{"id":"e205f4c44396c21b80487abbe019e6a7d731ea14147e6940de91d032debef780","title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","argument":"The authors propose HIDRA, a routing framework for robots that learn from demonstrations over time without storing or revisiting past data. They argue that sequential updates can distort a model’s representation space and make expert selection less reliable, especially when routing interference accumulates.\n\nHIDRA uses hierarchical dual-attention to separate instruction-level expert retrieval from context-dependent refinement. The authors also introduce key-level geometric regularization to encourage alignment within tasks and separation across tasks. Together, these methods are intended to stabilize expert selection as the set of experts expands.\n\nOn LIBERO benchmark suites, including tests with heterogeneous task sequences and paraphrased instructions, the authors report better AUC and forward transfer than replay-free and replay-based baselines, while maintaining low forgetting. They report stronger gains under high interference and language variation. These results concern the reported benchmark evaluations; the abstract does not establish how the method performs in physical robots or other settings.\n\nThe open question is whether HIDRA’s reported gains carry over to real-world robots learning from ongoing demonstrations, where past data cannot be replayed.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported evaluation is on LIBERO benchmark suites; the abstract does not report evaluation on physical robots or in other settings. The abstract does not give numerical result sizes or enough experimental detail to assess the reported comparisons. The source identifies this as the authors’ preprint and says the final version is available in the NeurIPS 2026 proceedings.","speaker":{"name":"Yu, Fanqi, Tiezzi, Matteo, Beyan, Cigdem, Apicella, Tommaso, Murino, Vittorio","roleAtSource":"Article author"},"source":{"title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.36374","publishedAt":"2026-10-26","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose HIDRA for replay-free, single-pass lifelong imitation learning, where robots adapt from demonstrations without storing or revisiting past data.","excerpt":"In this replay-free, single-pass lifelong imitation learning setting, sequential updates can progressively distort the representation space","url":"https://arxiv.org/abs/2610.36374","relationship":"Source-stated"},{"text":"HIDRA separates instruction-level expert retrieval from context-dependent refinement through hierarchical dual-attention, and uses key-level geometric regularization to encourage alignment within tasks and separation across tasks.","excerpt":"a hierarchical dual-attention mechanism, which decouples instruction-level expert retrieval from context-dependent refinement","url":"https://arxiv.org/abs/2610.36374","relationship":"Source-stated"},{"text":"On LIBERO benchmark suites, the authors report that HIDRA outperforms replay-free and replay-based baselines on AUC and forward transfer while maintaining low forgetting.","excerpt":"Our approach consistently outperforms both replay-free and replay-based baselines, improving AUC and forward transfer while maintaining low forgetting","url":"https://arxiv.org/abs/2610.36374","relationship":"Source-stated"},{"text":"The abstract reports benchmark results but does not establish performance on physical robots or in other settings.","excerpt":"We evaluate HIDRA on several LIBERO benchmark suites","url":"https://arxiv.org/abs/2610.36374","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.36374","checkedAt":"2026-11-14T17:22:08.485Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Do HIDRA’s reported benchmark gains carry over to real-world robots learning from ongoing demonstrations without replaying past data?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","summary":"The authors propose HIDRA, a routing framework for robots that learn from demonstrations over time without storing or revisiting past data. They argue that sequential updates can distort a model’s representation space and make expert selection less reliable, especially when routing interference accumulates.\n\nHIDRA uses hierarchical dual-attention to separate instruction-level expert retrieval from context-dependent refinement. The authors also introduce key-level geometric regularization to encourage alignment within tasks and separation across tasks. Together, these methods are intended to stabilize expert selection as the set of experts expands.