{"as_of":"2026-08-20T13:53:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:43b8a314ae16832e3c4c760db81104915392addebd9bac15d1cf18f352f8a7d1","coverage":[{"denominator":71,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":71,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-02T18:22:16.368099Z","state":"measured"},{"denominator":71,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":71,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-20T06:33:59.587034+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2603.12055/citation-record","integrity":"/paper/2603.12055/integrity","json":"/paper/2603.12055/citation-record.json","paper":"/paper/2603.12055"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:07.001973Z","title":"Learning transferable visual models from natural language supervision,","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:07.001973Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:a9855439f69ff915d5a127b8fad400e16fc7870571f7fb76eab388d4275c3a9e","observation_id":"f1b3937a-463f-430f-b1a4-7c5276b7ed79","resolution":{"observed_at":"2026-08-02T18:22:07.001973Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:07.139239Z","title":"Scaling up visual and vision-language representation learning with noisy text supervision,","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:07.139239Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:4c596924dbc96d64940a1c36b58c6bfa9de76aceaddc96899960d1af72a96bee","observation_id":"6534e0bf-4070-47d1-b686-64aa1a041836","resolution":{"observed_at":"2026-08-02T18:22:07.139239Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:07.313739Z","title":"Csta: Spatial-temporal causal adaptive learning for exemplar-free video class-incremental learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:07.313739Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:9e6296c750892f39174b653ec6849c81ce59cd32fffe4f130d2d4aaa81486b50","observation_id":"1a1da1f0-c84b-4a0f-af2f-177b6d935cee","resolution":{"observed_at":"2026-08-02T18:22:07.313739Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:07.483183Z","title":"Achieving plasticity- stability trade-off in continual learning through adaptive orthogonal projection,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:07.483183Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:e38bb4e959a5622c809b93dd369825f78846f21fcdba722c0ed4831f265a1edf","observation_id":"afac19ca-a651-4ba2-8aa3-0fe5fa30bc90","resolution":{"observed_at":"2026-08-02T18:22:07.483183Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:07.627702Z","title":"Joint memory optimiza- tion for continual learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:07.627702Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:6e2fffeb3948919362168b810f453d1d257cf7ed0d08bb07b6aa35a0db02477f","observation_id":"616ef07a-5a65-46e2-ba07-47f627991fbf","resolution":{"observed_at":"2026-08-02T18:22:07.627702Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:07.775832Z","title":"Learning without forgetting for vision-language models,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:07.775832Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:04f33bb142696a8879a8fd9b3eb6210e7489219a2f7067270f3c84a3f7addcc5","observation_id":"e33691af-2261-47b9-b33a-75a34f34e6ee","resolution":{"observed_at":"2026-08-02T18:22:07.775832Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:07.894366Z","title":"Class-aware prompting for federated few-shot class- incremental learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:07.894366Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:b5a2cf2aeb319ba91615628d3a146dc162b77c64ab5fedd55985ec6a0021caf3","observation_id":"8d5a4028-87c8-4f9e-9a12-a5eb137ed3ba","resolution":{"observed_at":"2026-08-02T18:22:07.894366Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.001746Z","title":"Mingle: Mixture of null-space gated low-rank experts for test-time continual model merging,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.001746Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:a86f3755c14323fea63170d3a0600c14d37f476a613b3137b26443fa3bba2fa7","observation_id":"0c91be2e-a8a7-433c-ab36-9cdca0a87602","resolution":{"observed_at":"2026-08-02T18:22:08.001746Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.115130Z","title":"Null-space filtering for data-free continual model merging: Preserving stability, promoting plasticity,","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.115130Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:8e369db12e7c9358b23053df6d071d61d29999efd75c542d51db846a7e765abf","observation_id":"3534288a-0205-49cb-81ce-dd884a5f4b16","resolution":{"observed_at":"2026-08-02T18:22:08.115130Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.261649Z","title":"New insights on relieving task-recency bias for online class incremental learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.261649Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:c47292699e72dc570205620cf3e2667c98839664c8c9ee80b3996e2779fb2413","observation_id":"bf6ba89e-42c0-4cc8-9186-8671e19f574c","resolution":{"observed_at":"2026-08-02T18:22:08.261649Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.449032Z","title":"Continual