{"as_of":"2026-08-08T16:56:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:d0f366198f4d3e605dd7a92a716953f2261c1f8cb2224a97934d1a88f0324e11","coverage":[{"denominator":72,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":72,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-06T16:26:35.498234Z","state":"measured"},{"denominator":72,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":72,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-08T06:32:00.761636+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/2507.13568/citation-record","integrity":"/paper/2507.13568/integrity","json":"/paper/2507.13568/citation-record.json","paper":"/paper/2507.13568"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:37.089945Z","title":"Expert gate: Lifelong learning with a network of experts","venue":null,"work_id":"a5e6fd31-87b3-46d4-9e45-fca6ec8d0dc6","year":2017},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:34.861720Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:bfa25e2bd530c9a6f6b3978c06f887998f8aa7b50517934c3664fb2e8219b05d","observation_id":"1d98ae2c-db50-4ec2-8ec8-44a749aad14c","resolution":{"observed_at":"2026-08-06T16:26:37.098243Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:37.063579Z","title":"Memory aware synapses: Learning what (not) to forget","venue":null,"work_id":"853144be-06bd-4bd3-a552-ba689d86f662","year":2018},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:34.982473Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:ce282fc60fe2bbb2ab760a808f22f8cb008d1e80ad92a1324577f8fdfc127ef2","observation_id":"6fd3ef03-bba8-4bdb-bddf-868670e11e75","resolution":{"observed_at":"2026-08-06T16:26:37.072139Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:37.034585Z","title":"Food-101 - mining discriminative components with random forests","venue":null,"work_id":"6ad5a8dd-8469-4ef8-86fd-2fe896798792","year":2014},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.028158Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:983fb394613878505fbfc712a86ca98eb48067e4041329fe02eea83eee6331fb","observation_id":"ec3c6642-54e1-4dd6-857a-5fa594269b65","resolution":{"observed_at":"2026-08-06T16:26:37.044648Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:37.013518Z","title":"Sharegpt4v: Improving large multi-modal models with better captions","venue":null,"work_id":"d23073c3-e873-416f-b6f6-ab0b8cf09b35","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.035376Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:2691ae2a91733c37e31f3c1a84e4e40eb72d1dd03108b05140c6d9729d6819ce","observation_id":"6d23d2f0-cff1-44bf-909e-7a52a1b9d338","resolution":{"observed_at":"2026-08-06T16:26:37.020157Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.992219Z","title":"Describing textures in the wild","venue":null,"work_id":"6bc87123-1bca-4aeb-8e98-18dcbc34067f","year":2014},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.042437Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:cca06ee69a85b17dcd5b82750b8a0261eaca4df346ed7fe0043576475e46eac1","observation_id":"165e3d35-f5b8-433e-9365-da4f6ae8867f","resolution":{"observed_at":"2026-08-06T16:26:36.999495Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.970964Z","title":"The MNIST database of handwritten digit images for machine learning research [best of the web].IEEE Signal Processing Magazine, 29, 2012","venue":null,"work_id":"28ec4579-aa79-4710-b9a6-eee31b0503a7","year":2012},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.049318Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:51def74655ecacf3d7b80109e202512fa292b409cf27c6834345eafe61719d37","observation_id":"324a7e5b-cc6f-4de9-9ec9-85ecb03ebff9","resolution":{"observed_at":"2026-08-06T16:26:36.977020Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.055892Z","title":"Learning without mem- orizing","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.055892Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:79e85b51aba170c944e4a13b95b8ed7a7b6cfba805dcf19b986afbe0417ef4e6","observation_id":"5ef582b7-e63b-4786-8c8b-09e1cb679efa","resolution":{"observed_at":"2026-08-06T16:26:35.055892Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.932306Z","title":"An image is worth 16x16 words: Transformers for image recognition at scale","venue":null,"work_id":"a6bb1b33-dd57-4e0e-b5ee-87d6cd0e4c4c","year":2021},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.062362Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:39c7a5762f84377cb5ee1338d92701cde2ab26d3b141f0d9403241b6780820d8","observation_id":"69c684e3-6f56-4fa1-b018-bc013ee96ff2","resolution":{"observed_at":"2026-08-06T16:26:36.938980Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.909569Z","title":"Podnet: Pooled outputs distil- lation for small-tasks incremental learning","venue":null,"work_id":"cf1c73c3-47cc-4b35-9e51-408e79892399","year":2020},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.069727Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:39c2dfedb26d8e61708582df28fc7e143eee600b3eab844daa613a82d7b6c137","observation_id":"c8d57873-943a-4894-a9a8-65e062e68a02","resolution":{"observed_at":"2026-08-06T16:26:36.917921Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.884570Z","title":"Dytox: Transformers for continual learn- ing with