Tech

Mage-VL-Cowork: A Full Training Recipe to Learn Hybrid Computer Use From Evolved Harnesses

By Xiaoyi Zhang, Haoqing Wang, Xinjie Zhang, Xiang Ming, Xing W · arXiv cs.CL

cs.CLcs.LG

arXiv:2610.40747v1 Announce Type: new Abstract: Effective computer-use agents (CUAs) should seamlessly combine visual GUI interaction with code and terminal execution within a single policy. We present Mage-VL-Cowork, a 4B hybrid…

Published: November 2, 2026 (1 h ago)
Outlet: arXiv cs.CL
Section: Tech
Length here: Headline and 30-word snippet. The full article is at the source.

More from arXiv cs.CL and around the same time

An open laptop and a stack of books on a desk

Adaptive Personalization Under Preference Epistemic UncertaintyarXiv:2610.40930v1 Announce Type: new Abstract: Human preferences change over time: an established preference may remain valid over time, become uncertain as context shifts, or be explicitly revised. Effective personalization…arXiv cs.CL · 1 h ago · reads at arXiv cs.CL

Better Retrievers, Worse Labels: How Incomplete Relevance Judgments Distort Abstention Evaluation in Conversational RAGarXiv:2610.40966v1 Announce Type: new Abstract: Conversational RAG benchmarks such as MTRAG and RETECO evaluate retrieval and generation end to end: each system answers from its own retrieval, so abstention must be scored against…arXiv cs.CL · 1 h ago · reads at arXiv cs.CL

CoFiBreak: A Coarse-to-Fine Diversity Reward for Effective Multi-Turn JailbreakarXiv:2610.40926v1 Announce Type: new Abstract: Automated red teaming trains an attacker large language model (LLM) to elicit harmful responses from a victim LLM, but reinforcement learning can collapse attackers onto a narrow…arXiv cs.CL · 1 h ago · reads at arXiv cs.CL

Do Open Language Models Invent Scripture? BILAL-BENCH: Benchmarking Islamic-Source Fabrication in BengaliarXiv:2610.40996v1 Announce Type: new Abstract: When a language model invents a Qur'anic verse or hadith, it gives false authority to text that readers may treat as sacred. Existing Islamic hallucination benchmarks are largely…arXiv cs.CL · 1 h ago · reads at arXiv cs.CL

EDU-Jev Training System-One Decision Models for EducationarXiv:2610.41079v1 Announce Type: new Abstract: Many automated decisions in education involve choosing from a fixed set of options: assigning a rubric score, checking whether an answer meets a marking criterion, or identifying…arXiv cs.CL · 1 h ago · reads at arXiv cs.CL

Fewer Heads, Fewer Memories: Pruning Redundant Attention Heads Suppresses Verbatim MemorisationarXiv:2610.40974v1 Announce Type: new Abstract: Large language models have been shown to memorise and reproduce verbatim sequences from their training data, raising concerns about privacy leakage, intellectual property exposure…arXiv cs.CL · 1 h ago · reads at arXiv cs.CL

This page holds the headline and a short snippet as published by arXiv cs.CL, whose feed they came from. The story itself, the photography and the rights stay with them, and the button above goes straight to their site. We keep this page so the headline has a stable address and a record of when it ran.