Critical Error Detection on WMT 22 (test)
0.92MCCGPT-4o-mini
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4o-miniFine-tuning=true, Inference Mode=zero-shot, Decision Rule=majority voting2026.02 | 0.92 | 91 | 99 | |
| LLaMA-3.1-8BFine-tuning=true, Inference Mode=zero-shot, Decision Rule=majority voting2026.02 | 0.87 | 87 | 99 | |
| GPT-OSS-20BFine-tuning=true, Inference Mode=zero-shot, Decision Rule=majority voting2026.02 | 0.39 | 39 | 91 |