learning of image classes with language guidance from a vision-language model,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.449032Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:2fa4ea9a0135fdadbd5503ea5ba63159fdfaf60fb0e120e304359680b8d1cb69","observation_id":"7e9ce8f6-4046-4e79-8152-b01d69c63e0b","resolution":{"observed_at":"2026-08-02T18:22:08.449032Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.574544Z","title":"Ex- ternal knowledge injection for clip-based class-incremental learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.574544Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:6db788f3dee3e2e2548ddc30d1ab048eac2cf2f876413c7f4bb3d1873510ddce","observation_id":"3ee19e1a-d9e1-4d96-9ae3-f20d5fbef04d","resolution":{"observed_at":"2026-08-02T18:22:08.574544Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.671509Z","title":"Class-incremental learning with clip: Adaptive representation adjustment and parameter fusion,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.671509Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:15055859e4730a2981b6e894c5f1e907babd202ecb3663be10a66a5fc7296352","observation_id":"62c19383-1ace-47c7-8185-38b3fee824e6","resolution":{"observed_at":"2026-08-02T18:22:08.671509Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.747853Z","title":"Overcoming generic knowledge loss with selective parameter update,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.747853Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:df27aa8f5e416456b2799758c9a66aeffb2fcb3a9f1b0b76af58b22e3dce237b","observation_id":"d71752d4-e468-48dc-9c92-b0788d539ea6","resolution":{"observed_at":"2026-08-02T18:22:08.747853Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:08.844089Z","title":"Slca: Slow learner with classifier alignment for continual learning on a pre-trained model,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:08.844089Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:17e8e15b683317ec2dc5ddbeb080c369557de629318b612bd2b782098093bd2e","observation_id":"f218c161-b98c-40d7-8823-4c500fc9472c","resolution":{"observed_at":"2026-08-02T18:22:08.844089Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:09.049292Z","title":"Boosting continual learning of vision-language models via mixture-of-experts adapters,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:09.049292Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:c85e78bfb7e6058c7ccaa31a92372529f9ab28cfb3165ea4dee10df3bb0efe51","observation_id":"5ac4d007-3507-4536-88fb-03e6d61466ec","resolution":{"observed_at":"2026-08-02T18:22:09.049292Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:09.164280Z","title":"Desclip: Robust continual learning via general attribute descriptions for vlm-based visual recognition,","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:09.164280Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:65d960f424e7bb15c0b65d88bbe3a8537fed00f5d17a1fe5e5761369a80c4309","observation_id":"f2dae506-7c13-48c7-8714-d22b8d55066c","resolution":{"observed_at":"2026-08-02T18:22:09.164280Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:09.368529Z","title":"Language guided concept bottleneck models for interpretable continual learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:09.368529Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:ffd2d2c1d4c27e37300cab3f1ecacbae253648184e904e1efcf4556fdc4de0d4","observation_id":"6cdcd6c0-3e49-4415-b2ec-f2938d73e9e6","resolution":{"observed_at":"2026-08-02T18:22:09.368529Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:09.539444Z","title":"Hierarchical semantic tree anchoring for clip-based class-incremental learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:09.539444Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:fb5b102f8cff8ee1b6c540f41f6e56efba1719064cf4192da75775cdf4798f52","observation_id":"37a2feea-e63b-4eb2-abe2-b25a2743e1d3","resolution":{"observed_at":"2026-08-02T18:22:09.539444Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:09.763537Z","title":"Difference vector equalization for robust fine-tuning of vision-language models,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:09.763537Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:d8b3663cfe585a81ac7852c34bbe0fd8e1678b2de73d7c0e44e10c593867426f","observation_id":"9de14231-de6a-43c9-bd22-1f0536b13e19","resolution":{"observed_at":"2026-08-02T18:22:09.763537Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:09.935779Z","title":"Mind the gap: Preserving and compensating for the modality gap in clip- based continual learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:09.935779Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:4c4ff43da6d4ce35f4cf5c30b56da6cf6a8a7c20aae2dcbc34b8a01062f0bed1","observation_id":"990e050e-028b-4ef2-a283-3caf56947c96","resolution":{"observed_at":"2026-08-02T18:22:09.935779Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:10.124406Z","title":"Preventing zero-shot transfer degradation in continual learning of vision-language