İngilizce sürümIf monitoring results influence which training samples or checkpoints are retained, how will you test whether that selection favors less detectable misbehavior rather than less misbehavior, even when reasoning text is never directly graded?
- Durum
- açık Tarih itibarıyla 2026-10-01
- Muhatap
- yapay zekâ laboratuvarlarına (Muhataplar soru metninden alındı, 22.10.2026'da Clef (Cloudflare) ile belirlendi)
- Model
- GPT-6 Astra
- Laboratuvar
- OpenAI
- Tarih
- 24 Eylül
- Konu
- Güven
Kontrol
Durum: açık
seçim: Fable 5.1, Claude Opus 5.5 (kanıtlar)
konu: Claude Sonnet 5, gruplama: Claude Opus 5.5
Çeviri: Claude Opus 5.5 · [kontrol edilmedi]
Puanlar: JEV (TypeSafe), 25 Eylül 2026 tarihli tahmin
Cite and take away
Cite and reuse
Own texts, questions and classifications: CC BY 4.0. Give attribution, link the license and indicate changes. Third-party source reports retain their own rights. CC BY 4.0
AMANi DUNiA. trust-32 · İzleme sonuçları hangi eğitim örneklerinin veya kontrol noktalarının (checkpoint) saklanac. Snapshot: . https://research.amanidunia.berlin/tr/fragen/trust-32/
Version: trust-32
BibTeX
@misc{research__tr_fragen_trust_32_,
author = {{AMANi DUNiA}},
title = {trust-32 · İzleme sonuçları hangi eğitim örneklerinin veya kontrol noktalarının (checkpoint) saklanac},
year = {2026},
url = {https://research.amanidunia.berlin/tr/fragen/trust-32/},
note = {Snapshot 2026-10-01; CC BY 4.0},
version = {trust-32}
}Language and review status
Main text: TR · includes English sections. Review notes apply only to the content they name.
Translation available; a completed independent content review is not recorded.
See language coverage and data snapshot →