dynamic token expansion","venue":null,"work_id":"32cbabc9-09ec-4521-a29d-b2f7aad82d51","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.077808Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:4df4d6cb8a51aaf38367e142774a66c1909f40607e90f1d7dea52ec8361f4da8","observation_id":"0b217111-7bfb-4753-9ea4-8171fa230b19","resolution":{"observed_at":"2026-08-06T16:26:36.890189Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.865127Z","title":"Scaling laws of synthetic images for model training","venue":null,"work_id":"d2e529f1-d3dd-4604-937d-e4bf00bf5f81","year":null},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.084155Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:014634765afa8ef986aab6870f033a3ae892e6b20f6d70bafbcfdc9e6b29def3","observation_id":"1577edf1-a715-4c28-a786-90a2cda48e1b","resolution":{"observed_at":"2026-08-06T16:26:36.870814Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.839676Z","title":"Learning gen- erative visual models from few training examples: An incre- mental bayesian approach tested on 101 object categories","venue":null,"work_id":"7ba3ecd8-6e39-43e6-a725-2f55894759f8","year":2007},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.090009Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:bfb3b09078f4033164089e80955b638f3a283dfac02b5574cfd5362ed6c9d7cb","observation_id":"c18e0157-591a-48df-9663-a8abe66b929f","resolution":{"observed_at":"2026-08-06T16:26:36.848306Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.809105Z","title":"Clip-adapter: Better vision-language models with feature adapters","venue":null,"work_id":"61046cd8-b864-47bf-940e-51c8ef3ee80f","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.095902Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:ca4a01c1afdbcae15ebd459d4f530b721011cc5e77dffc9087504a192ce3d87c","observation_id":"9ef151ed-61fc-460a-96ad-32bd1edf5166","resolution":{"observed_at":"2026-08-06T16:26:36.820195Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.791836Z","title":"DDGR: continual learning with deep diffusion-based generative replay","venue":null,"work_id":"e6c1a155-9a5e-4a9b-9b33-e191798a63fd","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.104482Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:3337671d164d334f87cd842a2df81d3cd89b4daf91521e72e28787b3a0aa8ce3","observation_id":"c8506f8b-8e51-4eac-9b13-9dc3aaf3c4fc","resolution":{"observed_at":"2026-08-06T16:26:36.797173Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.774195Z","title":"Finetune like you pretrain: Im- proved finetuning of zero-shot vision models","venue":null,"work_id":"af01523f-f853-4f64-8379-7e8d6b8f2eab","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.112527Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:ddc2db09e98838afd751a00d5d6e2d5da07aaf53676a23ef7c71b20792276fc1","observation_id":"32f364f8-0256-428d-9ac6-fe483baa48f8","resolution":{"observed_at":"2026-08-06T16:26:36.779911Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2402.01832","last_updated":"2024-07-18T10:21:29Z","snapshot_observed_at":"2026-08-04T15:45:57.991866Z","submitted_at":"2024-02-02T18:59:58Z","title":"SynthCLIP: Are We Ready for a Fully Synthetic CLIP Training?","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2402.01832","snapshot_observed_at":"2026-08-06T16:26:35.118609Z","title":"Synthclip: Are we ready for a fully synthetic CLIP training? CoRR, abs/2402.01832, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.118609Z"},"links":{"cited_paper":"/paper/2402.01832","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:5e541ba56b6141aaa8f6d04161226221c23795b49e6af8ff255b4a818c23988f","observation_id":"a64db165-4b06-40ab-9da7-53b8db24df41","resolution":{"observed_at":"2026-08-06T16:26:35.118609Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.754677Z","title":"EuroSAT: A novel dataset and deep learn- ing benchmark for land use and land cover classification","venue":null,"work_id":"533ec210-37dd-435d-993d-cc1541274a19","year":2019},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.124750Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:d3b77c84b1c2c70dbdaa1875c82930b0db24b9c5efade17cc27e2732f6315974","observation_id":"aec17fe2-ebbe-40d3-acc3-acdd891f504b","resolution":{"observed_at":"2026-08-06T16:26:36.759932Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.735180Z","title":"Feedback-guided data synthesis for imbalanced classifica- tion","venue":null,"work_id":"bb851268-72d8-4f18-a509-1ac23cffc29e","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.131957Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:35f3bca4a306c25ed9ddc8f69fad441610da5d159f81f4b407c4ae5d6f201571","observation_id":"72776ca2-0d88-46b3-9bd7-2f07a0b42aae","resolution":{"observed_at":"2026-08-06T16:26:36.741090Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.137907Z","title":"Learning a unified classifier incrementally via