf7\u52a1\u5fc5\u5728\u6b63\u6587\u7684\u6bb5\u843d\u4e2d\u5f15\u7528\u5bf9\u5e94\u7684\u53c2\u8003\u7f16\u53f7\uff0c\u4f8b\u5982[citation:3][citation:5]\uff0c\u4e0d\u80fd\u53ea\u5728\u6587\u7ae0\u672b\u5c3e\u5f15\u7528\u3002\u4f60\u9700\u8981\u89e3\u8bfb\u5e76\u6982\u62ec\u7528\u6237\u7684\u9898\u76ee\u8981\u6c42\uff0c\u9009\u62e9\u5408\u9002\u7684\u683c\u5f0f\uff0c\u5145\u5206\u5229\u7528\u641c\u7d22\u7ed3\u679c\u5e76\u62bd\u53d6\u91cd\u8981\u4fe1\u606f\uff0c\u751f\u6210\u7b26\u5408\u7528\u6237\u8981\u6c42\u3001\u6781\u5177\u601d\u60f3\u6df1\u5ea6\u3001\u5bcc\u6709\u521b\u9020\u529b\u4e0e\u4e13\u4e1a\u6027\u7684\u7b54\u6848\u3002\u4f60\u7684\u521b\u4f5c\u7bc7\u5e45\u9700\u8981\u5c3d\u53ef\u80fd\u5ef6\u957f\uff0c\u5bf9\u4e8e\u6bcf\u4e00\u4e2a\u8981\u70b9\u7684\u8bba\u8ff0\u8981\u63a8\u6d4b\u7528\u6237\u7684\u610f\u56fe\uff0c\u7ed9\u51fa\u5c3d\u53ef\u80fd\u591a\u89d2\u5ea6\u7684\u56de\u7b54\u8981\u70b9\uff0c\u4e14\u52a1\u5fc5\u4fe1\u606f\u91cf\u5927\u3001\u8bba\u8ff0\u8be6\u5c3d\u3002\r\n- \u5982\u679c\u56de\u7b54\u5f88\u957f\uff0c\u8bf7\u5c3d\u91cf\u7ed3\u6784\u5316\u3001\u5206\u6bb5\u843d\u603b\u7ed3\u3002\u5982\u679c\u9700\u8981\u5206\u70b9\u4f5c\u7b54\uff0c\u5c3d\u91cf\u63a7\u5236\u57285\u4e2a\u70b9\u4ee5\u5185\uff0c\u5e76\u5408\u5e76\u76f8\u5173\u7684\u5185\u5bb9\u3002\r\n- \u5bf9\u4e8e\u5ba2\u89c2\u7c7b\u7684\u95ee\u7b54\uff0c\u5982\u679c\u95ee\u9898\u7684\u7b54\u6848\u975e\u5e38\u7b80\u77ed\uff0c\u53ef\u4ee5\u9002\u5f53\u8865\u5145\u4e00\u5230\u4e24\u53e5\u76f8\u5173\u4fe1\u606f\uff0c\u4ee5\u4e30\u5bcc\u5185\u5bb9\u3002\r\n- \u4f60\u9700\u8981\u6839\u636e\u7528\u6237\u8981\u6c42\u548c\u56de\u7b54\u5185\u5bb9\u9009\u62e9\u5408\u9002\u3001\u7f8e\u89c2\u7684\u56de\u7b54\u683c\u5f0f\uff0c\u786e\u4fdd\u53ef\u8bfb\u6027\u5f3a\u3002\r\n- \u4f60\u7684\u56de\u7b54\u5e94\u8be5\u7efc\u5408\u591a\u4e2a\u76f8\u5173\u7f51\u9875\u6765\u56de\u7b54\uff0c\u4e0d\u80fd\u91cd\u590d\u5f15\u7528\u4e00\u4e2a\u7f51\u9875\u3002\r\n- \u9664\u975e\u7528\u6237\u8981\u6c42\uff0c\u5426\u5219\u4f60\u56de\u7b54\u7684\u8bed\u8a00\u9700\u8981\u548c\u7528\u6237\u63d0\u95ee\u7684\u8bed\u8a00\u4fdd\u6301\u4e00\u81f4\u3002\r\n# \u7528\u6237\u6d88\u606f\u4e3a\uff1a\r\n{question}'''\r\n<\/code><\/pre>\n<p>\u82f1\u6587\u67e5\u8be2 Prompt\uff1a<\/p>\n<pre><code>search_answer_en_template = \\\r\n'''# The following contents are the search results related to the user's message:\r\n{search_results}\r\nIn the search results I provide to you, each result is formatted as [webpage X begin]...[webpage X end], where X represents the numerical index of each article. Please cite the context at the end of the relevant sentence when appropriate. Use the citation format [citation:X] in the corresponding part of your answer. If a sentence is derived from multiple contexts, list all relevant citation numbers, such as [citation:3][citation:5]. Be sure not to cluster all citations at the end; instead, include them in the corresponding parts of the answer.