{"paper":{"title":"Macaron-V1: Towards Open Continual Learning with Self-Improvement and Mixture-of-LoRA","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Ada Zhou, Adrian Zhou, Alexy Li, Allen Lin, Amelia Chen, Andrew Chen, Andrew Lei, Anya Zhang, Arthur Fu, Asher Cai, Atlas Zeng, Bo Wu, Charles Huang, Chen Ling, Cleon Cheng, Cole Qiao, Danney Zeng, Dawn Li, Di Zhang, Elliot Lin, Evelyn Ye, Fancy Kong, Fan Lin, Fiona Ye, Guian Qiu, Hailee Hou, Hera Feng, Huan Feng, Ina Ye, Jaron Lee, Jingwei Cao, Josh Ying, Jun Gao, Juno Zhu, Kaijie Chen, Kairus Liu, Kaixuan Fan, Kieran Liu, Kuss Koo, Logan Liu, Lucian Li, Maeve Luo, Mia Zhang, Miles Jiang, Mind Lab: Vin Bo, Murphy Zhuang, Neo Liu, Niko Song, Nolan Ho, Nora Jiang, Ori Hong, Piers Hua, Pony Ma, Pyke Han, Qiuyu Jin, Ray Li, Regis Ye, Ricardo Li, Rio Yang, Salmon Zhan, Sentry, Smith Li, Song Cao, Steven Chiang, Sueky Zhang, Theo Li, Verity Niu, Vic Cao, Vincent Wang, Vince Qu, Wei Zhao, Xiang Liu, Yuhua Zhou, Yuxin Lu, Yuyi Jiang","submitted_at":"2026-08-10T16:39:55Z","abstract_excerpt":"Macaron-V1 is an open agent-model family for experiential intelligence: learning from experience in real environments and continuing to learn after deployment. It is organized around two system goals. Adaptation is pursued through recursive improvement of versioned model-harness pairs, where experience from one configuration is evaluated under an external contract and used to construct its successor. Collaboration is pursued via the Mixture-of-LoRA (MoL) architecture that freezes a base model, composes specialist LoRA adapters, and selects one LoRA per user turn. The flagship Macaron-V1-Venti "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.09819","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2608.09819/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}