rebalancing","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.137907Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:b94b139d701e4ed18074cda3e6a6babf93a9d401bd49dc7b2fe3f57c05fd2152","observation_id":"856e9deb-edee-4a5a-902c-a66534d1a353","resolution":{"observed_at":"2026-08-06T16:26:35.137907Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.706337Z","title":"Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen","venue":null,"work_id":"df5842e1-1fc7-4dd4-8263-fbe673a0f5c6","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.148139Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:2f994217a01171a17f03c7811cdba52979c3b5a931c6ef58c0c501dcfa34c4c7","observation_id":"5765b131-0595-471a-9b77-d7145332c593","resolution":{"observed_at":"2026-08-06T16:26:36.711855Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.688596Z","title":"Dense network expansion for class in- cremental learning","venue":null,"work_id":"969c4d8a-0364-4c09-82f4-50d4e93dcfd4","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.165620Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:6355ad6e63bd808c45f842be1f5c70c0ca43cab84b1e79cc72faa3c009ed87ed","observation_id":"2c113551-971b-4d9d-a14b-85294427cf58","resolution":{"observed_at":"2026-08-06T16:26:36.694044Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2411.10928","last_updated":"2024-11-17T01:16:37Z","snapshot_observed_at":"2026-08-06T21:05:25.595199Z","submitted_at":"2024-11-17T01:16:37Z","title":"Learn from Downstream and Be Yourself in Multimodal Large Language Model Fine-Tuning","version":1},"cited_work":{"arxiv_id":"2411.10928","doi":null,"metadata_source":"pith","pith_arxiv_id":"2411.10928","snapshot_observed_at":"2026-08-06T16:26:35.777715Z","title":"Learn from Downstream and Be Yourself in Multimodal Large Language Model Fine-Tuning","venue":"cs.CL","work_id":"65baa3e2-ddc6-4f02-8a48-5c26cf36906a","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.174592Z"},"links":{"cited_paper":"/paper/2411.10928","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:2a663226d6c12cdc58b1d5be96ff5c126bb263f3ea06c21e250366e1a6d8a189","observation_id":"3dd389d7-4e46-4f77-95f2-713ec43ba291","resolution":{"observed_at":"2026-08-06T16:26:35.785326Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.182303Z","title":"Le, Yun-Hsuan Sung, Zhen Li, and Tom Duerig","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.182303Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:b181d859f9c0e2e91e5486034745a591a32b160e7f20b4fdede05d67542e8ac4","observation_id":"9af946eb-01c9-41dc-9125-fe6ceb8653a8","resolution":{"observed_at":"2026-08-06T16:26:35.182303Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.659419Z","title":"Belongie, Bharath Hariharan, and Ser-Nam Lim","venue":null,"work_id":"40a1c824-b6a5-4d4d-85e2-49567cf8f260","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.188400Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:b712028e1d06511c550bca68a181778624de7a26935618661c80ea7aaebc516a","observation_id":"aded1f6e-44b9-49a2-a786-1a3ad4c2f768","resolution":{"observed_at":"2026-08-06T16:26:36.663921Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.643511Z","title":"Class-incremental learning using diffusion model for distillation and replay","venue":null,"work_id":"6e0df7b9-6de8-42d3-b765-207fc141664f","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.196415Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:4202986d34c5b3ee2469631491f27090c070366829f6bed4ce8af66218795ac3","observation_id":"d1b094c3-8f44-4e15-988a-c54431765405","resolution":{"observed_at":"2026-08-06T16:26:36.648700Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.626188Z","title":"SDDGR: stable diffusion- based deep generative replay for class incremental object de- tection","venue":null,"work_id":"b5c7b620-25d3-4435-a4d2-e5dda613c7b4","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.204074Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:94242b6f69479cee67afab7acb8bee71ed25a6e9f8f0f13ff16ca5115bc7ad1e","observation_id":"117c2913-6d52-4663-9bc4-aaed4ea791bc","resolution":{"observed_at":"2026-08-06T16:26:36.632208Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1612.00796","last_updated":"2017-01-25T13:01:51Z","snapshot_observed_at":"2026-08-04T14:09:02.234596Z","submitted_at":"2016-12-02T19:18:37Z","title":"Overcoming catastrophic forgetting in neural networks","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1612.00796","snapshot_observed_at":"2026-08-06T16:26:35.214569Z","title":"Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.214569Z"},"links":{"cited_paper":"/paper/1612.00796","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:c511ec9eb90865a982c312e78c9f9dd1e4c28c954b78a72b81d07572c527f85f","observation_id":"a2069725-81b2-47ba-a55d-9320e92613b9","resolution":{"observed_at":"2026-08-06T16:26:35.214569Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.609121Z","title":"3d object representations for fine-grained categorization","venue":null,"work_id":"db116173-ba4a-45c3-81c0-d2506f375fed","year":2013},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.221684Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:7b43708c1585cda63a633c3fc4262b1c13517ce442670d48db3836ad2e4cfa28","observation_id":"65427441-6997-43f1-abcb-783af87867ee","resolution":{"observed_at":"2026-08-06T16:26:36.613804Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.591053Z","title":"Learning multiple layers of features from tiny