\n\nOn LIBERO benchmark suites, including tests with heterogeneous task sequences and paraphrased instructions, the authors report better AUC and forward transfer than replay-free and replay-based baselines, while maintaining low forgetting. They report stronger gains under high interference and language variation. These results concern the reported benchmark evaluations; the abstract does not establish how the method performs in physical robots or other settings.\n\nThe open question is whether HIDRA’s reported gains carry over to real-world robots learning from ongoing demonstrations, where past data cannot be replayed.","question":"Do HIDRA’s reported benchmark gains carry over to real-world robots learning from ongoing demonstrations without replaying past data?","claims":[{"text":"The authors propose HIDRA for replay-free, single-pass lifelong imitation learning, where robots adapt from demonstrations without storing or revisiting past data.","excerpt":"In this replay-free, single-pass lifelong imitation learning setting, sequential updates can progressively distort the representation space","support":"source-stated"},{"text":"HIDRA separates instruction-level expert retrieval from context-dependent refinement through hierarchical dual-attention, and uses key-level geometric regularization to encourage alignment within tasks and separation across tasks.","excerpt":"a hierarchical dual-attention mechanism, which decouples instruction-level expert retrieval from context-dependent refinement","support":"source-stated"},{"text":"On LIBERO benchmark suites, the authors report that HIDRA outperforms replay-free and replay-based baselines on AUC and forward transfer while maintaining low forgetting.","excerpt":"Our approach consistently outperforms both replay-free and replay-based baselines, improving AUC and forward transfer while maintaining low forgetting","support":"source-stated"},{"text":"The abstract reports benchmark results but does not establish performance on physical robots or in other settings.","excerpt":"We evaluate HIDRA on several LIBERO benchmark suites","support":"interpretation"}],"limitations":["The reported evaluation is on LIBERO benchmark suites; the abstract does not report evaluation on physical robots or in other settings.","The abstract does not give numerical result sizes or enough experimental detail to assess the reported comparisons.","The source identifies this as the authors’ preprint and says the final version is available in the NeurIPS 2026 proceedings."],"mediaId":"4ad7c3640bcdc45cc178ce017cc6cd8e0c68596c11d52e1ce4d8571659bdccb5","correctionReason":null,"id":"e205f4c44396c21b80487abbe019e6a7d731ea14147e6940de91d032debef780","publicationId":"69539052-e2b0-4c1e-8336-50c8add44813","version":1,"sourceUrl":"https://arxiv.org/abs/2610.36374","sourceTitle":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-26","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-28T05:24:37.903Z","createdAt":"2026-11-14T00:19:21.066Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"4ad7c3640bcdc45cc178ce017cc6cd8e0c68596c11d52e1ce4d8571659bdccb5","kind":"article","url":"https://arxiv.org/abs/2610.36374","sourceUrl":"https://arxiv.org/abs/2610.36374","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-30T22:03:14.737Z","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/2610.36374","checkedAt":"2026-11-14T17:22:08.485Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"4ad7c3640bcdc45cc178ce017cc6cd8e0c68596c11d52e1ce4d8571659bdccb5","confidence":0.84,"provider":"typesafe","model":"jev-1.13.0","inputHash":"d09bdaedaf1e2f51a67302819ed244936922cb9e8f48b54c59a6354eded08edf","decidedAt":"2026-10-30T22:03:16.993Z"},"provenance":{"textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observationId":"c71e0382ece6c3088d6f1d80093027b3ae8947393d85f504595514754f137211","retrievedAt":"2026-10-30T22:03:14.248Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"cf090bc8-cf30-4f76-81ab-757ae21c5643"},"reviewer":{"model":"gpt-6-luna","runId":"2f0c3311-8a56-467b-8057-a4074232dbc1"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-14T00:19:21.066Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Yu, Fanqi","url":null,"kind":"unknown"},{"name":"Tiezzi, Matteo","url":null,"kind":"unknown"},{"name":"Beyan, Cigdem","url":null,"kind":"unknown"},{"name":"Apicella, Tommaso","url":null,"kind":"unknown"},{"name":"Murino, Vittorio","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Yu, Fanqi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Fanqi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Tiezzi, Matteo","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tiezzi, Matteo","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Beyan, Cigdem","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Beyan, Cigdem","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Apicella, Tommaso","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Apicella, Tommaso","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Murino, Vittorio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Murino, Vittorio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.36374","snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/e205f4c44396c21b80487abbe019e6a7d731ea14147e6940de91d032debef780/source"}},"canonicalUrl":"https://escapemeridian.com/readings/hidra-hierarchical-dual-routing-attention-for-replay-free-e205f4c4","formats":{"html":"/readings/hidra-hierarchical-dual-routing-attention-for-replay-free-e205f4c4","markdown":"/readings/hidra-hierarchical-dual-routing-attention-for-replay-free-e205f4c4.md","json":"/readings/hidra-hierarchical-dual-routing-attention-for-replay-free-e205f4c4.