{"id":"935f65f704b4d7c19f68413502e957c8237160a02eec4c73d53b2caac0e85ef9","title":"arXiv: ALD-Font separates content and style for few-shot Chinese font generation","argument":"Authors Jie Li, Suorong Yang, Jian Zhao, and Furao Shen propose ALD-Font, a model for generating Chinese characters in a target style from a handful of reference images. They argue that existing approaches can separate content and style at the feature level, but then allow the generator to combine them again, distorting character content or weakening style fidelity. ALD-Font instead uses an asymmetric design: SimSun-style glyphs provide content templates, while style features from the reference images guide generation as modulation signals. The authors say this architecture-level separation is theoretically validated. They also describe a parameter-efficient fine-tuning strategy that updates only style-related modules, intended to adapt to a new style without overfitting to the reference characters’ content. A Background Noise Removal module addresses noise in complex stroke regions, and the authors introduce Grey and OCR metrics to evaluate generated characters’ content quality. In their reported experiments, ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods. The abstract gives no numerical comparisons. The proposed design makes the separation of what a character says from how it looks an architectural choice rather than relying only on learned feature separation.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the arXiv abstract, which gives no numerical comparisons or detailed experimental results for the reported performance.","speaker":{"name":"Li, Jie, Yang, Suorong, Zhao, Jian, Shen, Furao","roleAtSource":"Article author"},"source":{"title":"Disentangle to create: Architecture-level disentangled diffusion model for few-shot Chinese font generation","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2602.18874","publishedAt":"2026-02-21","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose ALD-Font for generating Chinese characters in a target style from a small number of reference images.","excerpt":"Few-shot Chinese font generation aims to synthesize new characters in a target style using only a handful of reference images.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"},{"text":"The authors argue that feature-level disentanglement can be undone by the generator, causing content distortion and weaker style fidelity.","excerpt":"most existing approaches achieve only feature-level disentanglement and allow the generator to re-entangle these features, leading to content distortion and degraded style fidelity.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"},{"text":"ALD-Font uses SimSun-style glyphs as content templates and injects style features from target-style images as modulation signals.","excerpt":"The U-Net backbone takes SimSun-style glyphs as content templates, concatenated with noisy latents to construct a spatial canvas. Style features extracted from target-style images are injected as modulation signals to guide generation.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"},{"text":"The authors report that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","excerpt":"Experimental results show that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2602.18874","checkedAt":"2026-11-10T00:40:09.497Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What experimental comparisons and results support the authors’ report of significantly higher style fidelity with comparable content accuracy?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv: ALD-Font separates content and style for few-shot Chinese font generation","summary":"Authors Jie Li, Suorong Yang, Jian Zhao, and Furao Shen propose ALD-Font, a model for generating Chinese characters in a target style from a handful of reference images. They argue that existing approaches can separate content and style at the feature level, but then allow the generator to combine them again, distorting character content or weakening style fidelity. ALD-Font instead uses an asymmetric design: SimSun-style glyphs provide content templates, while style features from the reference images guide generation as modulation signals. The authors say this architecture-level separation is theoretically validated. They also describe a parameter-efficient fine-tuning strategy that updates only style-related modules, intended to adapt to a new style without overfitting to the reference characters’ content. A Background Noise Removal module addresses noise in complex stroke regions, and the authors introduce Grey and OCR metrics to evaluate generated characters’ content quality. In their reported experiments, ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods. The abstract gives no numerical comparisons. The proposed design makes the separation of what a character says from how it looks an architectural choice rather than relying only on learned feature separation.","question":"What experimental comparisons and results support the authors’ report of significantly higher style fidelity with comparable content accuracy?","claims":[{"text":"The authors propose ALD-Font for generating Chinese characters in a target style from a small number of reference images.","excerpt":"Few-shot Chinese font generation aims to synthesize new characters in a target style using only a handful of reference images.","support":"source-stated"},{"text":"The authors argue that feature-level disentanglement can be undone by the generator, causing content distortion and weaker style fidelity.","excerpt":"most existing approaches achieve only feature-level disentanglement and allow the generator to re-entangle these features, leading to content distortion and degraded style fidelity.","support":"source-stated"},{"text":"ALD-Font uses SimSun-style