images","venue":null,"work_id":"ff8a477d-d6bb-4fcb-8aae-802ae7a9498e","year":2009},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.229905Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:8c867c45b09d83b74057ddc8bbdbd0d2bd221d606d445563c1c9271e7497e07a","observation_id":"0f0051ca-2a83-4199-b2de-3c18278d2634","resolution":{"observed_at":"2026-08-06T16:26:36.596586Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"1810.10612","last_updated":"2018-10-24T20:41:13Z","snapshot_observed_at":"2026-07-06T07:10:25.716310Z","submitted_at":"2018-10-24T20:41:13Z","title":"Continual Classification Learning Using Generative Models","version":1},"cited_work":{"arxiv_id":"1810.10612","doi":null,"metadata_source":"pith","pith_arxiv_id":"1810.10612","snapshot_observed_at":"2026-08-06T16:26:35.713282Z","title":"Continual Classification Learning Using Generative Models","venue":"cs.LG","work_id":"9442cae0-ba07-46f4-b438-61dcfb15e5d4","year":2018},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.240398Z"},"links":{"cited_paper":"/paper/1810.10612","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:e5587e266bd1fb5adfb9a0895bd14939158ea442984c20a4d7f41f3f06ea7a1b","observation_id":"0d94308c-8d96-4a93-83e1-da95f5c1497e","resolution":{"observed_at":"2026-08-06T16:26:35.720758Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.568988Z","title":"Overcoming catastrophic forgetting by incremental moment matching","venue":null,"work_id":"ee8b0294-5ea8-4693-b7b9-23ab67b4e0c1","year":2017},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.247251Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:f7c3eb5f75a0fb3e24801f729359ba04b08528ecb9e87ad26b86b12bc52007ca","observation_id":"43a9a2fd-7585-407f-9d1b-74481df0a5bc","resolution":{"observed_at":"2026-08-06T16:26:36.576623Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2406.08478","last_updated":"2024-06-18T11:47:26Z","snapshot_observed_at":"2026-07-06T18:29:49.284504Z","submitted_at":"2024-06-12T17:59:07Z","title":"What If We Recaption Billions of Web Images with LLaMA-3?","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2406.08478","snapshot_observed_at":"2026-08-06T16:26:35.252979Z","title":"What if we recaption billions of web images with llama-3? CoRR, abs/2406.08478, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.252979Z"},"links":{"cited_paper":"/paper/2406.08478","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:24ab7bb8a23a180939cdbc91f336f17b8e746eb403e151717f77a4662fdc7cd3","observation_id":"4337a5b9-0985-43a8-89b6-cff4964300a7","resolution":{"observed_at":"2026-08-06T16:26:35.252979Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.546951Z","title":"Learning without forgetting","venue":null,"work_id":"9cac9721-26b7-4970-86b2-ca5acab1274c","year":2016},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.258294Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:b07e763ea592fb627504548d66c0e29d1f696c05d92aebb3b011a0e975c6ffec","observation_id":"2df8038a-0dbc-4edb-8560-756b12ae8124","resolution":{"observed_at":"2026-08-06T16:26:36.553269Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2312.03805","last_updated":"2023-12-06T15:54:05Z","snapshot_observed_at":"2026-07-06T16:57:59.000365Z","submitted_at":"2023-12-06T15:54:05Z","title":"SYNC-CLIP: Synthetic Data Make CLIP Generalize Better in Data-Limited Scenarios","version":1},"cited_work":{"arxiv_id":"2312.03805","doi":null,"metadata_source":"pith","pith_arxiv_id":"2312.03805","snapshot_observed_at":"2026-08-06T16:26:35.655890Z","title":"SYNC-CLIP: Synthetic Data Make CLIP Generalize Better in Data-Limited Scenarios","venue":"cs.CV","work_id":"d3340384-339b-4f6c-96ec-6ed2bd8f6520","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.264922Z"},"links":{"cited_paper":"/paper/2312.03805","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:a18cdbdefd4aa9db00911aea07005ac5354dc56626fc29e1c9c9cd79a1bedcf8","observation_id":"3da346e3-fa88-4f43-98b8-c57ac1a5e868","resolution":{"observed_at":"2026-08-06T16:26:35.665466Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2407.20756","last_updated":"2025-08-11T14:20:55Z","snapshot_observed_at":"2026-08-04T05:33:48.921191Z","submitted_at":"2024-07-30T11:57:40Z","title":"SynthVLM: Towards High-Quality and Efficient Synthesis of Image-Caption Datasets for Vision-Language Models","version":5},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2407.20756","snapshot_observed_at":"2026-08-06T16:26:35.272038Z","title":"Synthvlm: High-efficiency and high- quality synthetic data for vision language models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.272038Z"},"links":{"cited_paper":"/paper/2407.20756","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:49eb73f32be6f217a6fb07da0ce03a8096a243efb9d22006df9b4017362356c2","observation_id":"76b10177-e5f0-4fae-af7a-fe391e89e964","resolution":{"observed_at":"2026-08-06T16:26:35.272038Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.524511Z","title":"Gradient episodic memory for continual learning","venue":null,"work_id":"a489591b-f2ef-4825-9d2d-c71de5d02da6","year":2017},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.278214Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:4a6e3108f71bf3a4c0371c83c47ab7ab120e956a766e0a263adf7b324398128e","observation_id":"124c8f5d-7fa3-487b-a436-4041ae7b66c4","resolution":{"observed_at":"2026-08-06T16:26:36.530408Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.283302Z","title":"Decoupled