{"version":2,"relationshipURLBase":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html","relationshipResolution":"Resolve relative relationship URLs against relationshipURLBase; complete records and absolute URLs are in blocks.","title":"学习路径： 图像怎样约束句子与回答","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html","markdown":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.md","pages":[{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","unit":"学习路径： 图像怎样约束句子与回答","unitId":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","chapter":"course","number":1}],"aliases":[{"oldId":"b-d5a93129c6d1f73416012bd2","targetId":"b-d5a93129c6d1f73416012bd2","page":"reading-opening","operation":"move","beforeSHA256":"db08960319942d9119c9cad2cf7d0b78a68c914f8e19ab3bd375531d27c96204","afterSHA256":"db08960319942d9119c9cad2cf7d0b78a68c914f8e19ab3bd375531d27c96204","reason":"Same source body in authored semantic page"},{"oldId":"b-71bce6de5e464b5588da8b94","targetId":"b-71bce6de5e464b5588da8b94","page":"reading-opening","operation":"move","beforeSHA256":"e5df180ed030b50c84a576c8911ba2a69c96ab192fcfac5e7686b63ab5af678a","afterSHA256":"e5df180ed030b50c84a576c8911ba2a69c96ab192fcfac5e7686b63ab5af678a","reason":"Same source body in authored semantic page"},{"oldId":"b-0b4a0711cf2ab09768666d4b","targetId":"b-0b4a0711cf2ab09768666d4b","page":"reading-opening","operation":"move","beforeSHA256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","afterSHA256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","reason":"Same source body in authored semantic page"},{"oldId":"b-6886195655be63a4ec6f227e","targetId":"b-6886195655be63a4ec6f227e","page":"reading-opening","operation":"move","beforeSHA256":"6a611aef342a7535362f318ee0c33b63170ce4ab853fa44a05f76a04873fd320","afterSHA256":"6a611aef342a7535362f318ee0c33b63170ce4ab853fa44a05f76a04873fd320","reason":"Same source body in authored semantic page"},{"oldId":"b-5755719121461e96a574df42","targetId":"b-5755719121461e96a574df42","page":"reading-opening","operation":"move","beforeSHA256":"a57513e59eb07960710e5ece7951cc57af4a94c446630eb5fc6d70b460982042","afterSHA256":"a57513e59eb07960710e5ece7951cc57af4a94c446630eb5fc6d70b460982042","reason":"Same source body in authored semantic page"},{"oldId":"b-190499b12a23be90b36dcc22","targetId":"b-190499b12a23be90b36dcc22","page":"reading-opening","operation":"move","beforeSHA256":"6d15dd7cd1e410c3b462ad12e5c096e99f52620e6ea410d32e1010982856d3fc","afterSHA256":"6d15dd7cd1e410c3b462ad12e5c096e99f52620e6ea410d32e1010982856d3fc","reason":"Same source body in authored semantic page"},{"oldId":"b-2133286774150c82a760a479","targetId":"b-2133286774150c82a760a479","page":"reading-opening","operation":"move","beforeSHA256":"936c1f791e72aa9829fba35a2f8a6cbed45e786138bccc4c8cbea1187ef81f38","afterSHA256":"936c1f791e72aa9829fba35a2f8a6cbed45e786138bccc4c8cbea1187ef81f38","reason":"Same source body in authored semantic page"},{"oldId":"b-873fa1d13f3faa3c5fb31d42","targetId":"b-873fa1d13f3faa3c5fb31d42","page":"reading-opening","operation":"move","beforeSHA256":"dbd8267617e65afb9b361c38ab5ecd9deb08703c519c39e9a9fa45ca552b9794","afterSHA256":"dbd8267617e65afb9b361c38ab5ecd9deb08703c519c39e9a9fa45ca552b9794","reason":"Same source body in authored semantic page"},{"oldId":"b-eda48fefd5ba55ccc9eb849a","targetId":"b-eda48fefd5ba55ccc9eb849a","page":"reading-opening","operation":"move","beforeSHA256":"890fbc23abc501b6a296dd283f5d91269fcdc5fe37222a92fa880db818085b66","afterSHA256":"f6aedf109a28f615cb59b69833b6929e30ff48023b1eabc37990766d7cfd0565","reason":"Same source body in authored semantic page"},{"oldId":"b-fde7905d9e7a204d35ea2006","targetId":"b-fde7905d9e7a204d35ea2006","page":"reading-opening","operation":"move","beforeSHA256":"1149558f23d8ace57289208a1034f60a2abcfdd08f4c6474343263d32efe6c1e","afterSHA256":"1149558f23d8ace57289208a1034f60a2abcfdd08f4c6474343263d32efe6c1e","reason":"Same source body in authored semantic page"}],"blocks":[{"id":"b-d5a93129c6d1f73416012bd2","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-d5a93129c6d1f73416012bd2","content":"[LAB 13 · 看图生成一句话：训练、解码与消融](/course/notebooks/3da47f3ee3554ee0a36e0bb6a2fdc61f.html)","sha256":"db08960319942d9119c9cad2cf7d0b78a68c914f8e19ab3bd375531d27c96204","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-71bce6de5e464b5588da8b94","prerequisites":[],"relatedBlocks":[]},{"id":"b-71bce6de5e464b5588da8b94","type":"h1","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-71bce6de5e464b5588da8b94","content":"# 学习路径： 