{"id":5274,"date":"2024-12-06T11:44:26","date_gmt":"2024-12-06T03:44:26","guid":{"rendered":"https:\/\/nullthought.net\/?p=5274"},"modified":"2024-12-25T14:29:42","modified_gmt":"2024-12-25T06:29:42","slug":"paligemma-2-%e7%94%a8%e4%ba%8e%e8%bf%81%e7%a7%bb%e5%ad%a6%e4%b9%a0%e7%9a%84%e5%a4%9a%e5%8a%9f%e8%83%bd%e8%a7%86%e8%a7%89-%e8%af%ad%e8%a8%80%e6%a8%a1%e5%9e%8b%ef%bc%88vlm%ef%bc%89%e5%ae%b6%e6%97%8f","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=5274","title":{"rendered":"PaliGemma 2: \u7528\u4e8e\u8fc1\u79fb\u5b66\u4e60\u7684\u591a\u529f\u80fd\u89c6\u89c9-\u8bed\u8a00\u6a21\u578b\uff08VLM\uff09\u5bb6\u65cf"},"content":{"rendered":"\n<p>\u8bba\u6587<strong><a href=\"https:\/\/arxiv.org\/abs\/2412.03555\" target=\"_blank\" rel=\"noreferrer noopener\">PaliGemma 2: A Family of Versatile VLMs for Transfer<\/a><\/strong>\uff08\u300aPaliGemma 2: \u7528\u4e8e\u8fc1\u79fb\u5b66\u4e60\u7684\u591a\u529f\u80fd\u89c6\u89c9-\u8bed\u8a00\u6a21\u578b\u5bb6\u65cf\u300b\uff09\u8be6\u7ec6\u4ecb\u7ecd\u4e86<strong><a href=\"https:\/\/nullthought.net\/?p=4084\" target=\"_blank\" rel=\"noreferrer noopener\">PaliGemma\u6a21\u578b<\/a><\/strong>\u7684\u5347\u7ea7\u8fc7\u7a0b\u3002PaliGemma 2\u57fa\u4e8eGemma 2\u7cfb\u5217\u8bed\u8a00\u6a21\u578b\uff0c\u662f\u4e00\u79cd\u591a\u529f\u80fd\u7684<strong><a href=\"https:\/\/nullthought.net\/?p=4920\" target=\"_blank\" rel=\"noreferrer noopener\">\u89c6\u89c9-\u8bed\u8a00\u6a21\u578b\uff08Vision-Language Model, VLM\uff09<\/a><\/strong>\uff0c\u4e13\u4e3a\u5e7f\u6cdb\u7684\u8fc1\u79fb\u4efb\u52a1\u800c\u8bbe\u8ba1\uff0c\u76ee\u6807\u662f\u901a\u8fc7\u6a21\u578b\u89c4\u6a21\u3001\u5206\u8fa8\u7387\u4ee5\u53ca\u8bad\u7ec3\u7b56\u7565\u7684\u521b\u65b0\uff0c\u8fdb\u4e00\u6b65\u63a8\u52a8\u89c6\u89c9\u548c\u8bed\u8a00\u4efb\u52a1\u7684\u8fdb\u5c55\u3002<\/p>\n\n\n\n<p>PaliGemma 2\u4f5c\u4e3a\u4e00\u4e2a\u5f00\u6e90\u7684\u6a21\u578b\u5bb6\u65cf\uff0c\u5176\u63d0\u4f9b\u7684\u5e7f\u6cdb\u57fa\u7840\u6a21\u578b\u53ef\u4ee5\u4f5c\u4e3a\u73b0\u6709PaliGemma\u6a21\u578b\u7684\u76f4\u63a5\u66ff\u4ee3\uff0c\u5177\u5907\u66f4\u5f3a\u7684\u8fc1\u79fb\u80fd\u529b\u548c\u66f4\u5e7f\u6cdb\u7684\u5e94\u7528\u573a\u666f\uff0c\u9884\u8ba1\u5c06\u5728\u591a\u4e2a\u9886\u57df\u7684\u89c6\u89c9-\u8bed\u8a00\u4efb\u52a1\u4e2d\u63a8\u52a8\u65b0\u7684\u8fdb\u5c55\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u4f5c\u8005\u4e3aAndreas