glyphs as content templates and injects style features from target-style images as modulation signals.","excerpt":"The U-Net backbone takes SimSun-style glyphs as content templates, concatenated with noisy latents to construct a spatial canvas. Style features extracted from target-style images are injected as modulation signals to guide generation.","support":"source-stated"},{"text":"The authors report that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","excerpt":"Experimental results show that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","support":"source-stated"}],"limitations":["The supplied evidence is the arXiv abstract, which gives no numerical comparisons or detailed experimental results for the reported performance."],"mediaId":"39a68bddc9632dc0995c4da0886a0f902fcc4d820510f0653472269dc9379a99","correctionReason":null,"id":"935f65f704b4d7c19f68413502e957c8237160a02eec4c73d53b2caac0e85ef9","publicationId":"83acd355-15fa-48f9-89f4-27bf765bc918","version":1,"sourceUrl":"https://arxiv.org/abs/2602.18874","sourceTitle":"Disentangle to create: Architecture-level disentangled diffusion model for few-shot Chinese font generation","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-02-21","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:03:30.119Z","createdAt":"2026-11-10T00:08:36.906Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"39a68bddc9632dc0995c4da0886a0f902fcc4d820510f0653472269dc9379a99","kind":"article","url":"https://arxiv.org/abs/2602.18874","sourceUrl":"https://arxiv.org/abs/2602.18874","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T14:37:59.956Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2602.18874","checkedAt":"2026-11-10T00:40:09.497Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"39a68bddc9632dc0995c4da0886a0f902fcc4d820510f0653472269dc9379a99","confidence":0.88,"provider":"typesafe","model":"jev-1.13.0","inputHash":"c595435e6bec1910f9c852eaf200a247220c70f8347ee431ed85b50ac7c5d082","decidedAt":"2026-11-09T14:38:00.929Z"},"provenance":{"textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observationId":"4927220c8e8d2596b9747366fd9e8fec4f79ef86e0c54c5fb91ea08e6324b76d","retrievedAt":"2026-11-09T14:37:59.833Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"53ae86a2-40ec-4655-8a4b-d9bc4dab5d52"},"reviewer":{"model":"gpt-6-luna","runId":"58000212-678b-41d9-85fa-6e300960586a"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T00:08:36.906Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Li, Jie","url":null,"kind":"unknown"},{"name":"Yang, Suorong","url":null,"kind":"unknown"},{"name":"Zhao, Jian","url":null,"kind":"unknown"},{"name":"Shen, Furao","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Li, Jie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Jie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}},{"actor":{"name":"Yang, Suorong","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yang, Suorong","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}},{"actor":{"name":"Zhao, Jian","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhao, Jian","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}},{"actor":{"name":"Shen, Furao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Shen, Furao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}}],"participationCoverage":{"total":4,"returned":4,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2602.18874","snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/935f65f704b4d7c19f68413502e957c8237160a02eec4c73d53b2caac0e85ef9/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-ald-font-separates-content-and-style-for-few-shot-chinese-935f65f7","formats":{"html":"/readings/arxiv-ald-font-separates-content-and-style-for-few-shot-chinese-935f65f7","markdown":"/readings/arxiv-ald-font-separates-content-and-style-for-few-shot-chinese-935f65f7.md","json":"/readings/arxiv-ald-font-separates-content-and-style-for-few-shot-chinese-935f65f7.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-zero-shot","name":"Zero-shot","explanation":"Doing a task without worked examples (zero-shot) or with only a few examples given in the prompt (few-shot).","referenceUrl":"https://developers.google.com/machine-learning/glossary#zero-shot-prompting"},{"id":"term-fine-tuning","name":"Fine-tuning","explanation":"Further training of an existing model to adapt it to a particular task or use.","referenceUrl":"https://developers.google.com/machine-learning/glossary#fine-tuning"},{"id":"term-overfitting","name":"Overfitting","explanation":"When a model learns its training examples too closely and performs worse on new data.","referenceUrl":"https://developers.google.com/machine-learning/glossary#overfitting"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"935f65f704b4d7c19f68413502e957c8237160a02eec4c73d53b2caac0e85ef9","revision":"83acd355-15fa-48f9-89f4-27bf765bc918","revisionKind":"publication-id","title":"arXiv: ALD-Font separates content and style for few-shot Chinese font generation","url":"https://escapemeridian.com/readings/arxiv-ald-font-separates-content-and-style-for-few-shot-chinese-935f65f7"},"source":{"type":"source","id":"https://arxiv.org/abs/2602.18874","revision":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","revisionKind":"sha256-source-text","title":"Disentangle to create: Architecture-level disentangled diffusion model for few-shot Chinese font generation","url":"https://arxiv.org/abs/2602.18874"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"935f65f704b4d7c19f68413502e957c8237160a02eec4c73d53b2caac0e85ef9","title":"arXiv: ALD-Font separates content and style for few-shot Chinese font generation","argument":"Authors Jie Li, Suorong Yang, Jian Zhao, and Furao Shen propose ALD-Font, a model for generating Chinese characters in a target style from a handful of reference images. They argue that existing approaches can separate