图像怎样约束句子与回答","sha256":"e5df180ed030b50c84a576c8911ba2a69c96ab192fcfac5e7686b63ab5af678a","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-d5a93129c6d1f73416012bd2","nextBlock":"b-0b4a0711cf2ab09768666d4b","prerequisites":[{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-0b4a0711cf2ab09768666d4b","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-0b4a0711cf2ab09768666d4b","content":"按问题继续","sha256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-71bce6de5e464b5588da8b94","nextBlock":"b-6886195655be63a4ec6f227e","prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-6886195655be63a4ec6f227e","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-6886195655be63a4ec6f227e","content":"真实CNN–GRU条件训练与greedy/beam生成、遮图/错图消融","sha256":"6a611aef342a7535362f318ee0c33b63170ce4ab853fa44a05f76a04873fd320","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-0b4a0711cf2ab09768666d4b","nextBlock":"b-5755719121461e96a574df42","prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-5755719121461e96a574df42","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-5755719121461e96a574df42","content":"**学习路径：** [图像怎样约束句子与回答](/course/study/caption-vqa.html)","sha256":"a57513e59eb07960710e5ece7951cc57af4a94c446630eb5fc6d70b460982042","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-6886195655be63a4ec6f227e","nextBlock":"b-190499b12a23be90b36dcc22","prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-190499b12a23be90b36dcc22","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-190499b12a23be90b36dcc22","content":"**同样以`BOS a`开头，换一张图能否生成不同句子？** 先读[P51](/course/pages/l3-r2-p51.html)与[逐词目标表](/course/blocks/caption-shift-and-stop.html)，再对照本册第2节输入/标签、第3节视觉进入路径、第4/5节实际生成和消融。","sha256":"6d15dd7cd1e410c3b462ad12e5c096e99f52620e6ea410d32e1010982856d3fc","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-5755719121461e96a574df42","nextBlock":"b-2133286774150c82a760a479","prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-2133286774150c82a760a479","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-2133286774150c82a760a479","content":"本册确实用RGB图像训练CNN＋GRU并接续自身输出。2026-09-22本地基准保存了正确图、像素置零和循环错图的结果；18种语义组合在训练/保留两侧都出现，检验的是新渲染样本，不是未见组合或自然照片泛化。[结果和完整协议](/course/blocks/caption-visual-controls.html)。","sha256":"936c1f791e72aa9829fba35a2f8a6cbed45e786138bccc4c8cbea1187ef81f38","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-190499b12a23be90b36dcc22","nextBlock":"b-873fa1d13f3faa3c5fb31d42","prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-873fa1d13f3faa3c5fb31d42","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-873fa1d13f3faa3c5fb31d42","content":"读完应能拿一组图片与生成句子说明视觉是否影响输出，以及为何teacher-forced交叉熵不能替代生成评价。动手复现时按原代码顺序执行；GRU内部与beam搜索可留后续课，不影响先理解图像条件。另一个输出接口见[LAB 14 问答](https://deepnote.com/project/38755e4e-23f6-4f5f-8c68-8a2e4114b1c1/notebook/11d9f74871544a77b42024a9fe06b81a)。","sha256":"dbd8267617e65afb9b361c38ab5ecd9deb08703c519c39e9a9fa45ca552b9794","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-2133286774150c82a760a479","nextBlock":"b-eda48fefd5ba55ccc9eb849a","prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-eda48fefd5ba55ccc9eb849a","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-eda48fefd5ba55ccc9eb849a","content":"![固定前缀下的真实生成](/assets/display/9ebe76d49a449b9940b6470f.webp)","sha256":"f6aedf109a28f615cb59b69833b6929e30ff48023b1eabc37990766d7cfd0565","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-873fa1d13f3faa3c5fb31d42","nextBlock":"b-fde7905d9e7a204d35ea2006","prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-fde7905d9e7a204d35ea2006","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/71a055b4f2af4411829bd08c5f697050.html#b-fde7905d9e7a204d35ea2006","content":"图中句子由已训练模型生成。下一项问题是“同图换问会怎样”，进入 [LAB 14 · 图像问答：视觉条件与问题先验](https://deepnote.com/workspace/ise3309jingao-5ba66b0a-4283-4f5a-ae06-3f68d3201e99/project/38755e4e-23f6-4f5f-8c68-8a2e4114b1c1/notebook/11d9f74871544a77b42024a9fe06b81a?utm_source=openai&utm_medium=mcp&utm_campaign=openaimcp&utm_content=11d9f74871544a77b42024a9fe06b81a&utm_term=get_notebook)。","sha256":"1149558f23d8ace57289208a1034f60a2abcfdd08f4c6474343263d32efe6c1e","chapter":{"id":"course","title":"学习路径： 图像怎样约束句子与回答"},"unit":{"id":"notebook-blocks/71a055b4f2af4411829bd08c5f697050","title":"学习路径： 图像怎样约束句子与回答"},"page":{"id":"reading-opening","title":"学习路径： 图像怎样约束句子与回答","number":1},"sectionTitle":"学习路径： 图像怎样约束句子与回答","currentQuestion":"学习路径： 图像怎样约束句子与回答","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-eda48fefd5ba55ccc9eb849a","nextBlock":null,"prerequisites":[{"id":"b-71bce6de5e464b5588da8b94","url":"#b-71bce6de5e464b5588da8b94","role":"section-title"},{"id":"b-d5a93129c6d1f73416012bd2","url":"#b-d5a93129c6d1f73416012bd2","role":"problem-context"}],"relatedBlocks":[]}]}