Steiner, Andr\u00e9 Susano Pinto, Michael Tschannen, Daniel Keysers, Xiao Wang, Yonatan Bitton, Alexey Gritsenko, Matthias Minderer, Anthony Sherbondy, Shangbang Long, Siyang Qin, Reeve Ingle, Emanuele Bugliarello, Sahar Kazemzadeh, Thomas Mesnard, Ibrahim Alabdulmohsin, Lucas Beyer, Xiaohua Zhai\uff0c\u6765\u81eaGoogle DeepMind\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"621\" height=\"636\" src=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/12\/image-6.png\" alt=\"\" class=\"wp-image-5275\" srcset=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/12\/image-6.png 621w, https:\/\/nullthought.net\/wp-content\/uploads\/2024\/12\/image-6-293x300.png 293w\" sizes=\"auto, (max-width: 621px) 100vw, 621px\" \/><figcaption class=\"wp-element-caption\"><strong><a href=\"https:\/\/arxiv.org\/abs\/2412.03555\" target=\"_blank\" rel=\"noreferrer noopener\">PaliGemma 2: A Family of Versatile VLMs for Transfer<\/a><\/strong><\/figcaption><\/figure>\n\n\n\n<p><a href=\"https:\/\/nullthought.net\/?p=4084\" target=\"_blank\" rel=\"noreferrer noopener\">PaliGemma\u6700\u521d\u662f\u4e00\u4e2a\u57fa\u4e8eSigLIP\u89c6\u89c9\u7f16\u7801\u5668\u548cGemma\u8bed\u8a00\u6a21\u578b\u76843B\u89c4\u6a21\u7684\u89c6\u89c9-\u8bed\u8a00\u6a21\u578b\uff0c\u901a\u8fc7\u5bf9\u89c6\u89c9\u548c\u8bed\u8a00\u7684\u7ed3\u5408\u5b9e\u73b0\u4e86\u6bd4\u4ee5\u5f80\u66f4\u5f3a\u7684\u8fc1\u79fb\u80fd\u529b<\/a>\u3002\u7136\u800c\uff0cPaliGemma\u7684\u8bed\u8a00\u6a21\u578b\u4ecd\u6709\u5347\u7ea7\u7a7a\u95f4\u3002\u56e0\u6b64\uff0c\u7814\u7a76\u56e2\u961f\u5c06\u5176\u66f4\u65b0\u4e3aGemma 2\u8bed\u8a00\u6a21\u578b\uff0c\u751f\u6210\u4e86\u65b0\u7684PaliGemma 2\u57fa\u7840VLM\u7cfb\u5217\uff0c\u6db5\u76d63B\u300110B\u300128B\u4e09\u79cd\u4e0d\u540c\u6a21\u578b\u89c4\u6a21\uff0c\u5e76\u5728224px\u00b2\u3001448px\u00b2\u3001896px\u00b2\u4e09\u4e2a\u5206\u8fa8\u7387\u4e0b\u8bad\u7ec3\u3002\u8fd9\u4e9b\u6a21\u578b\u4e0d\u4ec5\u5728\u5206\u8fa8\u7387\u548c\u6a21\u578b\u89c4\u6a21\u4e0a\u505a\u4e86\u5347\u7ea7\uff0c\u800c\u4e14\u5728\u8fc1\u79fb\u4efb\u52a1\u7684\u79cd\u7c7b\u548c\u8986\u76d6\u8303\u56f4\u4e0a\u4e5f\u6709\u4e86\u5927\u5e45\u5ea6\u589e\u