{"id":"2ecbaaa36b11d8d78ca695a4d910a8346461738ca49ba4cb963f5efd6d41fa81","title":"Wu, Guo and Xia propose input-adaptive computation for frozen LLMs","argument":"Junxi Wu, Yiwen Guo and Shu-Tao Xia introduce SelfMorph, a method that lets a frozen LLM choose how to organize its internal computation for each prompt. Rather than only generating more tokens, it selects among the original forward path, serial reuse of a layer span, or local branching and merging of perturbed hidden states before generating a response.\n\nThe authors say these options have complementary benefits, so a lightweight selector reads the model’s depth-wise hidden state trajectory on the prompt and chooses a computation shape. They report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks, with the strongest performance in 24 of 27 settings.\n\nThe idea is to adapt not just how much computation a model uses, but its structure. The abstract reports benchmark results, not evidence about performance or cost in deployed systems. How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The evidence provided is the abstract, which does not describe benchmark tasks, comparison methods, or how computational cost was measured. The abstract reports benchmark evaluations, not results from deployed systems or real-world tasks. Term-library review may be needed for SelfMorph, test-time scaling, depth-wise hidden state trajectory, and other technical terms in the reading.","speaker":{"name":"Wu, Junxi, Guo, Yiwen, Xia, Shu-Tao","roleAtSource":"Article author"},"source":{"title":"Compute Has Shape: Self-Morphing the Internal Computation of Frozen LLMs","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.38382","publishedAt":"2026-10-28","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors introduce SelfMorph, which makes the shape of a frozen LLM’s internal computation an input-dependent inference-time choice.","excerpt":"We introduce SelfMorph, which makes computational shape an input-dependent inference-time decision.","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"SelfMorph selects among the original forward path, serial reuse of a layer span, and local branching and merging of perturbed hidden states.","excerpt":"It unifies the original forward path (BASE), serial reuse of a layer span (REPEAT), and local branching and merging of perturbed hidden states (BRANCH)","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"The authors report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","excerpt":"SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"The authors report that SelfMorph had the strongest performance in 24 of 27 settings.","excerpt":"It achieves the strongest performance in 24 of 27 settings.","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"The abstract reports benchmark results, but does not establish performance or cost in deployed systems.","excerpt":"Evaluated on three LLMs spanning 4B-27B, SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","url":"https://arxiv.org/abs/2610.38382","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38382","checkedAt":"2026-11-13T05:42:08.813Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Wu, Guo and Xia propose input-adaptive computation for frozen LLMs","summary":"Junxi Wu, Yiwen Guo and Shu-Tao Xia introduce SelfMorph, a method that lets a frozen LLM choose how to organize its internal computation for each prompt. Rather than only generating more tokens, it selects among the original forward path, serial reuse of a layer span, or local branching and merging of perturbed hidden states before generating a response.\n\nThe authors say these options have complementary benefits, so a lightweight selector reads the model’s depth-wise hidden state trajectory on the prompt and chooses a computation shape. They report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks, with the strongest performance in 24 of 27 settings.\n\nThe idea is to adapt not just how much computation a model uses, but its structure. The abstract reports benchmark results, not evidence about performance or cost in deployed systems. How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?","question":"How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?","claims":[{"text":"The authors introduce SelfMorph, which makes the shape of a frozen LLM’s internal computation an input-dependent inference-time choice.","excerpt":"We introduce SelfMorph, which makes computational shape an input-dependent inference-time decision.","support":"source-stated"},{"text":"SelfMorph selects among the original forward path, serial reuse of a layer span, and local branching and merging of perturbed hidden states.","excerpt":"It unifies the original