Sentiment Classification on REAL-AAE (held-out evaluation set)
84.9DGIGemma-3
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemma-3Parameters=27B2026.07 | 84.9 | 84.4 | 84.4 | |
| Phi-4Parameters=14B2026.07 | 81.4 | 78.8 | 82.6 | |
| Mistral-Small-3.1-24BParameters=24B2026.07 | 79.5 | 74.9 | 81.8 | |
| Phi-3-MediumParameters=14B2026.07 | 78 | 76.1 | 78.7 | |
| DeepSeek-R1Full Name=DeepSeek-R1-Distill-Qwen-32B, Parameters=32B2026.07 | 49.1 | 30.2 | 59.8 | |
| Llama-3.1-70BParameters=70B2026.07 | 46.8 | 27.8 | 53.1 |