52a0\uff0c\u6db5\u76d6\u8868\u7ed3\u6784\u8bc6\u522b\u3001\u5206\u5b50\u7ed3\u6784\u8bc6\u522b\u3001\u97f3\u4e50\u4e50\u8c31\u8bc6\u522b\u3001\u957f\u7bc7\u7ec6\u7c92\u5ea6\u56fe\u50cf\u63cf\u8ff0\u751f\u6210\u3001\u7a7a\u95f4\u63a8\u7406\u3001\u4ee5\u53ca\u653e\u5c04\u5b66\u62a5\u544a\u751f\u6210\u7b49\u4efb\u52a1\u3002<\/p>\n\n\n\n<p>PaliGemma 2\u7684\u8bad\u7ec3\u91c7\u7528\u4e86\u4e09\u9636\u6bb5\u8bad\u7ec3\u7b56\u7565\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u9636\u6bb51\uff1a\u57fa\u7840\u591a\u6a21\u6001\u4efb\u52a1\u8bad\u7ec3<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u5728\u7b2c\u4e00\u9636\u6bb5\u4e2d\uff0cPaliGemma 2\u5c06\u9884\u8bad\u7ec3\u7684SigLIP-So400m\u89c6\u89c9\u7f16\u7801\u5668\u4e0eGemma 2\u8bed\u8a00\u6a21\u578b\u7ed3\u5408\uff0c\u5bf9\u5305\u542b10\u4ebf\u4f8b\u7684\u591a\u6a21\u6001\u4efb\u52a1\u6570\u636e\u96c6\u8fdb\u884c\u8bad\u7ec3\uff0c\u4ee5\u4fc3\u8fdb\u6a21\u578b\u5728\u5e7f\u6cdb\u4efb\u52a1\u4e2d\u7684\u8fc1\u79fb\u80fd\u529b\u3002\u8fd9\u4e00\u9636\u6bb5\u7684\u8bad\u7ec3\u5206\u8fa8\u7387\u4e3a224px\u00b2\uff0c\u6240\u6709\u6a21\u578b\u53c2\u6570\u5747\u672a\u88ab\u51bb\u7ed3\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u9636\u6bb52\uff1a\u9010\u6b65\u63d0\u9ad8\u5206\u8fa8\u7387\u7684\u8bad\u7ec3<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u5728\u9636\u6bb52\u4e2d\uff0c\u6a21\u578b\u9996\u5148\u4f7f\u7528448px\u00b2\u7684\u5206\u8fa8\u7387\u8bad\u7ec35000\u4e07\u4f8b\uff0c\u7136\u540e\u518d\u4ee5896px\u00b2\u7684\u5206\u8fa8\u7387\u8bad\u7ec31000\u4e07\u4f8b\u3002\u4e3a\u4e86\u9002\u5e94\u8f83\u9ad8\u5206\u8fa8\u7387\u7684\u4efb\u52a1\uff0c\u4efb\u52a1\u6df7\u5408\u96c6\u4e2d\u589e\u52a0\u4e86\u5bf9\u9ad8\u5206\u8fa8\u7387\u6709\u76ca\u7684\u4efb\u52a1\u7684\u6743\u91cd\uff0c\u5e76\u5ef6\u957f\u4e86\u8f93\u51fa\u5e8f\u5217\u7684\u957f\u5ea6\uff0c\u4ee5\u4fbf\u66f4\u597d\u5730\u8fdb\u884c\u590d\u6742OCR\u4efb\u52a1\u7684\u5b66\u4e60\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u9636\u6bb53\uff1a\u5fae\u8c03\u5230\u5177\u4f53\u4efb\u52a1<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u9636\u6bb53\u5bf9\u9636\u6bb51\u6216\u9636\u6bb52\u751f\u6210\u7684\u68c0\u67e5\u70b9\u8fdb\u884c\u5fae\u8c03\uff0c\u4ee5\u9002\u5e94\u7279\u5b9a\u7684\u76ee\u6807\