Classification Probing on Sarcasm (test)
96.3Probe Acc (Best Layer)xRFM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| xRFMBackbone=Llama-3.1-70B-Instruct2026.01 | 96.3 | 92.9 | |
| RAPTORBackbone=Llama-3.1-70B-Instruct2026.01 | 96.3 | 93 | |
| xRFMBackbone=Llama-3.3-70B-Instruct2026.01 | 95.4 | 91.8 | |
| RAPTORBackbone=Llama-3.3-70B-Instruct2026.01 | 95 | 92.2 | |
| xRFMBackbone=Qwen-32B-Instruct2026.01 | 94.7 | 89.5 | |
| RAPTORBackbone=Qwen-32B-Instruct2026.01 | 94.3 | 89.5 | |
| RAPTORBackbone=Llama-3.1-8B-Instruct2026.01 | 94.3 | 90.1 | |
| GCSBackbone=Llama-3.1-70B-Instruct2026.01 | 94.2 | 90.6 | |
| xRFMBackbone=Llama-3.1-8B-Instruct2026.01 | 94 | 90.2 | |
| GCSBackbone=Llama-3.3-70B-Instruct2026.01 | 93.6 | 89.8 | |
| RAPTORBackbone=Qwen-7B-Instruct2026.01 | 92.6 | 87.7 | |
| xRFMBackbone=Qwen-7B-Instruct2026.01 | 92.5 | 87.5 | |
| GCSBackbone=Qwen-32B-Instruct2026.01 | 92.3 | 87.1 | |
| xRFMBackbone=Qwen-3B-Instruct2026.01 | 91.9 | 85.9 | |
| GCSBackbone=Llama-3.1-8B-Instruct2026.01 | 91.7 | 87.6 | |
| RAPTORBackbone=Qwen-3B-Instruct2026.01 | 91.5 | 85.7 | |
| GCSBackbone=Qwen-7B-Instruct2026.01 | 90.2 | 85 | |
| RAPTORBackbone=Gemma-7B-it2026.01 | 90.2 | 85.2 | |
| xRFMBackbone=Gemma-7B-it2026.01 | 89.5 | 84.9 | |
| GCSBackbone=Qwen-3B-Instruct2026.01 | 89.4 | 82 | |
| GCSBackbone=Gemma-7B-it2026.01 | 88 | 82 |