forward path (BASE), serial reuse of a layer span (REPEAT), and local branching and merging of perturbed hidden states (BRANCH)","support":"source-stated"},{"text":"The authors report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","excerpt":"SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","support":"source-stated"},{"text":"The authors report that SelfMorph had the strongest performance in 24 of 27 settings.","excerpt":"It achieves the strongest performance in 24 of 27 settings.","support":"source-stated"},{"text":"The abstract reports benchmark results, but does not establish performance or cost in deployed systems.","excerpt":"Evaluated on three LLMs spanning 4B-27B, SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","support":"interpretation"}],"limitations":["The evidence provided is the abstract, which does not describe benchmark tasks, comparison methods, or how computational cost was measured.","The abstract reports benchmark evaluations, not results from deployed systems or real-world tasks.","Term-library review may be needed for SelfMorph, test-time scaling, depth-wise hidden state trajectory, and other technical terms in the reading."],"mediaId":"fad3a76801406086f19b1fb72a74ffb9efae1d67ff6a30da82569ee3b425eb79","correctionReason":null,"id":"2ecbaaa36b11d8d78ca695a4d910a8346461738ca49ba4cb963f5efd6d41fa81","publicationId":"41ddf19b-ea57-4642-8d7b-6ecf0eb8cc7a","version":1,"sourceUrl":"https://arxiv.org/abs/2610.38382","sourceTitle":"Compute Has Shape: Self-Morphing the Internal Computation of Frozen LLMs","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-28","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-30T06:45:35.314Z","createdAt":"2026-11-13T03:35:40.411Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"fad3a76801406086f19b1fb72a74ffb9efae1d67ff6a30da82569ee3b425eb79","kind":"article","url":"https://arxiv.org/abs/2610.38382","sourceUrl":"https://arxiv.org/abs/2610.38382","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-31T05:17:59.426Z","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/2610.38382","checkedAt":"2026-11-13T05:42:08.813Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"fad3a76801406086f19b1fb72a74ffb9efae1d67ff6a30da82569ee3b425eb79","confidence":0.89,"provider":"typesafe","model":"jev-1.13.0","inputHash":"e873e5c6f2db0a81f7a136e19d1c86400f6837a947d03520eff25450b9e795c3","decidedAt":"2026-10-31T05:18:02.886Z"},"provenance":{"textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observationId":"128846a9a1ee387cbc904afa61d0734778eb48ac3e2238a0989c3b652943c5f1","retrievedAt":"2026-10-31T05:17:56.337Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"39b27e82-de87-405c-81ca-84ef0d4c5974"},"reviewer":{"model":"gpt-6-luna","runId":"cf5402ae-9134-4bea-8ba5-dfd70192f7d2"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-13T03:35:40.411Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Wu, Junxi","url":null,"kind":"unknown"},{"name":"Guo, Yiwen","url":null,"kind":"unknown"},{"name":"Xia, Shu-Tao","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Wu, Junxi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wu, Junxi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38382","locator":"head/meta[name=author|citation_author]"},"assertionId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T05:42:08.813Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","sourceUrl":"https://arxiv.org/abs/2610.38382","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z"}},{"actor":{"name":"Guo, Yiwen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Guo, Yiwen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38382","locator":"head/meta[name=author|citation_author]"},"assertionId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T05:42:08.813Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","sourceUrl":"https://arxiv.org/abs/2610.38382","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z"}},{"actor":{"name":"Xia, Shu-Tao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Xia, Shu-Tao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38382","locator":"head/meta[name=author|citation_author]"},"assertionId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T05:42:08.813Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","sourceUrl":"https://arxiv.org/abs/2610.38382","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z"}}],"participationCoverage":{"total":3,"returned":3,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.38382","snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z","basis":"publisher-page-metadata","sourceLinks":42,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/2ecbaaa36b11d8d78ca695a4d910a8346461738ca49ba4cb963f5efd6d41fa81/source"}},"canonicalUrl":"https://escapemeridian.com/readings/wu-guo-and-xia-propose-input-adaptive-computation-for-frozen-2ecbaaa3","formats":{"html":"/readings/wu-guo-and-xia-propose-input-adaptive-computation-for-frozen-2ecbaaa3","markdown":"/readings/wu-guo-and-xia-propose-input-adaptive-computation-for-frozen-2ecbaaa3.md","json":"/readings/wu-guo-and-xia-propose-input-adaptive-computation-for-frozen-2ecbaaa3.