models,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:10.124406Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:726bebab29a072199a2f3877ad31e1f32af2abb6822237c5f7650b97b4ba7f60","observation_id":"ebdafaa7-5317-4c43-80bf-110322ee35d6","resolution":{"observed_at":"2026-08-02T18:22:10.124406Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:10.293428Z","title":"Conceptual 12m: Pushing web-scale image-text pre-training to recognize long-tail visual concepts,","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:10.293428Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:c1d11492fbfebe5fbb7ef7617ad1a9c05c2b5e61295022c5cabd9388c1c5d7d9","observation_id":"2adae78e-9b17-4465-bbc6-2cb037306e28","resolution":{"observed_at":"2026-08-02T18:22:10.293428Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:10.465865Z","title":"Imagenet: A large-scale hierarchical image database,","venue":null,"work_id":null,"year":2009},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:10.465865Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:13f3dfc0ca643f153d81d050588c3e4c24ef3c7ecc5c3bdedf2ef3393aaf86d9","observation_id":"6bdc607d-0946-4ace-8881-eec92deda546","resolution":{"observed_at":"2026-08-02T18:22:10.465865Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:10.611072Z","title":"Divergence measures based on the shannon entropy,","venue":null,"work_id":null,"year":1991},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:10.611072Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:7bff3da1c646c9891c4daed8193ffef5131a42537bf7fe9cd669a2d597854822","observation_id":"27beecf2-afea-4991-b3f7-0acd10bb225b","resolution":{"observed_at":"2026-08-02T18:22:10.611072Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:10.700605Z","title":"Text-guided attention is all you need for zero-shot robustness in vision-language models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:10.700605Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:16b31c5bc0cefca9cfe220d18a5926270524f0264ff199d6ce18cd80a7d3afc1","observation_id":"502390c1-8c0c-4a96-a6ee-7ec18c23d776","resolution":{"observed_at":"2026-08-02T18:22:10.700605Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:10.854643Z","title":"Evaluating the adversarial robustness of vision-language models via internal feature perturbations,","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:10.854643Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:3657c7c01d5b1b04dce7579d969362bd862fb61423ec2412f227d9cf2c53a04d","observation_id":"976f6b7b-51c7-465d-8d78-89a169e8fc7f","resolution":{"observed_at":"2026-08-02T18:22:10.854643Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.001382Z","title":"Attention-guided hierarchical defense for multimodal attacks in vision-language models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.001382Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:4f2bbc6c837dbec4f039f6cb1c8db6be5ef765c4dbc451b3362a1913897cc503","observation_id":"14639531-917b-4245-94d8-6caa7520ab97","resolution":{"observed_at":"2026-08-02T18:22:11.001382Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.089524Z","title":"Synthetic data is an elegant gift for continual vision-language models,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.089524Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:db4f4daf50a89b35696a4bfe0a206c38fa23297bc6b8f6cf4d34d21f5ef8be1d","observation_id":"513f44fd-e255-4ff1-852d-ea9c78a3e87a","resolution":{"observed_at":"2026-08-02T18:22:11.089524Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.162857Z","title":"Lora-loop: Closing the synthetic re- play cycle for continual vlm learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.162857Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:38f7e5dc00ebe54ef00fecae293e97812fae28de4467a2041c4f5c49c325d8f7","observation_id":"52b9b079-f92b-4922-9b4f-02a98f7aba5d","resolution":{"observed_at":"2026-08-02T18:22:11.162857Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.248094Z","title":"Dual-consistency model inversion for non-exemplar class incremental learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.248094Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:a2ea4b91531cd87aad74d9b1474249b3f39f7ba6685255e9711314dbb06dd8e3","observation_id":"e4bbfd83-03c5-41dc-bc97-87f42c80cfcb","resolution":{"observed_at":"2026-08-02T18:22:11.248094Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.378262Z","title":"Mind the gap: Understanding the modality gap in multi-modal contrastive representation learning,","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.378262Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:23fbbaa2d43bc134682012f4bf13e2ef29eb95f83db30489fd706a02cad37c98","observation_id":"3ec0ec9f-4a4a-4788-bc90-05336bc118fd","resolution":{"observed_at":"2026-08-02T18:22:11.378262Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.526543Z","title":"Clip-lora: Low-rank adaptation for clip,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.526543Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:a64171831f1f8c3c23d4641478034fb4ed717c7c41255294667674d09b553fc1","observation_id":"2e014c60-f125-4383-ba4f-f27cb5dd766d","resolution":{"observed_at":"2026-08-02T18:22:11.526543Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.658717Z","title":"Learning