CANLI
xAI, Imagine API’yi 2.0’a Yükseltmeye Hazırlanıyor: Görüntü ve Video Tek…·Microsoft MAI-Cyber-1-Flash’ı Duyurdu·Moonshot AI, Kimi K3 Model Ağırlıklarını ve Teknik Raporunu Açık…
12 Nov 2026 · 19:43 GMT+3
Ai Haber – Türkiyenin Yapay Zeka Haber Portalı
ARAşTıRMA · DOğAL DIL İşLEME arXiv:2611.14480 10 Kas 2026 · v1

Pterodactyl: Pushing the Limits of Efficient Automated Translation Evaluation

Evelyn Y. Garland, Carola F. Berger

YAYIN:10 Kas 2026 ALAN:cs.CL OKUMA:2

Özet

In this paper, we present 1020 Group's submission to the WMT26 shared task on automated quality evaluation. Our system, Pterodactyl, was engineered specifically for subtasks 2 and 3, that is, segment-level quality score prediction and the detection of error-free segments. Pterodactyl is a highly customizable, ultra-low cost system optimized for binary classification in translation triage and data curation. It is based on an LLM-as-a-judge approach, with prompt engineering and model selection guided by ROC analysis. We also developed several variants of Pterodactyl that were specifically adapted for subtasks 3.

Özetle: In this paper, we present 1020 Group's submission to the WMT26 shared task on automated quality evaluation.

Özet

In this paper, we present 1020 Group's submission to the WMT26 shared task on automated quality evaluation. Our system, Pterodactyl, was engineered specifically for subtasks 2 and 3, that is, segment-level quality score prediction and the detection of error-free segments. Pterodactyl is a highly customizable, ultra-low cost system optimized for binary classification in translation triage and data curation. It is based on an LLM-as-a-judge approach, with prompt engineering and model selection guided by ROC analysis. We also developed several variants of Pterodactyl that were specifically adapted for subtasks 3.

Orijinal Özet (İngilizce)

In this paper, we present 1020 Group's submission to the WMT26 shared task on automated quality evaluation. Our system, Pterodactyl, was engineered specifically for subtasks 2 and 3, that is, segment-level quality score prediction and the detection of error-free segments. Pterodactyl is a highly customizable, ultra-low cost system optimized for binary classification in translation triage and data curation. It is based on an LLM-as-a-judge approach, with prompt engineering and model selection guided by ROC analysis. We also developed several variants of Pterodactyl that were specifically adapted for subtasks 3.

Kaynak: arXiv:2611.14480 · PDF

BibTeX

@article{garland2026pterodactyl,
  title   = {Pterodactyl: Pushing the Limits of Efficient Automated Translation Evaluation},
  author  = {Evelyn Y. Garland and Carola F. Berger},
  journal = {arXiv preprint arXiv:2611.14480},
  year    = {2026},
  url     = {https://arxiv.org/abs/2611.14480}
}

Tartışma

Bu habere emoji ile tepki ver

Hizli:

Henüz yorum yok. İlk yorumu siz yapın!

Yapıcı ve saygılı yorumlar bekliyoruz. Topluluk kuralları