u4efb\u52a1\u3002\u8fd9\u4e00\u9636\u6bb5\u91c7\u7528\u4e86\u591a\u79cd\u5b66\u672f\u57fa\u51c6\u6d4b\u8bd5\uff0c\u5305\u62ec\u591a\u4e2a\u6d89\u53ca\u591a\u5f20\u56fe\u50cf\u548c\u77ed\u89c6\u9891\u7684\u4efb\u52a1\uff0c\u5982\u5e38\u89c1\u7684\u56fe\u7247\u6807\u6ce8\u3001\u89c6\u89c9\u95ee\u7b54\uff08VQA\uff09\u4efb\u52a1\u7b49\u3002\u6b64\u5916\uff0cPaliGemma 2\u8fd8\u5c1d\u8bd5\u4e86\u65b0\u7684\u5e94\u7528\uff0c\u5305\u62ec\u6587\u6863\u76f8\u5173\u4efb\u52a1\u3001\u957f\u7bc7\u63cf\u8ff0\u751f\u6210\u548c\u533b\u5b66\u56fe\u50cf\u7406\u89e3\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n\n\n\n<p>PaliGemma 2\u5728\u591a\u4e2a\u8fc1\u79fb\u4efb\u52a1\u4e0a\u8fdb\u884c\u4e86\u5e7f\u6cdb\u7684\u5b9e\u9a8c\uff0c\u4ee5\u4e0b\u662f\u5404\u65b9\u9762\u7684\u8be6\u7ec6\u5b9e\u9a8c\u7ed3\u679c\u53ca\u5176\u5206\u6790\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u6a21\u578b\u89c4\u6a21\u548c\u5206\u8fa8\u7387\u5bf9\u6027\u80fd\u7684\u5f71\u54cd<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u5b9e\u9a8c\u7ed3\u679c\u8868\u660e\uff0c\u589e\u52a0\u6a21\u578b\u7684\u89c4\u6a21\u548c\u56fe\u50cf\u5206\u8fa8\u7387\u901a\u5e38\u53ef\u4ee5\u663e\u8457\u63d0\u9ad8\u6a21\u578b\u5728\u8fc1\u79fb\u4efb\u52a1\u4e2d\u7684\u8868\u73b0\u3002\u7279\u522b\u662f\u5bf9\u4e00\u4e9b\u4f9d\u8d56\u89c6\u89c9\u7ec6\u8282\u7684\u4efb\u52a1\uff0c\u5982\u6587\u672c\u8bc6\u522b\u3001\u8868\u7ed3\u6784\u8bc6\u522b\u7b49\uff0c\u5206\u8fa8\u7387\u7684\u63d0\u5347\u5c24\u5176\u91cd\u8981\u3002\u4f8b\u5982\uff0c\u56fe\u8868\u95ee\u7b54\u4efb\u52a1\uff08ChartQA\uff09\u548c\u6587\u6863\u89c6\u89c9\u95ee\u7b54\uff08DocVQA\uff09\u7b49\u6d89\u53ca\u6587\u6863\u548c\u56fe\u8868\u7406\u89e3\u7684\u4efb\u52a1\uff0c\u663e\u8457\u53d7\u76ca\u4e8e\u5206\u8fa8\u7387\u7684\u589e\u52a0\uff0c\u800c\u6d89\u53ca\u591a\u8bed\u8a00\u6570\u636e\u7684\u4efb\u52a1\u5219\u66f4\u53d7\u76ca\u4e8e\u6a21\u578b\u89c4\u6a21\u7684\u589e\u52a0\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u8fc1\u79fb\u5b66\u4e60\u901f\u7387\u548c\u6a21\u578b\u89c4\u6a21\u7684\u5173\u7cfb<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u7814\u7a76\u53d1\u73b0\uff0c\u968f\u7740\u6a21\u578b\u89c4\u6a21\u7684\u589e\u52a0\uff0c\u6700\u4f18\u8fc1\u79fb\u5b66\u4e60\u901f\u7387\u9010\u6e10\u964d\u4f4e\u3002\u56e0\u6b64\uff0c\u5728\u8fc1\u79fb\u5230\u8f83\u5927\u7684\u6a21\u578b\u65f6\uff0c\u63a8\u8350\u4f7f\u7528\u8f83\u5c0f\u7684\u5b66\u4e60\u901f\u7387\u8fdb\u884c\u5fae\u8c03\u3002\u6b64\u5916\uff0c\u76f8\u6bd4\u4e8e\u65e9\u671f\u7684PaliGemma\u6a21\u578b\uff0cPaliGemma 