{"version":2,"relationshipURLBase":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html","relationshipResolution":"Resolve relative relationship URLs against relationshipURLBase; complete records and absolute URLs are in blocks.","title":"4×4，位置 0–15","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html","markdown":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.md","pages":[{"id":"reading-opening","title":"4×4，位置 0–15","unit":"4×4，位置 0–15","unitId":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","chapter":"course","number":1}],"aliases":[{"oldId":"b-45e6f911ccb6786b6c1d39ff","targetId":"b-45e6f911ccb6786b6c1d39ff","page":"reading-opening","operation":"move","beforeSHA256":"343aaa499ee36d5531eab9b8b2e10a88939c1d9efb4bb1251925f2c15e332b89","afterSHA256":"343aaa499ee36d5531eab9b8b2e10a88939c1d9efb4bb1251925f2c15e332b89","reason":"Same source body in authored semantic page"},{"oldId":"b-c14cf67282b6e2fcb5ec8e61","targetId":"b-c14cf67282b6e2fcb5ec8e61","page":"reading-opening","operation":"move","beforeSHA256":"d80e4f5a895801698579826b78ff75d9e2b907c78e8a74c5742558a51926f97c","afterSHA256":"d80e4f5a895801698579826b78ff75d9e2b907c78e8a74c5742558a51926f97c","reason":"Same source body in authored semantic page"},{"oldId":"b-a0db837ffb1badd224f097d9","targetId":"b-a0db837ffb1badd224f097d9","page":"reading-opening","operation":"move","beforeSHA256":"42cba31deccba2a350b9b23bd0ca1dcaa4c4c5cb9d35c773e1fcb89124788ccd","afterSHA256":"42cba31deccba2a350b9b23bd0ca1dcaa4c4c5cb9d35c773e1fcb89124788ccd","reason":"Same source body in authored semantic page"},{"oldId":"b-28552bdbe40668be834905ce","targetId":"b-28552bdbe40668be834905ce","page":"reading-opening","operation":"move","beforeSHA256":"e5185ec67dda4a18681db2d55dc5cbe0c312c9a5519949e3ded12e96d49c997f","afterSHA256":"e5185ec67dda4a18681db2d55dc5cbe0c312c9a5519949e3ded12e96d49c997f","reason":"Same source body in authored semantic page"},{"oldId":"b-14261b12393b0dd45205b0d7","targetId":"b-14261b12393b0dd45205b0d7","page":"reading-opening","operation":"move","beforeSHA256":"dab1e775bbe722f2ff40501c3fe9af28013fd16eecdb6fb5667a6a0d711547f3","afterSHA256":"dab1e775bbe722f2ff40501c3fe9af28013fd16eecdb6fb5667a6a0d711547f3","reason":"Same source body in authored semantic page"},{"oldId":"b-bfaee36548edafd5d2198626","targetId":"b-bfaee36548edafd5d2198626","page":"reading-opening","operation":"move","beforeSHA256":"eaf39b3c8b6c9e1fe40882fe5a4c71f1da47a23d77302c9afeb60d563776dd34","afterSHA256":"eaf39b3c8b6c9e1fe40882fe5a4c71f1da47a23d77302c9afeb60d563776dd34","reason":"Same source body in authored semantic page"},{"oldId":"b-10fbef8074ffbf55a8581c38","targetId":"b-10fbef8074ffbf55a8581c38","page":"reading-opening","operation":"move","beforeSHA256":"819ec8665bc0b217310168ce7d9474a46963c04e4a4df5c17d24c167a3696a42","afterSHA256":"819ec8665bc0b217310168ce7d9474a46963c04e4a4df5c17d24c167a3696a42","reason":"Same source body in authored semantic page"},{"oldId":"b-517a38e50456c252e3a084c0","targetId":"b-517a38e50456c252e3a084c0","page":"reading-opening","operation":"move","beforeSHA256":"fd47fe78c4c48c67be97d9fd2b1ebf960186c37faa1d1b30a00277b62bee7de8","afterSHA256":"fd47fe78c4c48c67be97d9fd2b1ebf960186c37faa1d1b30a00277b62bee7de8","reason":"Same source body in authored semantic