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-llm","name":"Large language model","explanation":"A language model with many learned numerical settings, used to process and generate language.","referenceUrl":"https://developers.google.com/machine-learning/glossary#large-language-model"},{"id":"term-benchmark","name":"Benchmark","explanation":"A defined set of tests used to compare systems. Results depend on the tasks and measures selected.","referenceUrl":"https://arxiv.org/abs/2211.09110"},{"id":"term-inference","name":"Model inference","explanation":"Using a trained model to produce a prediction or response to new input.","referenceUrl":"https://developers.google.com/machine-learning/glossary#inference"}],"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"},{"id":"time","name":"TIME","kind":"organization","context":"TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.","referenceUrls":["https://time.com/"],"path":"/entities/time","json":"/entities/time.json","markdown":"/entities/time.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"2ecbaaa36b11d8d78ca695a4d910a8346461738ca49ba4cb963f5efd6d41fa81","revision":"41ddf19b-ea57-4642-8d7b-6ecf0eb8cc7a","revisionKind":"publication-id","title":"Wu, Guo and Xia propose input-adaptive computation for frozen LLMs","url":"https://escapemeridian.com/readings/wu-guo-and-xia-propose-input-adaptive-computation-for-frozen-2ecbaaa3"},"source":{"type":"source","id":"https://arxiv.org/abs/2610.38382","revision":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","revisionKind":"sha256-source-text","title":"Compute Has Shape: Self-Morphing the Internal Computation of Frozen LLMs","url":"https://arxiv.org/abs/2610.38382"},"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":"2ecbaaa36b11d8d78ca695a4d910a8346461738ca49ba4cb963f5efd6d41fa81","title":"Wu, Guo and Xia propose input-adaptive computation for frozen LLMs","argument":"Junxi Wu, Yiwen Guo and Shu-Tao Xia introduce SelfMorph, a method that lets a frozen LLM choose how to organize its internal computation for each prompt. Rather than only generating more tokens, it selects among the original forward path, serial reuse of a layer span, or local branching and merging of perturbed hidden states before generating a response.\n\nThe authors say these options have complementary benefits, so a lightweight selector reads the model’s depth-wise hidden state trajectory on the prompt and chooses a computation shape. They report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks, with the strongest performance in 24 of 27 settings.\n\nThe idea is to adapt not just how much computation a model uses, but its structure. The abstract reports benchmark results, not evidence about performance or cost in deployed systems. How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The evidence provided is the abstract, which does not describe benchmark tasks, comparison methods, or how computational cost was measured. The abstract reports benchmark evaluations, not results from deployed systems or real-world tasks. Term-library review may be needed for SelfMorph, test-time scaling, depth-wise hidden state trajectory, and other technical terms in the reading.","speaker":{"name":"Wu, Junxi, Guo, Yiwen, Xia, Shu-Tao","roleAtSource":"Article author"},"source":{"title":"Compute Has Shape: Self-Morphing the Internal Computation of Frozen LLMs","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.38382","publishedAt":"2026-10-28","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors introduce SelfMorph, which makes the shape of a frozen LLM’s internal computation an input-dependent inference-time choice.","excerpt":"We introduce SelfMorph, which makes computational shape an input-dependent inference-time decision.","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"SelfMorph selects among the original forward path, serial reuse of a layer span, and local branching and merging of perturbed hidden states.","excerpt":"It unifies the original forward path (BASE), serial reuse of a layer span (REPEAT), and local branching and merging of perturbed hidden states (BRANCH)","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"The authors report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","excerpt":"SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"The