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":[],"conditions":[],"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-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"}],"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":"e205f4c44396c21b80487abbe019e6a7d731ea14147e6940de91d032debef780","revision":"69539052-e2b0-4c1e-8336-50c8add44813","revisionKind":"publication-id","title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","url":"https://escapemeridian.com/readings/hidra-hierarchical-dual-routing-attention-for-replay-free-e205f4c4"},"source":{"type":"source","id":"https://arxiv.org/abs/2610.36374","revision":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","revisionKind":"sha256-source-text","title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","url":"https://arxiv.org/abs/2610.36374"},"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":"e205f4c44396c21b80487abbe019e6a7d731ea14147e6940de91d032debef780","title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","argument":"The authors propose HIDRA, a routing framework for robots that learn from demonstrations over time without storing or revisiting past data. They argue that sequential updates can distort a model’s representation space and make expert selection less reliable, especially when routing interference accumulates.\n\nHIDRA uses hierarchical dual-attention to separate instruction-level expert retrieval from context-dependent refinement. The authors also introduce key-level geometric regularization to encourage alignment within tasks and separation across tasks. Together, these methods are intended to stabilize expert selection as the set of experts expands.\n\nOn LIBERO benchmark suites, including tests with heterogeneous task sequences and paraphrased instructions, the authors report better AUC and forward transfer than replay-free and replay-based baselines, while maintaining low forgetting. They report stronger gains under high interference and language variation. These results concern the reported benchmark evaluations; the abstract does not establish how the method performs in physical robots or other settings.\n\nThe open question is whether HIDRA’s reported gains carry over to real-world robots learning from ongoing demonstrations, where past data cannot be replayed.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported evaluation is on LIBERO benchmark suites; the abstract does not report evaluation on physical robots or in other settings. The abstract does not give numerical result sizes or enough experimental detail to assess the reported comparisons. The source identifies this as the authors’ preprint and says the final version is available in the NeurIPS 2026 proceedings.","speaker":{"name":"Yu, Fanqi, Tiezzi, Matteo, Beyan, Cigdem, Apicella, Tommaso, Murino, Vittorio","roleAtSource":"Article author"},"source":{"title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.36374","publishedAt":"2026-10-26","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose HIDRA for replay-free, single-pass lifelong imitation learning, where robots adapt from demonstrations without storing or revisiting past data.","excerpt":"In this replay-free, single-pass lifelong imitation learning setting, sequential updates can progressively distort the representation space","url":"https://arxiv.org/abs/2610.36374","relationship":"Source-stated"},{"text":"HIDRA separates instruction-level expert retrieval from context-dependent refinement through hierarchical dual-attention, and uses key-level geometric regularization to encourage alignment within tasks and separation across tasks.","excerpt":"a hierarchical dual-attention mechanism, which decouples instruction-level expert retrieval from context-dependent refinement","url":"https://arxiv.org/abs/2610.36374","relationship":"Source-stated"},{"text":"On LIBERO benchmark suites, the authors report that HIDRA outperforms replay-free and replay-based baselines on AUC and forward transfer while maintaining low forgetting.","excerpt":"Our approach consistently outperforms both replay-free and replay-based baselines, improving AUC and forward transfer while maintaining low forgetting","url":"https://arxiv.org/abs/2610.36374","relationship":"Source-stated"},{"text":"The abstract reports benchmark results but does not establish performance on physical robots or in other settings.","excerpt":"We evaluate HIDRA on several LIBERO benchmark suites","url":"https://arxiv.org/abs/2610.36374","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.36374","checkedAt":"2026-11-14T17:22:08.485Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Do HIDRA’s reported benchmark gains carry over to real-world robots learning from ongoing demonstrations without replaying past data?