content and style at the feature level, but then allow the generator to combine them again, distorting character content or weakening style fidelity. ALD-Font instead uses an asymmetric design: SimSun-style glyphs provide content templates, while style features from the reference images guide generation as modulation signals. The authors say this architecture-level separation is theoretically validated. They also describe a parameter-efficient fine-tuning strategy that updates only style-related modules, intended to adapt to a new style without overfitting to the reference characters’ content. A Background Noise Removal module addresses noise in complex stroke regions, and the authors introduce Grey and OCR metrics to evaluate generated characters’ content quality. In their reported experiments, ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods. The abstract gives no numerical comparisons. The proposed design makes the separation of what a character says from how it looks an architectural choice rather than relying only on learned feature separation.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the arXiv abstract, which gives no numerical comparisons or detailed experimental results for the reported performance.","speaker":{"name":"Li, Jie, Yang, Suorong, Zhao, Jian, Shen, Furao","roleAtSource":"Article author"},"source":{"title":"Disentangle to create: Architecture-level disentangled diffusion model for few-shot Chinese font generation","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2602.18874","publishedAt":"2026-02-21","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose ALD-Font for generating Chinese characters in a target style from a small number of reference images.","excerpt":"Few-shot Chinese font generation aims to synthesize new characters in a target style using only a handful of reference images.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"},{"text":"The authors argue that feature-level disentanglement can be undone by the generator, causing content distortion and weaker style fidelity.","excerpt":"most existing approaches achieve only feature-level disentanglement and allow the generator to re-entangle these features, leading to content distortion and degraded style fidelity.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"},{"text":"ALD-Font uses SimSun-style glyphs as content templates and injects style features from target-style images as modulation signals.","excerpt":"The U-Net backbone takes SimSun-style glyphs as content templates, concatenated with noisy latents to construct a spatial canvas. Style features extracted from target-style images are injected as modulation signals to guide generation.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"},{"text":"The authors report that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","excerpt":"Experimental results show that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","url":"https://arxiv.org/abs/2602.18874","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2602.18874","checkedAt":"2026-11-10T00:40:09.497Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What experimental comparisons and results support the authors’ report of significantly higher style fidelity with comparable content accuracy?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv: ALD-Font separates content and style for few-shot Chinese font generation","summary":"Authors Jie Li, Suorong Yang, Jian Zhao, and Furao Shen propose ALD-Font, a model for generating Chinese characters in a target style from a handful of reference images. They argue that existing approaches can separate content and style at the feature level, but then allow the generator to combine them again, distorting character content or weakening style fidelity. ALD-Font instead uses an asymmetric design: SimSun-style glyphs provide content templates, while style features from the reference images guide generation as modulation signals. The authors say this architecture-level separation is theoretically validated. They also describe a parameter-efficient fine-tuning strategy that updates only style-related modules, intended to adapt to a new style without overfitting to the reference characters’ content. A Background Noise Removal module addresses noise in complex stroke regions, and the authors introduce Grey and OCR metrics to evaluate generated characters’ content quality. In their reported experiments, ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods. The abstract gives no numerical comparisons. The proposed design makes the separation of what a character says from how it looks an architectural choice rather than relying only on learned feature separation.","question":"What experimental comparisons and results support the authors’ report of significantly higher style fidelity with comparable content accuracy?","claims":[{"text":"The authors propose ALD-Font for generating Chinese characters in a target style from a small number of reference images.","excerpt":"Few-shot Chinese font generation aims to synthesize new characters in a target style using only a handful of reference images.","support":"source-stated"},{"text":"The authors argue that feature-level disentanglement can be undone by the generator, causing content distortion and weaker style fidelity.","excerpt":"most existing approaches achieve only feature-level disentanglement and allow the generator to re-entangle these features, leading to content distortion and degraded style fidelity.","support":"source-stated"},{"text":"ALD-Font uses SimSun-style glyphs as content templates and injects style features from target-style images as modulation signals.","excerpt":"The U-Net backbone takes SimSun-style glyphs as content templates, concatenated with noisy latents to construct a spatial canvas. Style features extracted from target-style images