Fake news detection on Constraint
87.78AccuracyTELLER w/ Intervention
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TELLER w/ InterventionLLM=Llama2 (13B)2024.02 | 87.78 | 87.71 | |
| TELLERLLM=Llama2 (13B)2024.02 | 87.31 | 87.29 | |
| TELLER w/ InterventionLLM=Llama2 (7B)2024.02 | 85.13 | 85.04 | |
| TELLER w/ InterventionLLM=FLAN-T5-xl (3B)2024.02 | 83.95 | 83.88 | |
| TELLERLLM=FLAN-T5-xl (3B)2024.02 | 83.77 | 83.66 | |
| TELLERLLM=Llama2 (7B)2024.02 | 83.72 | 83.54 | |
| TELLER w/ InterventionLLM=FLAN-T5-xxl (11B)2024.02 | 83.62 | 83.54 | |
| TELLERLLM=FLAN-T5-xxl (11B)2024.02 | 83.39 | 83.24 | |
| DirectLLM=Llama2 (7B)2024.02 | 81.83 | 81.73 | |
| TELLER w/ InterventionLLM=FLAN-T5-large (780M)2024.02 | 80.46 | 80.31 | |
| TELLERLLM=FLAN-T5-large (780M)2024.02 | 80.32 | 80.11 | |
| DirectLLM=FLAN-T5-large (780M)2024.02 | 78.06 | 77.97 | |
| DirectLLM=FLAN-T5-xl (3B)2024.02 | 75.32 | 74.79 | |
| DirectLLM=FLAN-T5-xxl (11B)2024.02 | 74.8 | 73.23 | |
| DirectLLM=Llama2 (13B)2024.02 | 57.53 | 51.75 |