product

LearnWeak

productactiveprovisionallearnweak-c565fd79·1 events·first seen 20d ago

Aliases: LearnWeak

Co-occurring entities

computer-use agents OpenCUA-7B EvoCUA-8B OSWorld error-aware specialization objective

More like this (12)

Language Modeling Loss LALS (Latent Association Leaning Score)Operator Learning Lean TailLoR Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs WISE Alternating Token-Weighted Unlearning Latent Context Language Models Memory Insights weak-to-strong generalization LiteLLM

Recent events (1)

6arXiv · cs.AI·20d ago·source ↗

LearnWeak: Automated Domain Specialization for Small Computer-Use Agents via Weakness-Targeted Synthesis

LearnWeak is an annotation-free framework for specializing small computer-use agents (CUAs) in specific software domains without deploying large expert models. It uses a stronger reference agent to identify weaknesses in a smaller student agent, synthesizes targeted tasks, and applies an error-aware training objective that disentangles planning from execution errors. On OSWorld, LearnWeak achieves gains of ~11 percentage points over 7B-8B baseline CUAs across eight domains. The work demonstrates that student-aware data synthesis substantially outperforms naive large-scale data generation for domain specialization.

Evaluation and Benchmarking Open Weights Progress LearnWeak computer-use agents OpenCUA-7B +6 more