page"},{"oldId":"b-5c3b17e43a89a2f14932dd04","targetId":"b-5c3b17e43a89a2f14932dd04","page":"reading-opening","operation":"move","beforeSHA256":"a4c984796646c7282f868779d1340c72a36696d18cad65cdeea7a54318a491ef","afterSHA256":"a4c984796646c7282f868779d1340c72a36696d18cad65cdeea7a54318a491ef","reason":"Same source body in authored semantic page"},{"oldId":"b-1744861d6b390ab4fd501a7a","targetId":"b-1744861d6b390ab4fd501a7a","page":"reading-opening","operation":"move","beforeSHA256":"41f1852c4a7076365057faf0c061910b1cdfb6467f1bac36e502721c13dde45b","afterSHA256":"41f1852c4a7076365057faf0c061910b1cdfb6467f1bac36e502721c13dde45b","reason":"Same source body in authored semantic page"},{"oldId":"b-7e71d514370e937a783597c1","targetId":"b-7e71d514370e937a783597c1","page":"reading-opening","operation":"move","beforeSHA256":"b6785550af1ca90eee39a779b30b699f296c304901d2e2543d8db5ce06e47a57","afterSHA256":"b6785550af1ca90eee39a779b30b699f296c304901d2e2543d8db5ce06e47a57","reason":"Same source body in authored semantic page"},{"oldId":"b-69e62cad1e86f5fbf5c0ea28","targetId":"b-69e62cad1e86f5fbf5c0ea28","page":"reading-opening","operation":"move","beforeSHA256":"8b3005e9f4ad6c069da988920a4232f225408f8e65d3d980260c4a0e52798a93","afterSHA256":"8b3005e9f4ad6c069da988920a4232f225408f8e65d3d980260c4a0e52798a93","reason":"Same source body in authored semantic page"}],"blocks":[{"id":"b-45e6f911ccb6786b6c1d39ff","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-45e6f911ccb6786b6c1d39ff","content":"[CLS 01 · 学习信号、表示与迁移](/course/notebooks/fb4365d604974dd19df492cb22516e44.html)","sha256":"343aaa499ee36d5531eab9b8b2e10a88939c1d9efb4bb1251925f2c15e332b89","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-c14cf67282b6e2fcb5ec8e61","prerequisites":[],"relatedBlocks":[]},{"id":"b-c14cf67282b6e2fcb5ec8e61","type":"h1","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-c14cf67282b6e2fcb5ec8e61","content":"# 4×4，位置 0–15","sha256":"d80e4f5a895801698579826b78ff75d9e2b907c78e8a74c5742558a51926f97c","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-45e6f911ccb6786b6c1d39ff","nextBlock":"b-a0db837ffb1badd224f097d9","prerequisites":[{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-a0db837ffb1badd224f097d9","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-a0db837ffb1badd224f097d9","content":"主线阅读与优先实验","sha256":"42cba31deccba2a350b9b23bd0ca1dcaa4c4c5cb9d35c773e1fcb89124788ccd","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-c14cf67282b6e2fcb5ec8e61","nextBlock":"b-28552bdbe40668be834905ce","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-28552bdbe40668be834905ce","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-28552bdbe40668be834905ce","content":"标签供给、文本目标接口、视图/队列/MAE图与表示迁移证据","sha256":"e5185ec67dda4a18681db2d55dc5cbe0c312c9a5519949e3ded12e96d49c997f","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-a0db837ffb1badd224f097d9","nextBlock":"b-14261b12393b0dd45205b0d7","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-14261b12393b0dd45205b0d7","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-14261b12393b0dd45205b0d7","content":"**4×4，位置 0–15**","sha256":"dab1e775bbe722f2ff40501c3fe9af28013fd16eecdb6fb5667a6a0d711547f3","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-28552bdbe40668be834905ce","nextBlock":"b-bfaee36548edafd5d2198626","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-bfaee36548edafd5d2198626","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-bfaee36548edafd5d2198626","content":"![4×4，位置 