2\u76843B\u6a21\u578b\u901a\u5e38\u9700\u8981\u66f4\u4f4e\u7684\u5b66\u4e60\u901f\u7387\u624d\u80fd\u8fbe\u5230\u6700\u4f18\u6548\u679c\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u4e0d\u540c\u4efb\u52a1\u7684\u6027\u80fd\u8868\u73b0<\/strong>\n<ul class=\"wp-block-list\">\n<li><strong>\u8868\u7ed3\u6784\u8bc6\u522b<\/strong>\uff1aPaliGemma 2\u5728\u8868\u7ed3\u6784\u8bc6\u522b\u4efb\u52a1\u4e0a\u8868\u73b0\u4f18\u5f02\uff0c\u4f7f\u7528\u4e86FinTabNet\u548cPubTabNet\u6570\u636e\u96c6\uff0c\u8bc4\u4f30\u6307\u6807\u5305\u62ecTree Edit Distance Similarity (TEDS)\u548cGrid Table Similarity (GriTS)\u3002PaliGemma 2\u5728\u5927\u591a\u6570\u8fd9\u4e9b\u6307\u6807\u4e0a\u8bbe\u7f6e\u4e86\u65b0\u7684\u6700\u5148\u8fdb\u8bb0\u5f55\uff0c\u8fd9\u8868\u660e\u5b83\u5728\u5904\u7406\u8868\u683c\u7ed3\u6784\u7684\u590d\u6742\u6027\u548c\u7cbe\u5ea6\u65b9\u9762\u53d6\u5f97\u4e86\u663e\u8457\u8fdb\u5c55\u3002<\/li>\n\n\n\n<li><strong>\u5206\u5b50\u7ed3\u6784\u8bc6\u522b<\/strong>\uff1a\u5728\u5206\u5b50\u7ed3\u6784\u8bc6\u522b\u4efb\u52a1\u4e2d\uff0cPaliGemma 2\u88ab\u7528\u4e8e\u8bc6\u522b\u5206\u5b50\u56fe\u7ed3\u6784\uff08\u4ee5SMILES\u5b57\u7b26\u4e32\u8868\u793a\uff09\uff0c\u4f7f\u7528\u4e86\u4ecePubChem\u6570\u636e\u96c6\u4e2d\u83b7\u53d6\u7684100\u4e07\u5206\u5b50\u56fe\u50cf\u6570\u636e\u8fdb\u884c\u8bad\u7ec3\uff0c\u5e76\u5bf9ChemDraw\u5e93\u4e2d\u7684\u8bc4\u4f30\u96c6\u8fdb\u884c\u4e86\u8bc4\u4f30\u3002PaliGemma 2\u5728\u8be5\u4efb\u52a1\u4e0a\u8d85\u8d8a\u4e86\u4e4b\u524d\u7684\u6700\u4f73\u6a21\u578bMolScribe\u3002<\/li>\n\n\n\n<li><strong>\u5149\u5b66\u97f3\u4e50\u8bc6\u522b<\/strong>\uff1aPaliGemma 2\u5728\u5149\u5b66\u97f3\u4e50\u8bc6\u522b\u4efb\u52a1\u4e2d\uff0c\u901a\u8fc7\u5bf9GrandStaff\u6570\u636e\u96c6\u7684\u8bad\u7ec3\uff0c\u5c06\u94a2\u7434\u4e50\u8c31\u56fe\u50cf\u7ffb\u8bd1\u4e3a\u6570\u5b57\u97f3\u4e50\u8868\u793a\u683c\u5f0f\u3002\u4f7f\u7528\u7684\u8bc4\u4f30\u6307\u6807\u5305\u62ec\u5b57\u7b26\u9519\u8bef\u7387\uff08CER\uff09\u3001\u7b26\u53f7\u9519\u8bef\u7387\uff08SER\uff09\u548c\u884c\u9519\u8bef\u7387\uff08LER\uff09\uff0cPaliGemma 