weight decay regularization","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.283302Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:a8bdccf85d846dc8afc5da7969b024ee4d5e3bdbeb4c4f00625ee1d5e730f881","observation_id":"3c07401a-2939-4653-964f-42d76b5011a8","resolution":{"observed_at":"2026-08-06T16:26:35.283302Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1306.5151","last_updated":"2013-06-21T14:31:57Z","snapshot_observed_at":"2026-07-06T03:16:28.287173Z","submitted_at":"2013-06-21T14:31:57Z","title":"Fine-Grained Visual Classification of Aircraft","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1306.5151","snapshot_observed_at":"2026-08-06T16:26:35.288581Z","title":"Blaschko, and Andrea Vedaldi","venue":null,"work_id":null,"year":2013},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.288581Z"},"links":{"cited_paper":"/paper/1306.5151","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:79e726ceac7bff29699ebfa6994fc5ac8ac7feae21f1447b844bdcff45091e52","observation_id":"7068d601-5efd-403d-a23f-f12b0496113e","resolution":{"observed_at":"2026-08-06T16:26:35.288581Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.490962Z","title":"Diffclass: Diffusion-based class incremental learning","venue":null,"work_id":"1aa8a940-c80f-4043-8af4-8a60b64a0cb1","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":39,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.293918Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:fa9908819a2b7e9de1c17195f365e15fbf86f06530616e507ba6cc245a4cfc0e","observation_id":"c437f79a-1568-4088-9934-6e6146d5b8a3","resolution":{"observed_at":"2026-08-06T16:26:36.496960Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.468929Z","title":"Improving multimodal datasets with image captioning","venue":null,"work_id":"22ef4f1a-c1f4-49d2-8582-6720c14b915f","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":40,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.299357Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:507e53c7ad2e137e35b962387c130006ac8a2c04201ddc9cea6f80dbf0b58c85","observation_id":"8f6b41f2-88fd-4d59-a2f0-6ba49e0aa1fb","resolution":{"observed_at":"2026-08-06T16:26:36.475039Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.445456Z","title":"Automated flower classification over a large number of classes","venue":null,"work_id":"4b07208d-ebc4-4ad0-a54e-58d48a9048fb","year":2008},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":41,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.305779Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:1d109a0cd104450f7e237fde246935c7bc0760b5ed9bea8c9af7770fba863a2a","observation_id":"d13b4892-7eb1-450c-bcb5-b6714814dbbd","resolution":{"observed_at":"2026-08-06T16:26:36.452936Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.416124Z","title":"Parkhi, Andrea Vedaldi, Andrew Zisserman, and C","venue":null,"work_id":"6ad30557-c1d4-4768-93d1-20a56486b78d","year":null},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":42,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.311747Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:1b02f3fd7c1e4300434c719e7418e33114b89bce2a0eec0467f76028ed7caef7","observation_id":"ba9c580b-0120-4f1c-83db-c2c24b9d3eb7","resolution":{"observed_at":"2026-08-06T16:26:36.427848Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.391651Z","title":null,"venue":null,"work_id":"c079d686-62d9-4734-8043-5cc0b46461cb","year":2020},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":43,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.318366Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:038daddd61139648805de5b16b45c8c94ff9d4a6378d66f3453c6a222be41f89","observation_id":"bf3b9e51-6300-443d-8794-0d76ba815f9c","resolution":{"observed_at":"2026-08-06T16:26:36.402578Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.372209Z","title":"Learning transferable visual models from natural language supervision","venue":null,"work_id":"86ce8866-4060-44c0-9ea7-f2f49996d2df","year":2021},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":44,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.326613Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:ea8ddcf02df87600931f54247c1f66681d17a35610d730abc28cc2d4723568d7","observation_id":"8ce516a6-5abc-407d-90d7-fbcca6ba26fd","resolution":{"observed_at":"2026-08-06T16:26:36.378356Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.353334Z","title":null,"venue":null,"work_id":"218dc840-05bc-405b-9a99-15070b48c3f1","year":null},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":45,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.332299Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:4ca155dcf9d3f1b3f98e6704f1fab1c9d963f1b193dad7b978933f408d8ee985","observation_id":"5b7e59ca-c09c-4ece-b5cf-19ce0b591b02","resolution":{"observed_at":"2026-08-06T16:26:36.359269Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.333795Z","title":"High-resolution image syn- thesis with latent diffusion models","venue":null,"work_id":"27df8954-a9ea-4b39-927a-63d3814839e9","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":46,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.337737Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:4ef12936a887ef8ad0a03532a362857d14a5ec3973a18801ade78941bad9a185","observation_id":"02738c9b-c261-44f3-a3bf-a90dc36d775a","resolution":{"observed_at":"2026-08-06T16:26:36.338942Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.315683Z","title":"Imagenet