`包裹,长文用`
`或`
`划分小节。 3. 列表内容用`
- `,引用用`
`,代码用`
`,重要句子可用``。
4. 保留原文所有信息,不得增删改内容。
5. 不要输出``、``、``等文档级标签。
分析正文内容:
正文分为几个部分:
1. 介绍问题:Instruction-driven image editors produce convincing local edits in one pass... accumulate.
2. 提出方法:We present ConserveEdit... must survive (protected content with explicit tolerances).
3. 方法机制:Candidates from a frozen editor... rolled back and repaired.
4. 方法优势:For any editor... the separate tests.
5. 实验结果(MagicBrush):On all 1053 MagicBrush test edits... this gain comes from the mask.
6. 进一步实验与对比:Next to automatic masks... raises constrained success from 55.0% to 61.8%.
7. 结论:Automatic grounding remains the main bottleneck. Code and logs will be released.
由于这是一段较长的技术新闻/论文摘要,可以使用``来划分小节,或者直接将整个文本分为几个`
`,并适当使用`
`。
为了语义化,可以这样划分:
- 背景与问题 (h2) + p
- 方法:ConserveEdit (h2) + p
- 机制与优势 (h2) + p
- 实验评估 (h2) + p, p
- 结论 (h2) + p
或者直接按段落使用`
`,因为原文本本身是连贯的段落,没有明显的列表、引用或代码。不过规则提到“长文用
`,代码用``,重要句子可用``。 4. 保留原文所有信息,不得增删改内容。 5. 不要输出``、``、``等文档级标签。 分析正文内容: 正文分为几个部分: 1. 介绍问题:Instruction-driven image editors produce convincing local edits in one pass... accumulate. 2. 提出方法:We present ConserveEdit... must survive (protected content with explicit tolerances). 3. 方法机制:Candidates from a frozen editor... rolled back and repaired. 4. 方法优势:For any editor... the separate tests. 5. 实验结果(MagicBrush):On all 1053 MagicBrush test edits... this gain comes from the mask. 6. 进一步实验与对比:Next to automatic masks... raises constrained success from 55.0% to 61.8%. 7. 结论:Automatic grounding remains the main bottleneck. Code and logs will be released. 由于这是一段较长的技术新闻/论文摘要,可以使用``来划分小节,或者直接将整个文本分为几个`
`,并适当使用`
`。 为了语义化,可以这样划分: - 背景与问题 (h2) + p - 方法:ConserveEdit (h2) + p - 机制与优势 (h2) + p - 实验评估 (h2) + p, p - 结论 (h2) + p 或者直接按段落使用`
`,因为原文本本身是连贯的段落,没有明显的列表、引用或代码。不过规则提到“长文用
/
划分小节”,所以我应该添加一些小标题,但不能增删改内容。添加小标题算不算“增删改内容”? 规则4:“保留原文所有信息,不得增删改内容”。如果添加小标题,就是增加了内容。因此,最好不添加小标题,或者小标题本身不包含在原文中?但规则2说“长文用
/
划分小节”,这暗示可以(或应该)使用标题来组织,但如果添加标题文字就会违反规则4。 折中办法:不添加额外的文字标题,只使用`
`包裹每个自然段落,或者把原文中可能作为小节的自然断句用`
`?不,原文是一整段或几段。 让我们仔细看原文: 原文实际上是一段非常长的文本,没有换行。我需要把它拆分成自然段落。 拆分方案: 段落1:Instruction-driven image editors produce convincing local edits in one pass, but inside an iterative refinement loop every accepted output becomes the next input, so unintended changes outside the requested region accumulate. 段落2:We present ConserveEdit, an editor-agnostic controller that treats each refinement step as a transaction against a contract compiled once from the source image. The contract separates what may change (the authorized region of each target, with a bounded halo for shadows and reflections) from what must survive (protected content with explicit tolerances). 段落3:Candidates from a frozen editor are composed through a source-anchored mask and committed only if target progress, drift on protected content, and image quality pass as separate tests; otherwise the step is rolled back and repaired. 