{"id":940,"date":"2025-08-25T09:24:00","date_gmt":"2025-08-25T01:24:00","guid":{"rendered":"http:\/\/www.applicationsofllm.com\/?p=940"},"modified":"2025-09-17T09:27:38","modified_gmt":"2025-09-17T01:27:38","slug":"meta%e5%bc%80%e6%ba%90mobilellm-r1%e6%a8%a1%e5%9e%8b%ef%bc%8c%e4%b8%8d%e5%88%b01b%e5%8f%82%e6%95%b0%ef%bc%8c%e7%94%a81-10%e7%9a%84%e8%ae%ad%e7%bb%83%e5%b0%b1%e8%b6%85%e8%b6%8a%e4%ba%86qwen3-2","status":"publish","type":"post","link":"http:\/\/www.applicationsofllm.com\/index.php\/2025\/08\/25\/meta%e5%bc%80%e6%ba%90mobilellm-r1%e6%a8%a1%e5%9e%8b%ef%bc%8c%e4%b8%8d%e5%88%b01b%e5%8f%82%e6%95%b0%ef%bc%8c%e7%94%a81-10%e7%9a%84%e8%ae%ad%e7%bb%83%e5%b0%b1%e8%b6%85%e8%b6%8a%e4%ba%86qwen3-2\/","title":{"rendered":"Meta\u5f00\u6e90MobileLLM-R1\u6a21\u578b\uff0c\u4e0d\u52301B\u53c2\u6570\uff0c\u75281\/10\u7684\u8bad\u7ec3\u5c31\u8d85\u8d8a\u4e86Qwen3"},"content":{"rendered":"\n<p>\u4e0e\u5176\u4ed6\u5168\u5f00\u6e90\u6a21\u578b\u76f8\u6bd4\uff0c\u6027\u80fd\u63d0\u53472-5\u500d\u3002<\/p>\n\n\n\n<p>\u5c0f\u53c2\u6570\u6a21\u578b\u4e5f\u8fdb\u5165\u4e86 R1 \u65f6\u4ee3\uff0c\u8fd9\u6b21\u5f00\u6e90\u51fa\u65b0\u6280\u672f\u7684\u662f Meta\u3002<\/p>\n\n\n\n<p>\u672c\u5468\uff0cMeta AI \u56e2\u961f\u6b63\u5f0f\u53d1\u5e03\u4e86 MobileLLM-R1\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>HuggingFace \u94fe\u63a5\uff1ahttps:\/\/huggingface.co\/collections\/facebook\/mobilellm-r1-68c4597b104fac45f28f448e<\/li>\n\n\n\n<li>\u8bd5\u7528\u94fe\u63a5\uff1ahttps:\/\/huggingface.co\/spaces\/akhaliq\/MobileLLM-R1-950M<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u662f MobileLLM \u7684\u5168\u65b0\u9ad8\u6548\u63a8\u7406\u6a21\u578b\u7cfb\u5217\uff0c<strong>\u5305\u542b\u4e24\u7c7b\u6a21\u578b<\/strong>\uff1a\u57fa\u7840\u6a21\u578b MobileLLM-R1-140M-base\u3001MobileLLM-R1-360M-base\u3001MobileLLM-R1-950M-base \u548c\u5b83\u4eec\u76f8\u5e94\u7684\u6700\u7ec8\u6a21\u578b\u7248\u3002<\/p>\n\n\n\n<p>\u5b83\u4eec\u4e0d\u662f\u901a\u7528\u7684\u804a\u5929\u6a21\u578b\uff0c\u800c\u662f\u76d1\u7763\u5fae\u8c03 (SFT) \u6a21\u578b\uff0c\u4e13\u95e8\u9488\u5bf9\u6570\u5b66\u3001\u7f16\u7a0b\uff08Python\u3001C++\uff09\u548c\u79d1\u5b66\u95ee\u9898\u8fdb\u884c\u8bad\u7ec3\u3002<\/p>\n\n\n\n<p>\u9664\u4e86\u6a21\u578b\u672c\u8eab\u4e4b\u5916\uff0cMeta \u8fd8\u53d1\u5e03\u4e86\u5b8c\u6574\u7684\u8bad\u7ec3\u65b9\u6848\u548c\u6570\u636e\u6e90\uff0c\u4ee5\u786e\u4fdd\u53ef\u91cd\u590d\u6027\u5e76\u652f\u6301\u8fdb\u4e00\u6b65\u7684\u7814\u7a76\u3002<\/p>\n\n\n\n<p>\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u8be5\u7cfb\u5217\u53c2\u6570\u6700\u5927\u7684 MobileLLM-R1 950M \u6a21\u578b<strong>\u4ec5\u4f7f\u7528\u7ea6 2T \u9ad8\u8d28\u91cf token \u8fdb\u884c\u9884\u8bad\u7ec3\uff0c\u603b\u8bad\u7ec3 token \u91cf\u5c11\u4e8e 5T<\/strong>\uff0c\u4f46\u5728 MATH\u3001GSM8K\u3001MMLU \u548c LiveCodeBench \u57fa\u51c6\u6d4b\u8bd5\u4e2d\uff0c\u5176\u6027\u80fd\u4e0e\u4f7f\u7528 36T token \u8fdb\u884c\u8bad\u7ec3\u7684 Qwen3 0.6B \u76f8\u5f53\u6216\u66f4\u4f73\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"794\" src=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-252-1024x794.jpg\" alt=\"\" class=\"wp-image-946\" srcset=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-252-1024x794.jpg 1024w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-252-300x233.jpg 300w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-252-768x595.jpg 768w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-252.jpg 1108w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>\u4e0e\u73b0\u6709\u7684\u5b8c\u5168\u5f00\u6e90\u6a21\u578b\u76f8\u6bd4\uff0c\u5c3d\u7ba1\u53c2\u6570\u89c4\u6a21\u660e\u663e\u66f4\u5c0f\uff0cMobileLLM-R1 950M \u6a21\u578b\u5728 MATH \u57fa\u51c6\u4e0a\u7684\u51c6\u786e\u7387\u4e5f\u6bd4 Olmo 1.24B \u6a21\u578b<strong>\u9ad8\u51fa\u7ea6\u4e94\u500d<\/strong>\uff0c\u6bd4 SmolLM2 1.7B \u6a21\u578b<strong>\u9ad8\u51fa\u7ea6\u4e24\u500d<\/strong>\u3002\u6b64\u5916\uff0cMobileLLM-R1 950M \u5728\u7f16\u7801\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u7684\u8868\u73b0\u8fdc\u8d85 Olmo 1.24B \u548c SmolLM2 1.7B \uff0c\u5728\u5b8c\u5168\u5f00\u6e90\u6a21\u578b\u4e2d\u521b\u4e0b\u4e86\u65b0\u7684\u6700\u9ad8\u6c34\u5e73\u3002<\/p>\n\n\n\n<p>Token \u6548\u7387\u7684\u6bd4\u8f83\u5982\u4e0b\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"407\" src=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-249-1024x407.jpg\" alt=\"\" class=\"wp-image-943\" srcset=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-249-1024x407.jpg 1024w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-249-300x119.jpg 300w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-249-768x305.jpg 768w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-249.jpg 1108w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>\u540e\u8bad\u7ec3\u6bd4\u8f83\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"438\" src=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-247-1024x438.jpg\" alt=\"\" class=\"wp-image-941\" srcset=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-247-1024x438.jpg 1024w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-247-300x128.jpg 300w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-247-768x329.jpg 768w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-247.jpg 1108w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>\u6a21\u578b\u67b6\u6784\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"282\" src=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-251-1024x282.jpg\" alt=\"\" class=\"wp-image-945\" srcset=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-251-1024x282.jpg 1024w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-251-300x83.jpg 300w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-251-768x211.jpg 768w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-251.jpg 1108w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"583\" src=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-250-1024x583.jpg\" alt=\"\" class=\"wp-image-944\" srcset=\"http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-250-1024x583.jpg 1024w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-250-300x171.jpg 300w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-250-768x437.jpg 768w, http:\/\/www.applicationsofllm.com\/wp-content\/uploads\/2025\/09\/image-250.jpg 1108w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>MobileLLM-R1 \u7684\u53d1\u5e03\u5f15\u8d77\u4e86\u673a\u5668\u5b66\u4e60\u793e\u533a\u7684\u8ba8\u8bba\u3002\u4eba\u4eec\u6b22\u8fce\u901a\u4e49\u3001Meta \u7b49\u9876\u5c16\u5927\u6a21\u578b\u56e2\u961f\u57fa\u4e8e\u5c0f\u4f53\u91cf\u6a21\u578b\u7684\u63a2\u7d22\u3002\u8fd9\u4e00\u65b9\u5411\u7684\u8bad\u7ec3\u6210\u672c\u8f83\u4e3a\u4f4e\u5ec9\uff0c\u53ef\u4ee5\u66f4\u52a0\u65b9\u4fbf\u5c1d\u8bd5\u5404\u7c7b\u6700\u65b0\u8bba\u6587\u63d0\u51fa\u7684\u6280\u672f\uff0c\u66f4\u91cd\u8981\u7684\u662f\uff0c\u6a21\u578b\u4f53\u91cf\u7684\u4e0b\u964d\u4e5f\u610f\u5473\u7740\u5b83\u53ef\u4ee5\u8986\u76d6\u66f4\u591a\u7aef\u4fa7\u8bbe\u5907\uff0c\u5b9e\u73b0\u66f4\u5927\u9762\u79ef\u7684\u843d\u5730\u3002<\/p>\n\n\n\n<p>\u968f\u7740\u8bad\u7ec3\u6210\u672c\u666e\u904d\u4e0b\u964d\uff0c\u6211\u4eec\u5c06\u4f1a\u5f97\u5230\u66f4\u597d\u7684\u6a21\u578b\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u4e0e\u5176\u4ed6\u5168\u5f00\u6e90\u6a21\u578b\u76f8\u6bd4\uff0c\u6027\u80fd\u63d0\u53472-5\u500d\u3002 \u5c0f\u53c2\u6570\u6a21\u578b\u4e5f\u8fdb\u5165\u4e86 R1 \u65f6\u4ee3\uff0c\u8fd9\u6b21\u5f00\u6e90\u51fa\u65b0\u6280\u672f\u7684\u662f Meta\u3002 \u672c [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":942,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10],"tags":[],"class_list":["post-940","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-10"],"_links":{"self":[{"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/posts\/940","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/comments?post=940"}],"version-history":[{"count":1,"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/posts\/940\/revisions"}],"predecessor-version":[{"id":947,"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/posts\/940\/revisions\/947"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/media\/942"}],"wp:attachment":[{"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/media?parent=940"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/categories?post=940"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.applicationsofllm.com\/index.php\/wp-json\/wp\/v2\/tags?post=940"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}