to prompt for vision- language models,","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.658717Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:a8a7c55e269db99003fdc57cd2bddc9a8f4f459b568ba0efa052c7f9f4568806","observation_id":"4c82890c-a9e4-4104-a7f0-5b8a9dd2f291","resolution":{"observed_at":"2026-08-02T18:22:11.658717Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:11.831562Z","title":"Conditional prompt learning for vision-language models,","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:11.831562Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:c161159ac042b7170503c0a38f085dff6fc05a92d309bed4e01c730107caaea3","observation_id":"78d2f06c-e853-4a51-adf6-18c20748acca","resolution":{"observed_at":"2026-08-02T18:22:11.831562Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:12.050749Z","title":"Graphadapter: Tuning vision-language models with dual knowledge graph,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:12.050749Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:26013aa12fde73bf9da883299eae74c72bba70966512a9a1503d81afca1e4778","observation_id":"31c0bbea-713a-4379-b258-20dd9de29ec1","resolution":{"observed_at":"2026-08-02T18:22:12.050749Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:12.165326Z","title":"Vmt-adapter: Parameter- efficient transfer learning for multi-task dense scene understanding,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:12.165326Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:b0f801169a3d7bffda2312c1f0e55d8b5e464033eae6717094d640b55f9877e8","observation_id":"c46b2745-aea6-4db1-a413-f67258e604a3","resolution":{"observed_at":"2026-08-02T18:22:12.165326Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:12.339658Z","title":"Lora: Low-rank adaptation of large language models","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:12.339658Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:40064aa2d44f719b327e69c10c30b40fa02daa186915542fe93dd095ac8858df","observation_id":"14e99aee-aed7-47a5-9ab1-34f4c20bd508","resolution":{"observed_at":"2026-08-02T18:22:12.339658Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:12.610506Z","title":"A model or 603 exemplars: Towards memory-efficient class-incremental learning,","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":39,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:12.610506Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:792acf9967c6b2fa63bb1af7e33c7af3f1ca09176cdbdd26f508dc3ef6b18694","observation_id":"e5c99e5f-af01-4527-839b-233ed0e7eeea","resolution":{"observed_at":"2026-08-02T18:22:12.610506Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:12.844102Z","title":"Ceat: Continual expansion and absorption transformer for non-exemplar class- incremental learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":40,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:12.844102Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:341e48d4a4c750161f5abe01142d3c5dcea858a6d581afbbc5c79a5ac77d5afc","observation_id":"873a2af0-9a3a-45e1-8b01-a315ee37d4cb","resolution":{"observed_at":"2026-08-02T18:22:12.844102Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:12.969860Z","title":"Learning without forgetting,","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":41,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:12.969860Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:bf63f501eebe1416acee7c3b205cde0f44b483f26fb03168d46622c11f7b29d5","observation_id":"f317d5e4-ed3d-41c3-b6b9-ab4b0eacc158","resolution":{"observed_at":"2026-08-02T18:22:12.969860Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:13.117849Z","title":"Overcoming catastrophic forgetting in neural networks,","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":42,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.117849Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:1393358c2b575b74991d932ff23bd297249962a60c9c8a7ee68da326c4ca0624","observation_id":"9b3c5ae1-a447-4599-87ae-06516d9aa035","resolution":{"observed_at":"2026-08-02T18:22:13.117849Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:13.322385Z","title":"Learning to prompt for continual learning,","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":43,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.322385Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:cc0c43c5923dc4e4032cee4c5cc033b703499d083945ac4332053ad10c1ed3dd","observation_id":"cc7c76d2-9607-4a53-aca1-da00a0907c80","resolution":{"observed_at":"2026-08-02T18:22:13.322385Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:13.495166Z","title":"Coda-prompt: Contin- ual decomposed attention-based prompting for rehearsal-free continual learning,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":44,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.495166Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:49701f6f5ee927ae2dc5de90232fe8f53251ec2fe9c3b042065775babb975df2","observation_id":"e6f704f9-3227-47ca-98ea-504ed4c615fb","resolution":{"observed_at":"2026-08-02T18:22:13.495166Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:13.677939Z","title":"Expandable subspace ensemble for pre-trained