2\u5728\u6240\u6709\u8fd9\u4e9b\u6307\u6807\u4e0a\u5747\u6709\u663e\u8457\u964d\u4f4e\uff0c\u8868\u660e\u5b83\u5728\u97f3\u4e50\u8bc6\u522b\u4efb\u52a1\u4e2d\u7684\u5f3a\u5927\u80fd\u529b\u3002<\/li>\n\n\n\n<li><strong>\u653e\u5c04\u5b66\u62a5\u544a\u751f\u6210<\/strong>\uff1a\u4e3a\u4e86\u6d4b\u8bd5PaliGemma 2\u5728\u533b\u7597\u9886\u57df\u7684\u5e94\u7528\uff0c\u6a21\u578b\u88ab\u7528\u4e8e\u81ea\u52a8\u751f\u6210\u80f8\u90e8X\u5c04\u7ebf\u62a5\u544a\uff0c\u6570\u636e\u96c6\u4e3aMIMIC-CXR\u3002\u901a\u8fc7\u5fae\u8c03\u6a21\u578b\uff0cPaliGemma 2\u5728RadGraph\u8bc4\u4f30\u6307\u6807\u4e0a\u8fbe\u5230\u4e86\u5f53\u524d\u6700\u5148\u8fdb\u7684\u6027\u80fd\u3002\u8fd9\u663e\u793a\u51fa\u5176\u5728\u751f\u6210\u533b\u5b66\u62a5\u544a\u65b9\u9762\u7684\u6f5c\u529b\uff0c\u5c24\u5176\u662f\u5728\u63cf\u8ff0\u590d\u6742\u533b\u5b66\u56fe\u50cf\u5185\u5bb9\u65f6\u7684\u7ec6\u8282\u8868\u8fbe\u80fd\u529b\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n\n\n\n<p>PaliGemma 2\u7684\u7814\u7a76\u610f\u4e49\u5728\u4e8e\u5b83\u6210\u529f\u5730\u8bc1\u660e\u4e86\u901a\u8fc7\u7ed3\u5408\u9ad8\u8d28\u91cf\u89c6\u89c9\u7f16\u7801\u5668\u4e0e\u5f3a\u5927\u7684\u8bed\u8a00\u6a21\u578b\uff0c\u53ef\u4ee5\u5728\u591a\u79cd\u89c6\u89c9-\u8bed\u8a00\u4efb\u52a1\u4e2d\u5b9e\u73b0\u6700\u5148\u8fdb\u7684\u6027\u80fd\u3002\u5176\u521b\u65b0\u4e4b\u5904\u5728\u4e8e\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u6a21\u578b\u7684\u6574\u4f53\u89c4\u6a21\u548c\u5206\u8fa8\u7387\uff0c\u8fd8\u901a\u8fc7\u591a\u9636\u6bb5\u8bad\u7ec3\u7b56\u7565\uff0c\u786e\u4fdd\u6a21\u578b\u80fd\u591f\u5728\u5e7f\u6cdb\u7684\u4efb\u52a1\u4e2d\u8fdb\u884c\u9ad8\u6548\u8fc1\u79fb\u3002\u8fd9\u4e9b\u6a21\u578b\u7684\u591a\u6837\u6027\u4e5f\u4f7f\u5f97\u7814\u7a76\u56e2\u961f\u53ef\u4ee5\u66f4\u6df1\u5165\u5730\u5206\u6790\u6a21\u578b\u89c4\u6a21\u3001\u56fe\u50cf\u5206\u8fa8\u7387\u4ee5\u53ca\u4e0d\u540c\u4efb\u52a1\u7c7b\u578b\u4e4b\u95f4\u7684\u76f8\u4e92\u4f5c\u7528\u3002<\/p>\n\n\n\n<p><strong>\u672a\u6765\u65b9\u5411<\/strong>\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u8de8\u9886\u57df\u5e94\u7528<\/strong>\uff1aPaliGemma 