large scale visual recognition challenge","venue":null,"work_id":"2748b9de-b1c7-49da-a0be-f27014c0bb66","year":2015},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":47,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.343127Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:57cc1cef83a6662b992ab2d3b57af6ad5b8878c8d338a719042e59bf36e2c668","observation_id":"9baf47ff-08f1-4b30-890d-20557c5c4dcb","resolution":{"observed_at":"2026-08-06T16:26:36.321133Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.296928Z","title":"Continual learning with deep generative replay","venue":null,"work_id":"90503368-bc9f-4e6c-82ea-4c7307433e12","year":2017},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":48,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.348421Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:029714d5432235aced9dc7b870bdb74fb326f93fe4e55e20e78a36cd87c0a310","observation_id":"ed815fdb-309a-448f-8c91-594e38b795a9","resolution":{"observed_at":"2026-08-06T16:26:36.302563Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.278806Z","title":"Is synthetic data all we need? benchmarking the robustness of models trained with synthetic images","venue":null,"work_id":"aae105b4-719f-405a-a32e-ef68782d5a7f","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":49,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.353949Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:a4418af22f6515a014d8ddaae483aa227377ae65b57ceccc4732425db92689f3","observation_id":"27e1ce66-aa5c-45b0-96a3-c552b372bd2e","resolution":{"observed_at":"2026-08-06T16:26:36.284482Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.260112Z","title":"StableRep: Synthetic images from text-to- image models make strong visual representation learners","venue":null,"work_id":"04d4f6a2-532c-4285-954a-8b199d637808","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":50,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.360380Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:072bfa3e16359e5897f7181d14515ff48159d20bafe729439baed9a168914971","observation_id":"c398e235-75b9-48e2-bb0c-7aea10858343","resolution":{"observed_at":"2026-08-06T16:26:36.266137Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.239238Z","title":"Learning vision from models rivals learning vision from data","venue":null,"work_id":"7e8a002e-23da-4b78-ab7e-f7e80d0b0437","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":51,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.365255Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:45f01066db24ed32103669b6a716ba862be7a661e44d96eac87222287a449dc1","observation_id":"3c86d73c-5999-4dbe-a7a3-53f57636bb98","resolution":{"observed_at":"2026-08-06T16:26:36.245446Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.221107Z","title":"S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning","venue":null,"work_id":"5151ff4f-40ca-4ebe-ae36-6e2d4140d7a6","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":52,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.370233Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:431d7ecca641e51813f822003750d3b5a777eec1d344cd16a02aca67fef1e6d6","observation_id":"50f66cd8-1e6b-4d19-90bf-c60ccdcdc588","resolution":{"observed_at":"2026-08-06T16:26:36.226946Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.202160Z","title":"Dy, and Tomas Pfister","venue":null,"work_id":"6954fbbb-92ef-4918-8f1e-ba88023a1970","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":53,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.376146Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:58fe9e7c876c52459873842b609e0025ada785149b8299f1b6469052e12f4995","observation_id":"2540e778-982d-4e66-828d-d05790cb50fc","resolution":{"observed_at":"2026-08-06T16:26:36.207526Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.182250Z","title":"Dy, and Tomas Pfister","venue":null,"work_id":"63d4cf01-0e0a-4008-9e4a-139c13ad7028","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":54,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.381267Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:2817d3c9b2629f4a9404458dcf1b455d1ee1074daf199c6e352b81b87b953bc3","observation_id":"753247d8-4ae2-4c77-b798-4093bd302740","resolution":{"observed_at":"2026-08-06T16:26:36.188577Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2109.01903","last_updated":"2022-06-21T21:50:28Z","snapshot_observed_at":"2026-07-06T11:44:19.030296Z","submitted_at":"2021-09-04T17:11:28Z","title":"Robust fine-tuning of zero-shot models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2109.01903","snapshot_observed_at":"2026-08-06T16:26:35.386481Z","title":"Robust fine-tuning of zero-shot models","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":55,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.386481Z"},"links":{"cited_paper":"/paper/2109.01903","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:d30cbf1fe5676009e13aae9b766e6b3c08d9c92a07294048a5ec4f9dbfcb1558","observation_id":"8dcc18c0-98c1-454e-b6ea-c6189ddaeb56","resolution":{"observed_at":"2026-08-06T16:26:35.386481Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.158469Z","title":"Syn- thetic