{"slug": "continual-arc-measuring-skill-acquisition-over-a-working-lifetime", "title": "Continual-ARC: Measuring Skill Acquisition over a Working Lifetime", "summary": "A paper submitted to arXiv on 10 November 2026 proposes Continual-ARC, a lifelong benchmark built from the three ARC-AGI editions that measures skill acquisition by returning the same tasks after dormant gaps with fresh instances while new tasks keep arriving. In experiments with an agent harness on Qwen3.5-9B and GLM-5.3 Flash, harness persistence generally did not improve memorization unless given explicit expert guidance, most harness-based methods learned less than a floor that simply caches each task's demonstrations, and no harness-based method gained on unseen tasks, whereas weight-based learning showed substantial gains. The authors say the results open the door to weight-based continual learning that could teach models to drive harnesses in new domains.", "body_md": "# Computer Science > Machine Learning\n\n  [Submitted on 10 Nov 2026]\n\n# Title:Continual-ARC: Measuring Skill Acquisition over a Working Lifetime\n\n[View PDF](http://arxiv.org/pdf/2611.14505v1)\n\n[HTML (experimental)](https://arxiv.org/html/2611.14505v1)\n\nAbstract:Agent harnesses have emerged as a promising mechanism for continual learning, alongside test-time parametric learning. However, the current benchmarks have several limitations: they generally focus on a single applied domain, serve each instance only once, and assume a single family of learner, so memory-based and weight-based systems are rarely compared on the same scale. We propose Continual-ARC, a lifelong stream built from the three ARC-AGI editions, whose tasks need no domain knowledge. The same tasks return after dormant gaps with fresh instances, while new tasks keep arriving. On each instance, the learner may pay for demonstrations, submit answers, or act inside the harness, with retries permitted on incorrect submissions, which are paid for. Each instance is scored based on the share of this feedback budget consumed. Since a learner may keep anything between instances, memory-based and weight-based systems are scored on the same scale. After the stream, the learner is frozen and tested on fresh instances of the tasks it met, which measures memorization, and on unseen tasks, which measures generalization. We run an agent harness with frozen weights, with and without persistence, on Qwen3.5-9B and GLM-5.3 Flash, comparing them with two prompting floors, a program-induction method and Qwen3.5-9B's weights trained to imitate GLM-5.3 Flash. Persistence in the harness generally does not improve memorization unless provided with explicit expert guidance. Most harness-based methods still learn less than a floor that simply caches each task's demonstrations, across all models we tested. No harness-based method gains on unseen tasks, whereas weight-based learning shows substantial gains. This opens the door to future weight-based continual learning that could teach models to drive harnesses in new domains.\n    \n\n### Additional Features\n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer \n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers \n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps \n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations \n\n*(*[What are Smart Citations?](https://www.scite.ai/))\n# Code, Data and Media Associated with this Article\n\nalphaXiv \n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers \n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub \n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub \n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face \n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast \n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))\n# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower \n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender \n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))\nIArxiv Recommender\n\n*(*[What is IArxiv?](https://iarxiv.org/about))\n# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [**Learn more about arXivLabs**](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/continual-arc-measuring-skill-acquisition-over-a-working-lifetime", "canonical_source": "http://arxiv.org/abs/2611.14505v1", "published_at": "2026-11-11 16:49:50+00:00", "updated_at": "2026-11-11 17:21:12.459711+00:00", "lang": "en", "topics": ["ai-research", "machine-learning", "ai-agents", "large-language-models", "artificial-intelligence"], "entities": ["arXiv", "Continual-ARC", "ARC-AGI", "Qwen3.5-9B", "GLM-5.3 Flash"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/continual-arc-measuring-skill-acquisition-over-a-working-lifetime", "markdown": "https://wpnews.pro/news/continual-arc-measuring-skill-acquisition-over-a-working-lifetime.md", "text": "https://wpnews.pro/news/continual-arc-measuring-skill-acquisition-over-a-working-lifetime.txt", "jsonld": "https://wpnews.pro/news/continual-arc-measuring-skill-acquisition-over-a-working-lifetime.jsonld"}}