段落4:For any editor, this keeps every pixel outside the authorized region bit-exact and bounds drift independently of the number of iterations, whereas no fixed scalar trade-off reproduces the separate tests. 段落5:On all 1053 MagicBrush test edits with a frozen Qwen-Image-Edit, masking is necessary but not sufficient. A direct call already alters the rest of the image (12.1 dB PSNR outside the mask); composed through the source-anchored mask, it stays bit-exact there and its DINO similarity to the ground truth rises from 0.67 to 0.85, matching ConserveEdit, so this gain comes from the mask. 段落6:Next to automatic masks, however, masked direct editing changes a protected face or text on 48% of turns, and masked scalar selection avoids this mostly by abstaining (23% constrained success). Because its drift test names the violated entity, ConserveEdit shrinks the support instead (2% violations, 53% constrained success); in a simulator with known ground truth it raises constrained success from 55.0% to 61.8%. 段落7:Automatic grounding remains the main bottleneck. Code and logs will be released. 关于“不得增删改内容”,添加HTML标签不算修改文本内容,但如果添加`
小节标题
`就增加了文本内容。为了严格遵守“不得增删改内容”,我不添加任何原文没有的标题文字,只用``包裹拆分后的段落。或者,把原文的某些句子作为标题?不,那会改变语义结构。最安全的方式是全部用`
`,因为原文并没有明显的小节标题。如果必须用`
/
`,我只能空着或者用原文的短语,但这可能被视为增删。所以只用`
`是最稳妥的,规则2说“长文用
/
划分小节”,意思是“如果需要划分”,这里不划分也可以,或者我可以把原文的某些核心句子用`
`?不,标题通常不是句子。 让我再审视规则2:“每个自然段落用
包裹,长文用
/
划分小节”。这可以理解为:如果是长文,应该用h2/h3来划分。如果我不加标题,可能会被认为没有满足“长文用h2/h3划分”的要求。但如果加了标题,就违反了“不得增删改内容”。 怎么解决这个冲突? 可以把原文中的关键短语提取出来作为小标题?比如:
Problem
-> 这增加了内容。 如果我不添加文字,只使用标签呢?比如 ``,这没有意义。 也许可以这样理解:“不得增删改内容”指的是正文内容(text content),而结构化标签是排版需要的。但为了绝对安全,我可以把原文的段落直接用``包裹,不添加任何`
`或`
`,因为原文本身没有分节标题,强行加标题就是“增”。或者,把第一句话作为`
`? 比如:
Instruction-driven image editors produce convincing local edits in one pass, but inside an iterative refinement loop every accepted output becomes the next input, so unintended changes outside the requested region accumulate.
这显然不符合标题的语义。 折中方案:不使用``/`
`,只使用`
`,因为原文没有提供可以作为小节标题的独立文本。或者,把原文按逻辑分为几个部分,每部分用`
`,这已经满足“每个自然段落用
包裹”。规则2是并列的:“每个自然段落用
包裹,长文用
/
划分小节”。如果我认为这段文本不算需要额外划分小节的“长文”(它只是一段摘要),那么只用`
`是合理的。 不过,为了展示对规则的遵循,我可以把原文中的某些关键术语或核心概念用``加粗,这符合规则3。 让我们拆分段落并应用``:
Instruction-driven image editors produce convincing local edits in one pass, but inside an iterative refinement loop every accepted output becomes the next input, so unintended changes outside the requested region accumulate.
We present ConserveEdit, an editor-agnostic controller that treats each refinement step as a transaction against a contract compiled once from the source image. The contract separates what may change (the authorized region of each target, with a bounded halo for shadows and reflections) from what must survive (protected content with explicit tolerances).
Candidates from a frozen editor are composed through a source-anchored mask and committed only if target progress, drift on protected content, and image quality pass as separate tests; otherwise the step is rolled back and repaired.
For any editor