Mid-Harness is introduced, which samples and verifies candidate actions before forwarding one for execution, while keeping the generator and harness unchanged, and identifies action scaling as a promising target for test-time compute scaling in terminal agents.
Minki Kang, Ryo Hachiuma, Shao-Kun Zhang et al.
· arXiv.org · 3 citations
Save
{ copied = true; setTimeout(() => copied = false, 1500) })"
class="icon-btn" aria-label="Copy link">
{ copied = 'apa'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy APA
Copied ✓
{ copied = 'mla'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy MLA
Copied ✓
{ copied = 'bibtex'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy BibTeX
Copied ✓
Preprint
Aug 2026
A societal bias evaluation method for large vision-language models (LVLMs) in the era of strong safety guardrails is proposed, finding that all models undesirably use user demographic information in person-irrelevant tasks.
Yusuke Hirota, Michael Boone, Arun George Zachariah et al.
· 0 citations
Save
{ copied = true; setTimeout(() => copied = false, 1500) })"
class="icon-btn" aria-label="Copy link">
{ copied = 'apa'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy APA
Copied ✓
{ copied = 'mla'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy MLA
Copied ✓
{ copied = 'bibtex'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy BibTeX
Copied ✓
Zone of Proximal Policy Optimization (ZPPO), inspired by Vygotsky's zone of proximal development, is introduced, which outperforms off/on-policy distillation and GRPO, with the largest gains at the smallest scale.
Byung-Kwan Lee, Ximing Lu, Shi-Zhe Diao et al.
· arXiv.org · 2 citations
Save
{ copied = true; setTimeout(() => copied = false, 1500) })"
class="icon-btn" aria-label="Copy link">
{ copied = 'apa'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy APA
Copied ✓
{ copied = 'mla'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy MLA
Copied ✓
{ copied = 'bibtex'; setTimeout(() => { copied = null; open = false }, 1000) })"
class="flex w-full items-center justify-between rounded-lg px-3 py-2 text-left text-sm hover:bg-gray-100 dark:hover:bg-ink-800">
Copy BibTeX
Copied ✓