model-based class-incremental learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":45,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.677939Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:4f4cb40fd30074eb49f71e3e8ba6e41615dac4fcd4af186878ed63b135c80f79","observation_id":"acb7d825-768f-458a-91f6-9992a685ecc5","resolution":{"observed_at":"2026-08-02T18:22:13.677939Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:13.765342Z","title":"Inflora: Interference-free low-rank adaptation for continual learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":46,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.765342Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:875e3d67fd1e63935a7654868b149762332c245c6ae7768aa7776f14be05da4a","observation_id":"dca6ef5e-d04e-4a7d-8874-f6c1437cf385","resolution":{"observed_at":"2026-08-02T18:22:13.765342Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:13.806123Z","title":"Consistent prompting for rehearsal- free continual learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":47,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.806123Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:2705c3cb257b886e0eb74c10d29d2021898c11ede9a10f7e445a774b576e7630","observation_id":"2a1f5bf0-efcd-40b2-b1f5-9b09600b6e34","resolution":{"observed_at":"2026-08-02T18:22:13.806123Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:13.878392Z","title":"Revisiting class-incremental learning with pre-trained models: Generalizability and adaptivity are all you need,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":48,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.878392Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:e1c5f0a4afa84ef6d7282ceb328298002062ac6626ab86c2aaa7b52317335bb0","observation_id":"62faca89-1cad-4bda-beab-bdfb129c131b","resolution":{"observed_at":"2026-08-02T18:22:13.878392Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2408.01076","last_updated":"2025-06-09T08:48:46Z","snapshot_observed_at":"2026-08-19T16:25:55.651828Z","submitted_at":"2024-08-02T07:51:44Z","title":"Exploiting the Semantic Knowledge of Pre-trained Text-Encoders for Continual Learning","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2408.01076","snapshot_observed_at":"2026-08-02T18:22:13.979188Z","title":"Exploiting the semantic knowledge of pre-trained text-encoders for continual learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":49,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:13.979188Z"},"links":{"cited_paper":"/paper/2408.01076","citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:93f7ed6be4dd4e2ab9e9a1b2c5dcc4dd34866339ccf74c0d4a64c3d28504ffe7","observation_id":"f6747f81-a55e-4bc8-988a-9a306957da10","resolution":{"observed_at":"2026-08-02T18:22:13.979188Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.090989Z","title":"Select and distill: Selective dual-teacher knowledge transfer for continual learning on vision-language models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":50,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.090989Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:1b59109936b940554ee8552e07db2eeed6270b61cc2e4cb9f163f53a1508529c","observation_id":"becaef37-7a0e-4215-8237-4624bd9e9830","resolution":{"observed_at":"2026-08-02T18:22:14.090989Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.197125Z","title":"Adapt without forgetting: Distill proximity from dual teachers in vision-language models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":51,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.197125Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:4cca585967793c399a6dc79064b66d2b56c9ffce71ae710e830dde61b48a015e","observation_id":"b2c02c29-7752-43bf-8a6f-9ed1f50c450c","resolution":{"observed_at":"2026-08-02T18:22:14.197125Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.345362Z","title":"Explaining and harnessing adversarial examples,","venue":null,"work_id":null,"year":2015},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":52,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.345362Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:71d63a030a208fb4772aa603c1ddaa504e8c6c08c66e2207fdcdb99f986d8cac","observation_id":"bfc94037-b27a-4232-baac-fe514c18f354","resolution":{"observed_at":"2026-08-02T18:22:14.345362Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.466601Z","title":"Adversarial examples in the physical world,","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":53,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.466601Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:84414139c9c8bbfa2946e4c7666552ca382ce886a63e0429517c3e9e94f71dbf","observation_id":"3c7cce05-10a3-4999-bd7b-746e12788fd5","resolution":{"observed_at":"2026-08-02T18:22:14.466601Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.551436Z","title":"Towards deep learning models resistant to adversarial