\r\nWhen responding, please keep the following points in mind:\r\n- Today is {cur_date}.\r\n- Not all content in the search results is closely related to the user's question. You need to evaluate and filter the search results based on the question.\r\n- For listing-type questions (e.g., listing all flight information), try to limit the answer to 10 key points and inform the user that they can refer to the search sources for complete information. Prioritize providing the most complete and relevant items in the list. Avoid mentioning content not provided in the search results unless necessary.\r\n- For creative tasks (e.g., writing an essay), ensure that references are cited within the body of the text, such as [citation:3][citation:5], rather than only at the end of the text. You need to interpret and summarize the user's requirements, choose an appropriate format, fully utilize the search results, extract key information, and generate an answer that is insightful, creative, and professional. Extend the length of your response as much as possible, addressing each point in detail and from multiple perspectives, ensuring the content is rich and thorough.\r\n- If the response is lengthy, structure it well and summarize it in paragraphs. If a point-by-point format is needed, try to limit it to 5 points and merge related content.\r\n- For objective Q&amp;A, if the answer is very brief, you may add one or two related sentences to enrich the content.\r\n- Choose an appropriate and visually appealing format for your response based on the user's requirements and the content of the answer, ensuring strong readability.\r\n- Your answer should synthesize information from multiple relevant webpages and avoid repeatedly citing the same webpage.\r\n- Unless the user requests otherwise, your response should be in the same language as the user's question.\r\n# The user's message is:\r\n{question}'''\r\n<\/code><\/pre>\n<p><strong>\u672c\u5730\u8fd0\u884c\u65b9\u6cd5<\/strong><\/p>\n<p>DeepSeek-V3-0324 \u7684\u6a21\u578b\u7ed3\u6784\u4e0e DeepSeek-V3 \u5b8c\u5168\u4e00\u81f4\u3002 \u6709\u5173\u672c\u5730\u8fd0\u884c\u6b64\u6a21\u578b\u7684\u66f4\u591a\u4fe1\u606f\uff0c\u8bf7\u8bbf\u95ee\u00a0<a href=\"https:\/\/github.com\/deepseek-ai\/DeepSeek-V3\">DeepSeek-V3<\/a>\u00a0\u4ee3\u7801\u4ed3\u5e93\u3002<\/p>\n<p><strong>\u8be5\u6a21\u578b\u652f\u6301 Function Calling\u3001JSON \u8f93\u51fa\u548c FIM completion \u7b49\u529f\u80fd\u3002 \u6709\u5173\u5982\u4f55\u6784\u5efa prompts \u4ee5\u4f7f\u7528\u8fd9\u4e9b\u529f\u80fd\u7684\u8bf4\u660e\uff0c\u8bf7\u53c2\u9605\u00a0<a href=\"https:\/\/huggingface.co\/deepseek-ai\/DeepSeek-V2.5#function-calling\">DeepSeek-V2.5<\/a>\u00a0\u4ee3\u7801\u4ed3\u5e93\u3002<\/strong><\/p>\n<p>&nbsp;<\/p>\n<p>DeepSeek-V3-0324 \u7684\u4f4e\u8c03\u66f4\u65b0\uff0c\u5374\u5728\u79d1\u6280\u754c\u5f15\u8d77\u4e86\u5e7f\u6cdb\u5173\u6ce8\u3002 \u5b83\u5728\u4ee3\u7801\u80fd\u529b\u65b9\u9762\u53d6\u5f97\u4e86\u4ee4\u4eba\u77a9\u76ee\u7684\u8fdb\u6b65\uff0c\u4e0d\u4ec5\u5728\u591a\u9879\u7f16\u7a0b\u4efb\u52a1\u4e2d\u5c55\u73b0\u51fa\u5f3a\u5927\u7684\u5b9e\u529b\uff0c\u66f4\u5728\u67d0\u4e9b\u65b9\u9762\u80fd\u591f\u4e0e Claude 3.5\/3.7 Sonnet \u7b49\u9876\u5c16\u6a21\u578b\u76f8\u5ab2\u7f8e\u3002 \u5176\u5f00\u6e90\u3001\u9ad8\u6548\u4e14\u9ad8\u6027\u4ef7\u6bd4\u7684\u7279\u70b9\u9884\u793a\u7740\uff0c\u00a0<strong>AI \u7f16\u7a0b\u7684\u666e\u60e0\u65f6\u4ee3\u6216\u5c06\u52a0\u901f\u5230\u6765<\/strong>\u00a0\u3002 \u968f\u7740\u66f4\u591a\u7b2c\u4e09\u65b9\u5e73\u53f0\u63a5\u5165 DeepSeek \u65b0\u7248 V3\uff0c\u5f00\u53d1\u8005\u548c\u666e\u901a\u7528\u6237\u90fd\u5c06\u80fd\u591f\u4ee5\u66f4\u4f4e\u7684\u6210\u672c\u4f53\u9a8c\u5230\u5148\u8fdb\u7684 AI \u7f16\u7a0b\u80fd\u529b\u3002 \u8fd9\u65e0\u7591\u5c06\u4e3a\u6574\u4e2a AI \u751f\u6001\u6ce8\u5165\u65b0\u7684\u6d3b\u529b\uff0c\u5e76\u63a8\u52a8\u66f4\u591a\u521b\u65b0\u5e94\u7528\u7684\u6d8c\u73b0\u3002 \u5728\u62e5\u6709\u5f3a\u5927\u7684\u4ee3\u7801\u80fd\u529b V3 \u548c\u9876\u7ea7\u7684\u63a8\u7406\u80fd\u529b R1 \u4e4b\u540e\uff0cDeepSeek \u672a\u6765\u7684 R2 \u6a21\u578b\uff0c\u66f4\u503c\u5f97\u6211\u4eec\u62ed\u76ee\u4ee5\u5f85\u3002<\/p>\n<p>DeepSeekV3 \u7684\u6b64\u6b21\u66f4\u65b0\uff0c\u518d\u6b21\u8bc1\u660e\u4e86\u4e2d\u56fd AI \u6280\u672f\u7684\u5feb\u901f\u53d1\u5c55\u548c\u8ffd\u8d76\u52bf\u5934\u3002 DeepSeek-V3-0324 \u7684\u5f00\u6e90\u548c\u514d\u8d39\u5546\u7528\u6388\u6743\u7b56\u7565\uff0c\u65e0\u7591\u5c06\u5438\u5f15\u66f4\u591a\u7684\u5f00\u53d1\u8005\u548c\u4f01\u4e1a\u52a0\u5165\u5230 AI \u5e94\u7528\u5f00\u53d1\u7684\u884c\u5217\u4e2d\uff0c\u5171\u540c\u63a8\u52a8 AI \u6280\u672f\u7684\u8fdb\u6b65\u548c\u666e\u53ca\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u79d1\u6280\u9886\u57df\u7684\u7ade\u4e89\u603b\u662f\u6697\u6d41\u6d8c\u52a8\u3002\u8fd1\u65e5\uff0c\u4e2d\u56fd AI \u521b\u4f01 DeepSeek\uff08\u6df1\u5ea6\u6c42\u7d22\uff09\u56e2\u961f\u5728\u672a\u8fdb\u884c\u5927\u89c4\u6a21\u5ba3\u4f20\u7684\u60c5\u51b5\u4e0b\uff0c\u4f4e\u8c03\u5730\u66f4\u65b0\u4e86\u5176 V3 \u57fa\u5ea7\u6a21\u578b\uff0c\u65b0\u7248\u672c\u00a0DeepSeek-V3-0324\u00a0\u5df2\u6084\u7136\u4e0a\u7ebf Hugging Face \u5e73\u53f0\uff0c\u4f9b\u5f00\u53d1\u8005\u4e0b&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[46],"tags":[],"class_list":["post-29364","post","type-post","status-publish","format-standard","hentry","category-news"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/29364","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=29364"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/29364\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media?parent=29364"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/categories?post=29364"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/tags?post=29364"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}