{"id":16888,"date":"2026-11-02T03:47:59","date_gmt":"2026-11-01T20:47:59","guid":{"rendered":"https:\/\/blogdee.com\/?p=16888"},"modified":"2026-11-02T03:47:59","modified_gmt":"2026-11-01T20:47:59","slug":"amazon-sagemaker-ai-%e0%b8%a3%e0%b8%ad%e0%b8%87%e0%b8%a3%e0%b8%b1%e0%b8%9a%e0%b8%81%e0%b8%b2%e0%b8%a3%e0%b8%9b%e0%b8%a3%e0%b8%b1%e0%b8%9a%e0%b9%81%e0%b8%95%e0%b9%88%e0%b8%87%e0%b9%82%e0%b8%a1%e0%b9%80","status":"publish","type":"post","link":"https:\/\/blogdee.com\/?p=16888","title":{"rendered":"Amazon SageMaker AI \u0e23\u0e2d\u0e07\u0e23\u0e31\u0e1a\u0e01\u0e32\u0e23\u0e1b\u0e23\u0e31\u0e1a\u0e41\u0e15\u0e48\u0e07\u0e42\u0e21\u0e40\u0e14\u0e25\u0e41\u0e1a\u0e1a Serverless \u0e2a\u0e33\u0e2b\u0e23\u0e31\u0e1a Muse Glimmer, Qwen \u0e41\u0e25\u0e30 Ministral"},"content":{"rendered":"<p>Amazon SageMaker AI \u0e02\u0e13\u0e30\u0e19\u0e35\u0e49\u0e23\u0e2d\u0e07\u0e23\u0e31\u0e1a\u0e01\u0e32\u0e23\u0e1b\u0e23\u0e31\u0e1a\u0e41\u0e15\u0e48\u0e07\u0e42\u0e21\u0e40\u0e14\u0e25\u0e41\u0e1a\u0e1a Serverless \u0e2a\u0e33\u0e2b\u0e23\u0e31\u0e1a\u0e42\u0e21\u0e40\u0e14\u0e25 Muse Glimmer, Qwen3.8 27B \u0e41\u0e25\u0e30 Ministral 3 (3B, 8B, 14B) \u0e19\u0e2d\u0e01\u0e08\u0e32\u0e01\u0e01\u0e32\u0e23 Deploy \u0e1a\u0e19 SageMaker AI \u0e41\u0e25\u0e49\u0e27 \u0e1c\u0e39\u0e49\u0e43\u0e0a\u0e49\u0e22\u0e31\u0e07\u0e2a\u0e32\u0e21\u0e32\u0e23\u0e16\u0e1b\u0e23\u0e31\u0e1a\u0e41\u0e15\u0e48\u0e07\u0e42\u0e21\u0e40\u0e14\u0e25\u0e40\u0e2b\u0e25\u0e48\u0e32\u0e19\u0e35\u0e49\u0e43\u0e2b\u0e49\u0e40\u0e02\u0e49\u0e32\u0e01\u0e31\u0e1a\u0e42\u0e14\u0e40\u0e21\u0e19\u0e41\u0e25\u0e30\u0e40\u0e27\u0e34\u0e23\u0e4c\u0e01\u0e42\u0e1f\u0e25\u0e27\u0e4c\u0e40\u0e09\u0e1e\u0e32\u0e30\u0e17\u0e32\u0e07\u0e44\u0e14\u0e49\u0e14\u0e49\u0e27\u0e22\u0e40\u0e17\u0e04\u0e19\u0e34\u0e04\u0e15\u0e48\u0e32\u0e07\u0e46 \u0e40\u0e0a\u0e48\u0e19 supervised fine-tuning (SFT), direct preference optimization (DPO) \u0e41\u0e25\u0e30 reinforcement fine-tuning (RFT)<\/p>\n<p>\u0e01\u0e32\u0e23\u0e1b\u0e23\u0e31\u0e1a\u0e41\u0e15\u0e48\u0e07\u0e19\u0e35\u0e49\u0e2a\u0e32\u0e21\u0e32\u0e23\u0e16\u0e17\u0e33\u0e44\u0e14\u0e49\u0e17\u0e31\u0e49\u0e07\u0e41\u0e1a\u0e1a low-rank adaptation (LoRA) \u0e2b\u0e23\u0e37\u0e2d full fine-tuning \u0e42\u0e14\u0e22 SageMaker AI \u0e08\u0e30\u0e40\u0e02\u0e49\u0e32\u0e21\u0e32\u0e08\u0e31\u0e14\u0e01\u0e32\u0e23\u0e40\u0e23\u0e37\u0e48\u0e2d\u0e07\u0e01\u0e32\u0e23\u0e08\u0e31\u0e14\u0e40\u0e15\u0e23\u0e35\u0e22\u0e21\u0e42\u0e04\u0e23\u0e07\u0e2a\u0e23\u0e49\u0e32\u0e07\u0e1e\u0e37\u0e49\u0e19\u0e10\u0e32\u0e19\u0e41\u0e25\u0e30\u0e01\u0e32\u0e23\u0e1b\u0e23\u0e30\u0e2a\u0e32\u0e19\u0e07\u0e32\u0e19\u0e01\u0e32\u0e23\u0e1d\u0e36\u0e01\u0e2a\u0e2d\u0e19\u0e17\u0e31\u0e49\u0e07\u0e2b\u0e21\u0e14 \u0e17\u0e33\u0e43\u0e2b\u0e49\u0e1c\u0e39\u0e49\u0e43\u0e0a\u0e49\u0e2a\u0e32\u0e21\u0e32\u0e23\u0e16\u0e21\u0e38\u0e48\u0e07\u0e40\u0e19\u0e49\u0e19\u0e44\u0e1b\u0e17\u0e35\u0e48\u0e02\u0e49\u0e2d\u0e21\u0e39\u0e25\u0e41\u0e25\u0e30\u0e01\u0e32\u0e23\u0e1b\u0e23\u0e30\u0e40\u0e21\u0e34\u0e19\u0e1c\u0e25\u0e40\u0e17\u0e48\u0e32\u0e19\u0e31\u0e49\u0e19 \u0e41\u0e25\u0e30\u0e08\u0e48\u0e32\u0e22\u0e15\u0e32\u0e21\u0e01\u0e32\u0e23\u0e43\u0e0a\u0e49\u0e07\u0e32\u0e19\u0e08\u0e23\u0e34\u0e07<\/p>\n<p>\u0e1a\u0e23\u0e34\u0e01\u0e32\u0e23 Serverless customization \u0e19\u0e35\u0e49\u0e40\u0e1b\u0e34\u0e14\u0e43\u0e2b\u0e49\u0e1a\u0e23\u0e34\u0e01\u0e32\u0e23\u0e43\u0e19\u0e20\u0e39\u0e21\u0e34\u0e20\u0e32\u0e04 US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo) \u0e41\u0e25\u0e30 Europe (Ireland) \u0e42\u0e14\u0e22\u0e2a\u0e32\u0e21\u0e32\u0e23\u0e16\u0e40\u0e23\u0e34\u0e48\u0e21\u0e15\u0e49\u0e19\u0e44\u0e14\u0e49\u0e08\u0e32\u0e01\u0e2b\u0e19\u0e49\u0e32 Models \u0e43\u0e19 Amazon SageMaker Studio \u0e2b\u0e23\u0e37\u0e2d\u0e43\u0e0a\u0e49 SageMaker Python SDK<\/p>\n<p>\u0e17\u0e35\u0e48\u0e21\u0e32: <a href=\"https:\/\/aws.amazon.com\/about-aws\/whats-new\/2026\/10\/amazon-sagemaker-muse-qwen-ministral\/\" target=\"_blank\" rel=\"noopener nofollow\">AWS<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Amazon SageMaker AI \u0e40\u0e1b\u0e34\u0e14\u0e15\u0e31\u0e27\u0e23\u0e2d\u0e07\u0e23\u0e31\u0e1a\u0e01\u0e32\u0e23\u0e1b\u0e23\u0e31\u0e1a\u0e41\u0e15\u0e48\u0e07\u0e42\u0e21\u0e40\u0e14\u0e25 (model customization) \u0e41\u0e1a\u0e1a Serverless \u0e2a\u0e33\u0e2b\u0e23\u0e31\u0e1a\u0e42\u0e21\u0e40\u0e14\u0e25 Muse Glimmer, Qwen3.8 27B \u0e41\u0e25\u0e30 Ministral\u2026<\/p>\n","protected":false},"author":1,"featured_media":16887,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[50],"tags":[4608,88,5916,5936],"class_list":["post-16888","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cloud-computing","tag-fine-tuning","tag-llm","tag-sagemaker","tag-serverless"],"_links":{"self":[{"href":"https:\/\/blogdee.com\/index.php?rest_route=\/wp\/v2\/posts\/16888","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blogdee.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogdee.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogdee.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/blogdee.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=16888"}],"version-history":[{"count":0,"href":"https:\/\/blogdee.com\/index.php?rest_route=\/wp\/v2\/posts\/16888\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blogdee.com\/index.php?rest_route=\/wp\/v2\/media\/16887"}],"wp:attachment":[{"href":"https:\/\/blogdee.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=16888"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogdee.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=16888"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogdee.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=16888"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}