analiza aiqbench
This arXiv preprint proposes training language-model agents to learn reusable "experience" from interaction, using verifiable task families that share a common tool or workflow. The reported gains on SpreadsheetBench, APEX-Agents, and AutomationBench are notable, but they come from a single fine-tuned model (Qwen3.5-122B-A10B) and benchmarks chosen by the authors, so independent replication matters. The most useful signal for practitioners is the claim that cross-task experience can transfer as a model capability rather than living in external memory scaffolds.
MetaExp: Learning to Self-Improve through Experience via Verifiable Cross-Task Environment Scaling
arXiv:2610.21492v1 Announce Type: new Abstract: Language-model agents can accumulate interaction feedback, but reliably turning that feedback into reusable experience remains challenging. Existing methods either optimize external experience scaffolds or train experience-related behavior on specific task distributions, leaving open whether experience learning itself can become a transferable model capability. Training such a capability calls for supervision beyond isolated tasks, which mainly support in-episode problem solving rather than revealing what knowledge remains useful across episodes. In this work, we introduce MetaExp, a scalable framework that trains this capability through verifiable task families with controlled cross-task structure. Each family preserves a reusable factor while varying task-specific realizations, enabling transfer without direct solution reuse. We propose two complementary designs: Shared-Tool families keep black-box tools fixed across tasks to elicit operational experience, while Shared-Workflow families preserve latent procedures across changing realizations to elicit procedural experience. We scale environments under both designs, collect and filter sequential interaction trajectories, and fine-tune Qwen3.5-122B-A10B on 4,956 trajectories from 1,000 task families. With parameters frozen during downstream evolution, MetaExp improves by +8.5 points on SpreadsheetBench, peaks at +2.3 on APEX-Agents, and improves by +14.8 on AutomationBench. The gains persist under domain-excluded training, supporting experience learning as a transferable capability for continued non-parametric improvement after deployment.
Abstract:Language-model agents can accumulate interaction feedback, but reliably turning that feedback into reusable experience remains challenging. Existing methods either optimize external experience scaffolds or train experience-related behavior on specific task distributions, leaving open whether experience learning itself can become a transferable model capability. Training such a capability calls for supervision beyond isolated tasks, which mainly support in-episode problem solving rather than revealing what knowledge remains useful across episodes. In this work, we introduce MetaExp, a scalable framework that trains this capability through verifiable task families with controlled cross-task structure. Each family preserves a reusable factor while varying task-specific realizations, enabling transfer without direct solution reuse. We propose two complementary designs: Shared-Tool families keep black-box tools fixed across tasks to elicit operational experience, while Shared-Workflow families preserve latent procedures across changing realizations to elicit procedural experience. We scale environments under both designs, collect and filter sequential interaction trajectories, and fine-tune Qwen3.5-122B-A10B on 4,956 trajectories from 1,000 task families. With parameters frozen during downstream evolution, MetaExp improves by +8.5 points on SpreadsheetBench, peaks at +2.3 on APEX-Agents, and improves by +14.8 on AutomationBench. The gains persist under domain-excluded training, supporting experience learning as a transferable capability for continued non-parametric improvement after deployment.
Artificial Intelligence (cs.AI)
arXiv:2610.21492 [cs.AI]
(or arXiv:2610.21492v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.21492 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both 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.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 294 cuvinte
Subiecte conexe
Models Can't Remember Their Training. Neither Can You.
LessWrong · 19 iul. 2026
Game Design Creativity evals
LessWrong · 12 oct. 2026
Expenditure Horizon: Measuring Optimization Ability, with an Application to NanoGPT
METR · 21 iul. 2026
ASCENT: Online Test-Time Training of Long-Horizon Agents via Self-Distillation of Verified Experience
arXiv · 6 oct. 2026