are injected as modulation signals to guide generation.","support":"source-stated"},{"text":"The authors report that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","excerpt":"Experimental results show that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods.","support":"source-stated"}],"limitations":["The supplied evidence is the arXiv abstract, which gives no numerical comparisons or detailed experimental results for the reported performance."],"mediaId":"39a68bddc9632dc0995c4da0886a0f902fcc4d820510f0653472269dc9379a99","correctionReason":null,"id":"935f65f704b4d7c19f68413502e957c8237160a02eec4c73d53b2caac0e85ef9","publicationId":"83acd355-15fa-48f9-89f4-27bf765bc918","version":1,"sourceUrl":"https://arxiv.org/abs/2602.18874","sourceTitle":"Disentangle to create: Architecture-level disentangled diffusion model for few-shot Chinese font generation","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-02-21","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:03:30.119Z","createdAt":"2026-11-10T00:08:36.906Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"39a68bddc9632dc0995c4da0886a0f902fcc4d820510f0653472269dc9379a99","kind":"article","url":"https://arxiv.org/abs/2602.18874","sourceUrl":"https://arxiv.org/abs/2602.18874","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T14:37:59.956Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2602.18874","checkedAt":"2026-11-10T00:40:09.497Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"39a68bddc9632dc0995c4da0886a0f902fcc4d820510f0653472269dc9379a99","confidence":0.88,"provider":"typesafe","model":"jev-1.13.0","inputHash":"c595435e6bec1910f9c852eaf200a247220c70f8347ee431ed85b50ac7c5d082","decidedAt":"2026-11-09T14:38:00.929Z"},"provenance":{"textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observationId":"4927220c8e8d2596b9747366fd9e8fec4f79ef86e0c54c5fb91ea08e6324b76d","retrievedAt":"2026-11-09T14:37:59.833Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"53ae86a2-40ec-4655-8a4b-d9bc4dab5d52"},"reviewer":{"model":"gpt-6-luna","runId":"58000212-678b-41d9-85fa-6e300960586a"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T00:08:36.906Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Li, Jie","url":null,"kind":"unknown"},{"name":"Yang, Suorong","url":null,"kind":"unknown"},{"name":"Zhao, Jian","url":null,"kind":"unknown"},{"name":"Shen, Furao","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Li, Jie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Jie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}},{"actor":{"name":"Yang, Suorong","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yang, Suorong","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}},{"actor":{"name":"Zhao, Jian","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Zhao, Jian","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}},{"actor":{"name":"Shen, Furao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Shen, Furao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2602.18874","locator":"head/meta[name=author|citation_author]"},"assertionId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T00:40:09.497Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","sourceUrl":"https://arxiv.org/abs/2602.18874","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z"}}],"participationCoverage":{"total":4,"returned":4,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2602.18874","snapshotId":"eddfcef835e8c39401664a3153d26c1768fd44ef00192b05c45436c579506087","contentSha256":"ac3326c07339d3a954547a90fc42d54a8968442c5898b65fd72b01c1b0c29808","textHash":"dc2c5681c461c701f7daa7147e6d7e3c4d464bbf1e32cbaa606f1b089387893a","observedAt":"2026-11-10T00:40:09.497Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/935f65f704b4d7c19f68413502e957c8237160a02eec4c73d53b2caac0e85ef9/source"}}}],"audits":[{"publicationId":"83acd355-15fa-48f9-89f4-27bf765bc918","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading correctly credits ALD-Font to the four listed authors on arXiv. It keeps the method, the theoretical-validation claim and the experimental results attributed to them rather than presenting them as established facts.","excerpt":"Authors:Jie Li, Suorong Yang, Jian Zhao, Furao Shen"},{"dimension":"summary","status":"pass","explanation":"The title, summary and question faithfully convey the problem (few-shot font generation), the authors' critique of feature-level disentanglement, the asymmetric design, the PEFT and BNR components, the new metrics and the reported results. A lay reader can follow it. Technical terms such as SimSun, OCR and diffusion model can be covered by tooltips. The question is a genuine open issue, since the abstract gives no figures.","excerpt":"we propose the Architecture-Level Disentangled Diffusion Model (ALD-Font) with an asymmetric design"},{"dimension":"date","status":"pass","explanation":"The originalPublished date of 2026-02-21 matches the v1 submission date. The later revision (4 Nov 2026) and the feed date are correctly not used.","excerpt":"Submitted on 21 Feb 2026 (v1), last revised 4 Nov 2026 (this version, v2)"},{"dimension":"support","status":"pass","explanation":"Each claim matches the abstract verbatim in substance. The reading notes that the abstract gives no numbers and does not turn the reported results into proof of real-world reliability. The closing sentence restates the architecture-level framing and invents no benefits.","excerpt":"Experimental results show that ALD-Font achieves significantly higher style fidelity while maintaining comparable content accuracy to existing state-of-the-art methods."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct link to the arXiv abstract page, which is the original source. No better source media was found.","excerpt":null}],"createdAt":"2026-11-10T00:08:36.906Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}