data is an elegant GIFT for continual vision-language models","venue":null,"work_id":"318a491c-6d5d-42f4-8a8e-c386649a5f50","year":2025},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":56,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.392189Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:b714ba050f0dacbd465aaebbf64007465792bff40027a94dc75feb91e52b79b6","observation_id":"8ebd7431-77fc-405b-8664-65f10d92af55","resolution":{"observed_at":"2026-08-06T16:26:36.166201Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.397189Z","title":"Ehinger, Aude Oliva, and Antonio Torralba","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":57,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.397189Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:922ae7cf30f298603c8c2ab16b28fe2afd92d26b480f8929af453932477af955","observation_id":"ca3917d2-1d9c-4ea5-9cf6-b3f90a9a833a","resolution":{"observed_at":"2026-08-06T16:26:35.397189Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.117889Z","title":"DER: dynam- ically expandable representation for class incremental learn- ing","venue":null,"work_id":"0f22212c-4807-4ab4-8782-7821397f49d0","year":2021},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":58,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.405760Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:a136077d117bac74f9a69c112338751991b61f8b7102c6919686e11b435a76c5","observation_id":"5d853061-3916-4db2-b894-2dd23ad965c2","resolution":{"observed_at":"2026-08-06T16:26:36.124562Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.094972Z","title":null,"venue":null,"work_id":"cef7e530-6f3f-494d-a944-28da13d5ff90","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":59,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.412605Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:bdd448594fd5f41946326f65a14fac7a4c1d20e1abdde8038bfa29962fc07265","observation_id":"f947eeb3-e44e-4d56-bca6-9e3c863d7d95","resolution":{"observed_at":"2026-08-06T16:26:36.100836Z","resolver_source":"raw_fallback","status":"unresolved"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.071964Z","title":"Controlled training data generation with diffusion mod- els","venue":null,"work_id":"c05f9f78-daaa-40ae-9e04-5bb8773df19c","year":2025},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":60,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.419477Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:35b1dab296697cc461e8575bc6f09e0e1005ed00665993845e608a1d80a1242f","observation_id":"5ef6e0a5-ad4a-41d2-83e4-bdd5f9d72886","resolution":{"observed_at":"2026-08-06T16:26:36.079185Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.052858Z","title":"Lifelong learning with dynamically expandable net- works","venue":null,"work_id":"6210d4a1-ec00-469f-86fd-c919045dd0ac","year":2018},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":61,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.425895Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:3d016d49bc432cb5ce67ea301df5980c98570ea48e8b0630321cc325246b85f7","observation_id":"bcdf61bb-a8a0-4940-b5a3-4f30bfa7021f","resolution":{"observed_at":"2026-08-06T16:26:36.059223Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.030786Z","title":"Coca: Contrastive captioners are image-text foundation models","venue":null,"work_id":"949e9404-4377-4612-8d92-6b0acbc1febe","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":62,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.431898Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:1accabf97538bada611aee487d70aac6517d934598b7f72c2c1c3bf27502d381","observation_id":"393dc943-036d-4e33-bba2-ec070aa55fd2","resolution":{"observed_at":"2026-08-06T16:26:36.038689Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:36.009209Z","title":"Boosting continual learning of vision-language models via mixture-of-experts adapters","venue":null,"work_id":"5b86f331-6bc7-4c97-9ccc-f484709fc6ab","year":2024},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":63,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.438220Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:0d2b5d5068885336f2dd1649c80b7f2de47635592445cbe953bb06b3f9104afb","observation_id":"e9d2eb05-ad30-48e0-85e8-0a0bf417b3e8","resolution":{"observed_at":"2026-08-06T16:26:36.014985Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2111.11432","last_updated":"2021-11-22T18:59:55Z","snapshot_observed_at":"2026-07-06T12:11:02.119174Z","submitted_at":"2021-11-22T18:59:55Z","title":"Florence: A New Foundation Model for Computer Vision","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2111.11432","snapshot_observed_at":"2026-08-06T16:26:35.445457Z","title":"Florence: A new foundation model for computer vision","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":64,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.445457Z"},"links":{"cited_paper":"/paper/2111.11432","citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:e33fd79dc3dd42b89690e7e29db6ee07c5a9548ceb45d7d2bb1e93ac07c8fc52","observation_id":"73786bac-8b93-49e8-b287-e930a7f00028","resolution":{"observed_at":"2026-08-06T16:26:35.445457Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.984928Z","title":"Contin- ual