0–15](/course/knowledge-assets/a3e27515290178f1ab6a68b1.jpg)","sha256":"eaf39b3c8b6c9e1fe40882fe5a4c71f1da47a23d77302c9afeb60d563776dd34","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-14261b12393b0dd45205b0d7","nextBlock":"b-10fbef8074ffbf55a8581c38","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-10fbef8074ffbf55a8581c38","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-10fbef8074ffbf55a8581c38","content":"**仅位置 0/3/9/14 可见**","sha256":"819ec8665bc0b217310168ce7d9474a46963c04e4a4df5c17d24c167a3696a42","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-bfaee36548edafd5d2198626","nextBlock":"b-517a38e50456c252e3a084c0","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-517a38e50456c252e3a084c0","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-517a38e50456c252e3a084c0","content":"![仅位置 0/3/9/14 可见](/course/knowledge-assets/1b0dd8453a17629d0f9bf823.jpg)","sha256":"fd47fe78c4c48c67be97d9fd2b1ebf960186c37faa1d1b30a00277b62bee7de8","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-10fbef8074ffbf55a8581c38","nextBlock":"b-5c3b17e43a89a2f14932dd04","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-5c3b17e43a89a2f14932dd04","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-5c3b17e43a89a2f14932dd04","content":"16 块位置示意：可见位置为 0、3、9、14；其余是要预测的位置。完整源图提供目标，图中不展示虚构模型补全。","sha256":"a4c984796646c7282f868779d1340c72a36696d18cad65cdeea7a54318a491ef","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-517a38e50456c252e3a084c0","nextBlock":"b-1744861d6b390ab4fd501a7a","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-1744861d6b390ab4fd501a7a","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-1744861d6b390ab4fd501a7a","content":"MAE 把图像分成 4×4 的 16 块。这个教学例子保留 4 块，遮住 12 块：**编码器只读取可见块及其位置**；轻量解码器将可见表示放回正确位置，再在缺失处插入 mask token，构成完整位置序列。隐藏位置的预测与原图相应位置比较，形成训练损失。训练之后主要保留编码器服务识别任务。[MAE](https://arxiv.org/abs/2111.06377)","sha256":"41f1852c4a7076365057faf0c061910b1cdfb6467f1bac36e502721c13dde45b","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-5c3b17e43a89a2f14932dd04","nextBlock":"b-7e71d514370e937a783597c1","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-7e71d514370e937a783597c1","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-7e71d514370e937a783597c1","content":"**先预测：** 如果把相同四块的位置信息拿走，解码器还知道它们来自左上还是右下吗？把可见集合改为相邻四块，对推测全图会有什么影响？","sha256":"b6785550af1ca90eee39a779b30b699f296c304901d2e2543d8db5ce06e47a57","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-1744861d6b390ab4fd501a7a","nextBlock":"b-69e62cad1e86f5fbf5c0ea28","prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-69e62cad1e86f5fbf5c0ea28","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239.html#b-69e62cad1e86f5fbf5c0ea28","content":"对比目标强调变换后的关系；遮挡目标要求可见内容支持缺失内容预测。两种路线都有可能得到可复用表示，也都有选择代价：强变换可能删掉任务线索，重建可能把资源用在下游并不关心的纹理上。","sha256":"8b3005e9f4ad6c069da988920a4232f225408f8e65d3d980260c4a0e52798a93","chapter":{"id":"course","title":"4×4，位置 0–15"},"unit":{"id":"notebook-blocks/fdd3ba8228064e708e554a6bf0bc3239","title":"4×4，位置 0–15"},"page":{"id":"reading-opening","title":"4×4，位置 0–15","number":1},"sectionTitle":"4×4，位置 0–15","currentQuestion":"4×4，位置 0–15","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-7e71d514370e937a783597c1","nextBlock":null,"prerequisites":[{"id":"b-c14cf67282b6e2fcb5ec8e61","url":"#b-c14cf67282b6e2fcb5ec8e61","role":"section-title"},{"id":"b-45e6f911ccb6786b6c1d39ff","url":"#b-45e6f911ccb6786b6c1d39ff","role":"problem-context"}],"relatedBlocks":[]}]}