2\u5df2\u7ecf\u5c55\u793a\u4e86\u5728\u591a\u9886\u57df\uff08\u5982\u533b\u5b66\u3001\u5316\u5b66\u3001\u97f3\u4e50\u7b49\uff09\u4e2d\u7684\u5e94\u7528\u6f5c\u529b\uff0c\u672a\u6765\u53ef\u4ee5\u8fdb\u4e00\u6b65\u63a2\u7d22\u5176\u5728\u66f4\u591a\u8de8\u9886\u57df\u4efb\u52a1\u4e2d\u7684\u5e94\u7528\uff0c\u7279\u522b\u662f\u6d89\u53ca\u66f4\u52a0\u590d\u6742\u7684\u89c6\u89c9\u548c\u8bed\u8a00\u63a8\u7406\u7684\u4efb\u52a1\u3002<\/li>\n\n\n\n<li><strong>\u6a21\u578b\u7684\u8f7b\u91cf\u5316\u4e0e\u8bbe\u5907\u90e8\u7f72<\/strong>\uff1a\u5c3d\u7ba1\u76ee\u524dPaliGemma 2\u5df2\u7ecf\u8868\u73b0\u51fa\u4e86\u5f3a\u5927\u7684\u8fc1\u79fb\u80fd\u529b\uff0c\u672a\u6765\u7684\u7814\u7a76\u53ef\u4ee5\u66f4\u52a0\u5173\u6ce8\u5982\u4f55\u5bf9\u6a21\u578b\u8fdb\u884c\u8fdb\u4e00\u6b65\u7684\u8f7b\u91cf\u5316\u5904\u7406\uff0c\u4ee5\u4fbf\u5728\u8fb9\u7f18\u8bbe\u5907\u4e0a\u8fd0\u884c\u3002\u6b64\u5916\uff0c\u901a\u8fc7\u5bf9\u6a21\u578b\u8fdb\u884c\u91cf\u5316\uff08\u59828-bit\u91cf\u5316\uff09\uff0c\u53ef\u4ee5\u5728\u4fdd\u8bc1\u6027\u80fd\u7684\u524d\u63d0\u4e0b\u5927\u5e45\u964d\u4f4e\u63a8\u7406\u7684\u8ba1\u7b97\u5f00\u9500\uff0c\u4f7f\u5176\u66f4\u9002\u5408\u5728\u8d44\u6e90\u6709\u9650\u7684\u8bbe\u5907\u4e0a\u8fdb\u884c\u63a8\u7406\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>\u8bba\u6587PaliGemma 2: A Family of Versatile VLMs for Transfer\uff08 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8],"tags":[39,90,80,66],"class_list":["post-5274","post","type-post","status-publish","format-standard","hentry","category-tech","tag-ai","tag-multimodal","tag-cv","tag-google"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a>","rttpg_excerpt":"\u8bba\u6587PaliGemma 2: A Family of Versatile VLMs for Transfer\uff08&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5274","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=5274"}],"version-history":[{"count":1,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5274\/revisions"}],"predecessor-version":[{"id":5276,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5274\/revisions\/5276"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5274"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5274"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5274"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}