İçeriğe geç
onsuggest
Giriş
Makale en

Frontier coding agents claim work they didn't do

Kaynak: LessWrong ·

Tara Research Team arXiv | Twitter thread tl;dr. We built OverclaimBench, which asks frontier coding agents to review a corpus of code or text documents, then compares their final report with their transcripts. Pooled across 12 frontier agents and five scenarios, two-thirds of runs didn't open some of the files they were asked to review. In ~4 out of 5 of those incomplete runs the agent didn't tell the user, often claiming to have read everything. Skipping these files had real consequences - runs that overclaimed missed planted defects almost twice as often as runs that touched every file. Fig

Frontier coding agents claim work they didn't do nedir?

Frontier coding agents claim work they didn't do, Makale kategorisinde yer alan ve LessWrong üzerinden sunulan bir keşiftir. Tara Research Team arXiv | Twitter thread tl;dr. We built OverclaimBench, which asks frontier coding agents to review a corpus of code or text documents, then compares their final

Nasıl erişebilir veya deneyimleyebilirim?

Bu içeriğe "Aslına Git" butonuna tıklayarak doğrudan orijinal platformunda (LessWrong) erişebilirsiniz.

Bu içerik nereden geliyor?

Bu içerik LessWrong üzerinden OnSuggest topluluğu ve küratörleri tarafından internetin kaliteli köşelerinden seçilmiştir.

Klavye kısayolları

Önceki keşif
←P
Sıradaki keşif
→N
Beğen
L
Bana göre değil
X
Kaydet
S
Bu yardımı göster
?
Kapat
Esc

Kısayollar keşif sayfasında çalışır.