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","summary":"The authors propose HIDRA, a routing framework for robots that learn from demonstrations over time without storing or revisiting past data. They argue that sequential updates can distort a model’s representation space and make expert selection less reliable, especially when routing interference accumulates.\n\nHIDRA uses hierarchical dual-attention to separate instruction-level expert retrieval from context-dependent refinement. The authors also introduce key-level geometric regularization to encourage alignment within tasks and separation across tasks. Together, these methods are intended to stabilize expert selection as the set of experts expands.\n\nOn LIBERO benchmark suites, including tests with heterogeneous task sequences and paraphrased instructions, the authors report better AUC and forward transfer than replay-free and replay-based baselines, while maintaining low forgetting. They report stronger gains under high interference and language variation. These results concern the reported benchmark evaluations; the abstract does not establish how the method performs in physical robots or other settings.\n\nThe open question is whether HIDRA’s reported gains carry over to real-world robots learning from ongoing demonstrations, where past data cannot be replayed.","question":"Do HIDRA’s reported benchmark gains carry over to real-world robots learning from ongoing demonstrations without replaying past data?","claims":[{"text":"The authors propose HIDRA for replay-free, single-pass lifelong imitation learning, where robots adapt from demonstrations without storing or revisiting past data.","excerpt":"In this replay-free, single-pass lifelong imitation learning setting, sequential updates can progressively distort the representation space","support":"source-stated"},{"text":"HIDRA separates instruction-level expert retrieval from context-dependent refinement through hierarchical dual-attention, and uses key-level geometric regularization to encourage alignment within tasks and separation across tasks.","excerpt":"a hierarchical dual-attention mechanism, which decouples instruction-level expert retrieval from context-dependent refinement","support":"source-stated"},{"text":"On LIBERO benchmark suites, the authors report that HIDRA outperforms replay-free and replay-based baselines on AUC and forward transfer while maintaining low forgetting.","excerpt":"Our approach consistently outperforms both replay-free and replay-based baselines, improving AUC and forward transfer while maintaining low forgetting","support":"source-stated"},{"text":"The abstract reports benchmark results but does not establish performance on physical robots or in other settings.","excerpt":"We evaluate HIDRA on several LIBERO benchmark suites","support":"interpretation"}],"limitations":["The reported evaluation is on LIBERO benchmark suites; the abstract does not report evaluation on physical robots or in other settings.","The abstract does not give numerical result sizes or enough experimental detail to assess the reported comparisons.","The source identifies this as the authors’ preprint and says the final version is available in the NeurIPS 2026 proceedings."],"mediaId":"4ad7c3640bcdc45cc178ce017cc6cd8e0c68596c11d52e1ce4d8571659bdccb5","correctionReason":null,"id":"e205f4c44396c21b80487abbe019e6a7d731ea14147e6940de91d032debef780","publicationId":"69539052-e2b0-4c1e-8336-50c8add44813","version":1,"sourceUrl":"https://arxiv.org/abs/2610.36374","sourceTitle":"HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-26","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-28T05:24:37.903Z","createdAt":"2026-11-14T00:19:21.066Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"4ad7c3640bcdc45cc178ce017cc6cd8e0c68596c11d52e1ce4d8571659bdccb5","kind":"article","url":"https://arxiv.org/abs/2610.36374","sourceUrl":"https://arxiv.org/abs/2610.36374","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-30T22:03:14.737Z","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/2610.36374","checkedAt":"2026-11-14T17:22:08.485Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"4ad7c3640bcdc45cc178ce017cc6cd8e0c68596c11d52e1ce4d8571659bdccb5","confidence":0.84,"provider":"typesafe","model":"jev-1.13.0","inputHash":"d09bdaedaf1e2f51a67302819ed244936922cb9e8f48b54c59a6354eded08edf","decidedAt":"2026-10-30T22:03:16.993Z"},"provenance":{"textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observationId":"c71e0382ece6c3088d6f1d80093027b3ae8947393d85f504595514754f137211","retrievedAt":"2026-10-30T22:03:14.248Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"cf090bc8-cf30-4f76-81ab-757ae21c5643"},"reviewer":{"model":"gpt-6-luna","runId":"2f0c3311-8a56-467b-8057-a4074232dbc1"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-14T00:19:21.066Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Yu, Fanqi","url":null,"kind":"unknown"},{"name":"Tiezzi, Matteo","url":null,"kind":"unknown"},{"name":"Beyan, Cigdem","url":null,"kind":"unknown"},{"name":"Apicella, Tommaso","url":null,"kind":"unknown"},{"name":"Murino, Vittorio","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Yu, Fanqi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yu, Fanqi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Tiezzi, Matteo","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tiezzi, Matteo","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Beyan, Cigdem","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Beyan, Cigdem","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Apicella, Tommaso","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Apicella, Tommaso","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}},{"actor":{"name":"Murino, Vittorio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Murino, Vittorio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.36374","locator":"head/meta[name=author|citation_author]"},"assertionId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-14T17:22:08.485Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","sourceUrl":"https://arxiv.org/abs/2610.36374","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.36374","snapshotId":"72c5f3eb52446579972eb5eb7b02ef707f7383ecb006822d93626d3f2a04db42","contentSha256":"d1073e9d1e3899973c7496482fca54026e24a653132890b4441e5a5805a47fa0","textHash":"5e92137ea2ac858944d768856f8b8bf5464c50fd3a4e0396f8855fa5198f2372","observedAt":"2026-11-14T17:22:08.485Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/e205f4c44396c21b80487abbe019e6a7d731ea14147e6940de91d032debef780/source"}}}],"audits":[{"publicationId":"69539052-e2b0-4c1e-8336-50c8add44813","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading consistently attributes the proposal and results to the paper's authors (Yu, Tiezzi, Beyan, Apicella, Murino) and frames performance as what the authors report, not as established fact.","excerpt":"Authors:Fanqi Yu, Matteo Tiezzi, Cigdem Beyan, Tommaso Apicella, Vittorio Murino"},{"dimension":"summary","status":"uncertain","explanation":"The reading is faithful, but a non-specialist cannot follow its core idea. It never says plainly what the 'experts' are or what 'routing' or 'expert selection' means: the model chooses which specialised sub-module handles a task. AUC, forward transfer, forgetting and LIBERO are also unexplained, so the results are not readable directly. Smallest repair: keep the exact terms but add one plain-language sentence on experts and routing, plus brief glosses or tooltips for AUC, forward transfer and LIBERO. Minor: the abstract's 'consistently outperforms' is softened to 'better', which slightly weakens the authors' claim.","excerpt":"making expert-based approaches particularly sensitive to routing interference, causing expert selection to degrade over time"},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-10-26 matches the arXiv v1 submission date. It is not the later feed date of 2026-10-29.","excerpt":"[Submitted on 26 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"The source-stated claims match the abstract. Saying the abstract does not establish physical-robot performance is correctly labeled as interpretation; the source mentions only LIBERO evaluations. The open question about real-world transfer follows from the evidence and is not a leading claim. The limitation on the NeurIPS proceedings mirrors the source's comment.","excerpt":"We evaluate HIDRA on several LIBERO benchmark suites, including more challenging variants with heterogeneous task sequences and paraphrased language instructions"},{"dimension":"media","status":"pass","explanation":"The selected media ID is the verified direct arXiv abstract page, matching the source URL. No other source media was found.","excerpt":null}],"createdAt":"2026-11-14T00:19:21.066Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}