Binary Classification on LIAR Open
84.21Accuracyw/ Intervention
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| w/ InterventionLLM=Llama2 (13B)2024.02 | 84.21 | 84.03 | |
| TELLERLLM=Llama2 (13B)2024.02 | 82.52 | 82.37 | |
| w/ InterventionLLM=Llama2 (7B)2024.02 | 81.93 | 81.84 | |
| Few-shotLLM=GPT-3.5-turbo2024.02 | 81.02 | 81 | |
| w/ InterventionLLM=FLAN-T5-xxl (11B)2024.02 | 80.73 | 80.41 | |
| TELLERLLM=FLAN-T5-xxl (11B)2024.02 | 80.24 | 79.85 | |
| TELLERLLM=Llama2 (7B)2024.02 | 79.94 | 79.8 | |
| w/ InterventionLLM=FLAN-T5-xl (3B)2024.02 | 79.34 | 79.07 | |
| TELLERLLM=FLAN-T5-xl (3B)2024.02 | 78.75 | 78.55 | |
| TELLERLLM=FLAN-T5-large (780M)2024.02 | 77.76 | 77.32 | |
| w/ InterventionLLM=FLAN-T5-large (780M)2024.02 | 77.46 | 77.14 | |
| DirectLLM=GPT-3.5-turbo2024.02 | 76.27 | 74.21 | |
| DirectLLM=FLAN-T5-xl (3B)2024.02 | 75.97 | 75.67 | |
| DirectLLM=FLAN-T5-xxl (11B)2024.02 | 75.17 | 75.15 | |
| Few-shot COTLLM=GPT-3.5-turbo2024.02 | 74.48 | 76.21 | |
| DirectLLM=FLAN-T5-large (780M)2024.02 | 74.38 | 73.84 | |
| Zero-shot COTLLM=GPT-3.5-turbo2024.02 | 72.49 | 70.83 | |
| DirectLLM=Llama2 (7B)2024.02 | 72.29 | 69.63 | |
| DirectLLM=Llama2 (13B)2024.02 | 69.31 | 63.77 | |
| Few-shot LogicLLM=GPT-3.5-turbo2024.02 | 61.67 | 60.92 | |
| DirectLLM=FLAN-T5-base (250M)2024.02 | 61.47 | 61.43 | |
| DirectLLM=FLAN-T5-small (80M)2024.02 | 45.08 | 32.41 |