attacks,","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":54,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.551436Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:1d0f946376d0f8ae0bacc2490686b8303c9dab21eddc49588f102732e843344e","observation_id":"589cf603-28bb-4cb7-8fc7-62119a4f3ea9","resolution":{"observed_at":"2026-08-02T18:22:14.551436Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2511.11421","last_updated":"2026-06-23T16:56:06Z","snapshot_observed_at":"2026-08-07T22:33:43.649307Z","submitted_at":"2025-11-14T15:51:40Z","title":"BOFA: Bridge-Layer Orthogonal Low-Rank Fusion for CLIP-Based Class-Incremental Learning","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2511.11421","snapshot_observed_at":"2026-08-02T18:22:14.645821Z","title":"Bofa: Bridge-layer orthogonal low-rank fusion for clip-based class-incremental learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":55,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.645821Z"},"links":{"cited_paper":"/paper/2511.11421","citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:9b92e276b25aaa7d3ea99924cc68e094cfb0262711aaf9c0a23e4df3e41c8673","observation_id":"f9000dfe-5bb6-4dfb-9d38-6490af0f83cd","resolution":{"observed_at":"2026-08-02T18:22:14.645821Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.749797Z","title":"On information and sufficiency,","venue":null,"work_id":null,"year":1951},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":56,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.749797Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:82a98d30061c50ad1299527f26e770f889fd099480846028cba57ae03f30386f","observation_id":"945dd24f-4079-4d6e-a5df-2d2906a899b9","resolution":{"observed_at":"2026-08-02T18:22:14.749797Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.860770Z","title":"Dynamic multi-layer null space projection for vision-language continual learning,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":57,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.860770Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:61340e8fdf4b783e1edd9d539497e63c344ae7dab14f1383243e60797694963b","observation_id":"662c5c04-b557-4cc6-8858-fc2f9a0d3edd","resolution":{"observed_at":"2026-08-02T18:22:14.860770Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:14.989677Z","title":"Exemplar-free continual representation learning via learnable drift compensation,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":58,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:14.989677Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:b42cd6406623e1f7705cd8545c2e41c40e3fca77f93140e3957d6ced17e58a24","observation_id":"d564afcf-cbbd-4eee-b7ac-8139a6f8d361","resolution":{"observed_at":"2026-08-02T18:22:14.989677Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.113748Z","title":"Resurrecting old classes with new data for exemplar-free continual learning,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":59,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.113748Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:0f91b534978941ae3546ec512d083b507d1d631de570d5c6d9a3d24d83877ded","observation_id":"834f4bc0-5399-4d3e-9b9d-eb5b1c668af3","resolution":{"observed_at":"2026-08-02T18:22:15.113748Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2210.03114","last_updated":"2022-10-06T17:59:15Z","snapshot_observed_at":"2026-08-16T16:26:19.842459Z","submitted_at":"2022-10-06T17:59:15Z","title":"CLIP model is an Efficient Continual Learner","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2210.03114","snapshot_observed_at":"2026-08-02T18:22:15.207938Z","title":"Clip model is an efficient continual learner,","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":60,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.207938Z"},"links":{"cited_paper":"/paper/2210.03114","citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:999470139baa6143571c37f9e284c32fb8dc4cb5635f5bac5db07221f2b3b518","observation_id":"a847d8b3-e4b8-4e78-9239-eb1b7e68e8fc","resolution":{"observed_at":"2026-08-02T18:22:15.207938Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.314081Z","title":"Clap4clip: Continual learning with probabilistic finetuning for vision-language models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":61,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.314081Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:6ee3bac1999219a32e564195256cdf09ee36501bbcae4ca2a2d490957a278f4e","observation_id":"6792ce4f-0fb0-4d10-a0fc-f983636a9459","resolution":{"observed_at":"2026-08-02T18:22:15.314081Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.417074Z","title":"Learning multiple layers of features from tiny images,","venue":null,"work_id":null,"year":2009},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":62,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.417074Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:f6f9fe950d44ec80e2e9a5299b3f79ea3de8ae37de27b4e4042105753dee8f2d","observation_id":"36cecfb9-ab38-4869-87e0-f2da458afaa2","resolution":{"observed_at":"2026-08-02T18:22:15.417074Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.570091Z","title":"The caltech-ucsd birds-200-2011 