learning through synaptic intelligence","venue":null,"work_id":"81ae44bf-4b45-486f-93e6-577dc7c33a5e","year":2017},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":65,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.453258Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:2bfa5a2696c5b248cdb25de68aee06345b7fd90de61b0c54a42c6aedf091984a","observation_id":"86599f12-d084-45d7-9b64-6589340eba32","resolution":{"observed_at":"2026-08-06T16:26:35.992001Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.956270Z","title":"Lit: Zero-shot transfer with locked-image text tuning","venue":null,"work_id":"d16c201c-20c2-431e-b2b8-9c9163d09bb8","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":66,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.458238Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:a62ee5ffb4056016ef16df4ade5e1162aa42d776a1feffa0befa64a7f3a923fe","observation_id":"6e7835c4-c6ae-4439-986c-c65e1cb446e5","resolution":{"observed_at":"2026-08-06T16:26:35.963785Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.933761Z","title":"Preventing zero-shot transfer degradation in continual learning of vision-language models","venue":null,"work_id":"028a514e-cc5d-4da3-808a-e82a899d685d","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":67,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.463331Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:5f1263a407958e04e07a14b8101458b25d0a6202d4cd0497270e4e0a7a2e8f35","observation_id":"85aa4f5b-11d5-47ad-a0ba-cff6a482250c","resolution":{"observed_at":"2026-08-06T16:26:35.939775Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.907747Z","title":"Preventing zero-shot transfer degradation in continual learning of vision-language models","venue":null,"work_id":"eac4639b-0882-4a68-abda-7c37aa1fbe80","year":2023},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":68,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.469907Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:1e67e2717dee635335a8aa3648d00b9710fb2ab25f70ca82324658aee25e0824","observation_id":"01c01543-e306-4f2d-a47a-1a4d5567aaa0","resolution":{"observed_at":"2026-08-06T16:26:35.913806Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.888421Z","title":"Learning with- out forgetting for vision-language models","venue":null,"work_id":"35d366ff-f627-4fda-a61c-ed30add5ca2a","year":2025},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":69,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.475543Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:f3ab4644650b99ae392735bbe0f859cc945345b142e32479dd4f264a2fd229f8","observation_id":"defae5cf-7b45-4c8b-bd0a-e45b6d0a4e9f","resolution":{"observed_at":"2026-08-06T16:26:35.893970Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.867686Z","title":"Conditional prompt learning for vision-language models","venue":null,"work_id":"42803f2c-f5d8-4261-8c78-670e68d5d909","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":70,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.484596Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:e256c5172f83940c2b54c05e20264f9f93610457dc4c40087383db906755c6b6","observation_id":"1a7220a1-a264-4bb4-80ca-cfc24c55bd69","resolution":{"observed_at":"2026-08-06T16:26:35.873832Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.845307Z","title":"Learning to prompt for vision-language models","venue":null,"work_id":"72049bf7-96a6-4d42-9a90-0520c9c52192","year":2022},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":71,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.491002Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:cfed74b0c0216758f3da3bdbb755a7747b94ebeaeb9f594070429bd23af5ff81","observation_id":"e39ead18-e2bd-4384-b47c-ce70f9bd67ea","resolution":{"observed_at":"2026-08-06T16:26:35.851499Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-06T16:26:35.824548Z","title":"6 and Tab","venue":null,"work_id":"01c3d217-78f7-4578-a50c-8b7b7357c0e1","year":null},"citing_paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning","version":2},"reference_index":72,"source":"pdf_text","source_observed_at":"2026-08-06T16:26:35.498234Z"},"links":{"citing_paper":"/paper/2507.13568"},"observation_digest":"sha256:769e3a20599a4823493f6c63d5fc86df874d0d72b0a02d4965bca478cc76549d","observation_id":"1430810a-ddf8-4bfb-a7d2-9d6ee8e816f6","resolution":{"observed_at":"2026-08-06T16:26:35.829712Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}}],"paper":{"arxiv_id":"2507.13568","last_updated":"2025-07-29T03:29:57Z","latest_version":2,"primary_category":"cs.CV","snapshot_observed_at":"2026-08-06T16:19:31.949430Z","submitted_at":"2025-07-17T23:08:29Z","title":"LoRA-Loop: Closing the Synthetic Replay Cycle for Continual VLM Learning"},"reference_resolution":{"displayed":72,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":15,"verified_exact":3,"verified_fuzzy":54},"total_outbound_references":72},"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":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"thesis":"As of 8 August 2026, this Paper Citation Record lists 72 of 72 outbound references and 0 inbound Pith citation observations for arXiv:2507.13568."}