authors report that SelfMorph had the strongest performance in 24 of 27 settings.","excerpt":"It achieves the strongest performance in 24 of 27 settings.","url":"https://arxiv.org/abs/2610.38382","relationship":"Source-stated"},{"text":"The abstract reports benchmark results, but does not establish performance or cost in deployed systems.","excerpt":"Evaluated on three LLMs spanning 4B-27B, SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","url":"https://arxiv.org/abs/2610.38382","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38382","checkedAt":"2026-11-13T05:42:08.813Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Wu, Guo and Xia propose input-adaptive computation for frozen LLMs","summary":"Junxi Wu, Yiwen Guo and Shu-Tao Xia introduce SelfMorph, a method that lets a frozen LLM choose how to organize its internal computation for each prompt. Rather than only generating more tokens, it selects among the original forward path, serial reuse of a layer span, or local branching and merging of perturbed hidden states before generating a response.\n\nThe authors say these options have complementary benefits, so a lightweight selector reads the model’s depth-wise hidden state trajectory on the prompt and chooses a computation shape. They report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks, with the strongest performance in 24 of 27 settings.\n\nThe idea is to adapt not just how much computation a model uses, but its structure. The abstract reports benchmark results, not evidence about performance or cost in deployed systems. How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?","question":"How well does the selector’s benchmark advantage hold up across real-world tasks and inference costs?","claims":[{"text":"The authors introduce SelfMorph, which makes the shape of a frozen LLM’s internal computation an input-dependent inference-time choice.","excerpt":"We introduce SelfMorph, which makes computational shape an input-dependent inference-time decision.","support":"source-stated"},{"text":"SelfMorph selects among the original forward path, serial reuse of a layer span, and local branching and merging of perturbed hidden states.","excerpt":"It unifies the original forward path (BASE), serial reuse of a layer span (REPEAT), and local branching and merging of perturbed hidden states (BRANCH)","support":"source-stated"},{"text":"The authors report an 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","excerpt":"SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","support":"source-stated"},{"text":"The authors report that SelfMorph had the strongest performance in 24 of 27 settings.","excerpt":"It achieves the strongest performance in 24 of 27 settings.","support":"source-stated"},{"text":"The abstract reports benchmark results, but does not establish performance or cost in deployed systems.","excerpt":"Evaluated on three LLMs spanning 4B-27B, SelfMorph achieves a 8.06% mean relative improvement on five in-distribution benchmarks and 4.45% on four unseen benchmarks.","support":"interpretation"}],"limitations":["The evidence provided is the abstract, which does not describe benchmark tasks, comparison methods, or how computational cost was measured.","The abstract reports benchmark evaluations, not results from deployed systems or real-world tasks.","Term-library review may be needed for SelfMorph, test-time scaling, depth-wise hidden state trajectory, and other technical terms in the reading."],"mediaId":"fad3a76801406086f19b1fb72a74ffb9efae1d67ff6a30da82569ee3b425eb79","correctionReason":null,"id":"2ecbaaa36b11d8d78ca695a4d910a8346461738ca49ba4cb963f5efd6d41fa81","publicationId":"41ddf19b-ea57-4642-8d7b-6ecf0eb8cc7a","version":1,"sourceUrl":"https://arxiv.org/abs/2610.38382","sourceTitle":"Compute Has Shape: Self-Morphing the Internal Computation of Frozen LLMs","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-28","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-30T06:45:35.314Z","createdAt":"2026-11-13T03:35:40.411Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"fad3a76801406086f19b1fb72a74ffb9efae1d67ff6a30da82569ee3b425eb79","kind":"article","url":"https://arxiv.org/abs/2610.38382","sourceUrl":"https://arxiv.org/abs/2610.38382","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-31T05:17:59.426Z","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/2610.38382","checkedAt":"2026-11-13T05:42:08.813Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"fad3a76801406086f19b1fb72a74ffb9efae1d67ff6a30da82569ee3b425eb79","confidence":0.89,"provider":"typesafe","model":"jev-1.13.0","inputHash":"e873e5c6f2db0a81f7a136e19d1c86400f6837a947d03520eff25450b9e795c3","decidedAt":"2026-10-31T05:18:02.