Classification on GoEmotions (Accuracy and Macro-F1)
81.63Accuracyslm-judge
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| slm-judge2026.04 | 81.63 | 63.8 | |
| slm-judgeDropout=p = 0.12026.04 | 79.37 | 49.98 | |
| slm-judgeTraining Strategy=Early Stopping, Data Augmentation=false2026.04 | 79.17 | 56.85 | |
| slm-judgeTraining Strategy=Full SFT, Data Augmentation=false2026.04 | 78.19 | 49.67 | |
| GPT-5-nanoEvaluation Protocol=Zero-shot2026.04 | 51.77 | 39.69 | |
| GPT-5.2-chatEvaluation Protocol=Zero-shot2026.04 | 50.62 | 40.99 | |
| GPT-5.2-chatEvaluation Protocol=Few-shot2026.04 | 49.9 | 39.26 | |
| GPT-4oEvaluation Protocol=Zero-shot2026.04 | 47.41 | 37.32 | |
| GPT-5-mini-2025-08-07Evaluation Protocol=Zero-shot2026.04 | 47.32 | 38.75 | |
| GPT-4oEvaluation Protocol=Few-shot2026.04 | 45.97 | 23.09 |