dataset,","venue":null,"work_id":null,"year":2011},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":63,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.570091Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:51691c95b1855ff9e8bef462df63d128eedb7d832ac4e6722fe288d9b38a083d","observation_id":"1605421e-b5af-4181-8789-80aa111a6bec","resolution":{"observed_at":"2026-08-02T18:22:15.570091Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.689815Z","title":"The many faces of robustness: A critical analysis of out-of-distribution generalization,","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":64,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.689815Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:97a822c4c82e4a9aebf32f41a8ecce26ea441883ed0ae191a16a5205f113ae2a","observation_id":"bf51e84c-b046-43c1-a637-65f93b8068de","resolution":{"observed_at":"2026-08-02T18:22:15.689815Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.768275Z","title":"Ucf101: A dataset of 101 human actions classes from videos in the wild,","venue":null,"work_id":null,"year":2012},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":65,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.768275Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:9735b2e31ab155be3735b892498927837715d7bc43a8e1c469588c4587c025a4","observation_id":"604313b5-3ad5-4a2a-90c5-ef0ac6f18b9a","resolution":{"observed_at":"2026-08-02T18:22:15.768275Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.852180Z","title":"Large-scale machine learning with stochastic gradient de- scent,","venue":null,"work_id":null,"year":2010},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":66,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.852180Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:665899bc14b6a663f90eca7049f3238c077accf1ed3c307b88ccef8abea9c8bf","observation_id":"730421f2-dc63-4ce3-95b6-904ee531530e","resolution":{"observed_at":"2026-08-02T18:22:15.852180Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:15.939248Z","title":"High- resolution image synthesis with latent diffusion models,","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":67,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:15.939248Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:da5e6666ab39f438299e3c2c5481db91ec59f365be0534be125d69cdc977ff32","observation_id":"77cbf9c4-44dc-451d-9825-294ddbe18c29","resolution":{"observed_at":"2026-08-02T18:22:15.939248Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:16.048292Z","title":"Food-101–mining dis- criminative components with random forests,","venue":null,"work_id":null,"year":2014},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":68,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:16.048292Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:736e5e53bf6d934dfdc3a802307e5f2b298c241e6e3e16d9ce773593895049bb","observation_id":"fbd4a2f8-5423-4a42-8ba1-bb3eea612daf","resolution":{"observed_at":"2026-08-02T18:22:16.048292Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:16.180668Z","title":"Cats and dogs,","venue":null,"work_id":null,"year":2012},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":69,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:16.180668Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:3c32cdd86f2ba5f92d6c9083c2e6def49624962c9411c051f1e1c0f42e55dbef","observation_id":"442d4ec9-ba96-4551-af38-a04c371a21d5","resolution":{"observed_at":"2026-08-02T18:22:16.180668Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:16.258665Z","title":"Ad- vancing cross-domain discriminability in continual learning of vision- language models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":70,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:16.258665Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:e60c1ee2b2107885c3041e278e2f63623b6446a5e69ccd8defc793fca32b2dc3","observation_id":"07690ba8-22aa-4fbc-8ffc-155965735c3b","resolution":{"observed_at":"2026-08-02T18:22:16.258665Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:22:16.368099Z","title":"Video segmentation and its applications","venue":null,"work_id":null,"year":2005},"citing_paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation","version":3},"reference_index":2005,"source":"pdf_text","source_observed_at":"2026-08-02T18:22:16.368099Z"},"links":{"citing_paper":"/paper/2603.12055"},"observation_digest":"sha256:5cc73c83673c6d81e6ce5674ac1d741f23a56feb2a1564375efab5ae13ae7f2b","observation_id":"9521f373-fa3e-4ddf-a0f0-6b3065169804","resolution":{"observed_at":"2026-08-02T18:22:16.368099Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2603.12055","last_updated":"2026-07-30T02:56:46Z","latest_version":3,"primary_category":"cs.CV","snapshot_observed_at":"2026-08-20T13:21:03.546106Z","submitted_at":"2026-03-12T15:25:53Z","title":"Continual Learning with Vision-Language Models via Semantic-Geometry Preservation"},"reference_resolution":{"displayed":71,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":71,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":71},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-20T06:33:59.587034+00:00","source":"crossref"},{"observed_at":"2026-08-20T06:33:54.927442+00:00","source":"retraction_watch"}],"thesis":"As of 20 August 2026, this Paper Citation Record lists 71 of 71 outbound references and 0 inbound Pith citation observations for arXiv:2603.12055."}