886Z"},"provenance":{"textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observationId":"128846a9a1ee387cbc904afa61d0734778eb48ac3e2238a0989c3b652943c5f1","retrievedAt":"2026-10-31T05:17:56.337Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"39b27e82-de87-405c-81ca-84ef0d4c5974"},"reviewer":{"model":"gpt-6-luna","runId":"cf5402ae-9134-4bea-8ba5-dfd70192f7d2"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-13T03:35:40.411Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Wu, Junxi","url":null,"kind":"unknown"},{"name":"Guo, Yiwen","url":null,"kind":"unknown"},{"name":"Xia, Shu-Tao","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Wu, Junxi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wu, Junxi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38382","locator":"head/meta[name=author|citation_author]"},"assertionId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T05:42:08.813Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","sourceUrl":"https://arxiv.org/abs/2610.38382","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z"}},{"actor":{"name":"Guo, Yiwen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Guo, Yiwen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38382","locator":"head/meta[name=author|citation_author]"},"assertionId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T05:42:08.813Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","sourceUrl":"https://arxiv.org/abs/2610.38382","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z"}},{"actor":{"name":"Xia, Shu-Tao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Xia, Shu-Tao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38382","locator":"head/meta[name=author|citation_author]"},"assertionId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T05:42:08.813Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","sourceUrl":"https://arxiv.org/abs/2610.38382","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z"}}],"participationCoverage":{"total":3,"returned":3,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.38382","snapshotId":"632b3a020cd57c376e204745c822b466b50b6156f6313e23db9c90d1d1a31c56","contentSha256":"cbaa55b6b44c5aadf75029ce4f196be46c2550159b7890ec9c92fae2ab8dcb9e","textHash":"d9d839713d738db8550cd101fc98e0b741560f915f88c2bdc1d682eb430ba007","observedAt":"2026-11-13T05:42:08.813Z","basis":"publisher-page-metadata","sourceLinks":42,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/2ecbaaa36b11d8d78ca695a4d910a8346461738ca49ba4cb963f5efd6d41fa81/source"}}}],"audits":[{"publicationId":"41ddf19b-ea57-4642-8d7b-6ecf0eb8cc7a","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading correctly credits Junxi Wu, Yiwen Guo and Shu-Tao Xia as the authors. It keeps SelfMorph's design and results framed as the authors' introduction and reported findings rather than as established fact.","excerpt":"Authors:Junxi Wu, Yiwen Guo, Shu-Tao Xia"},{"dimension":"summary","status":"uncertain","explanation":"The core idea is faithfully stated: the shape of computation is chosen per input, beyond generating more tokens. But central terms are left unexplained for a general reader, including frozen LLM (weights unchanged), hidden states, layer span and in-distribution, and no tooltip glosses are shown. The reading also omits two things. First, the selector is trained to weigh task performance against computational cost, which bears directly on the closing question. Second, the tests covered three LLMs from 4B to 27B parameters. Smallest repair: briefly gloss frozen and in-distribution, and add one clause each on the cost-aware selector training and the tested model range.","excerpt":"We train the selector by distilling utility-based distribution from offline morphing that jointly account for task performance and computational cost."},{"dimension":"date","status":"pass","explanation":"The originalPublished date, 2026-10-28, matches the arXiv v1 submission date. The later feed date of Oct 31 is correctly not used.","excerpt":"[Submitted on 28 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"The stated mechanisms, the reported percentages and the 24-of-27 result are all grounded verbatim in the abstract. The interpretive claim that benchmark results do not establish deployed performance or cost is labeled as interpretation and stays within the evidence. The question is an open issue that follows from the evidence, not a leading claim.","excerpt":"It achieves the strongest performance in 24 of 27 settings."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper. It is the most direct way for a reader to access the source.","excerpt":null}],"createdAt":"2026-11-13T03:35:40.411Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}