Topic Classification on TREC (train)
80.87AccuracyCanDistadd
Evaluation results
| Method | Links | |
|---|---|---|
| CanDistaddLLM Annotator=GPT-3.5, SLM Backbone=RoBERTa-Base, Candidate Strategy=CAadd2025.06 | 80.87 | |
| CanDistallLLM Annotator=GPT-3.5, SLM Backbone=RoBERTa-Base, Candidate Strategy=CAall2025.06 | 79.73 | |
| FreeALLLM Annotator=GPT-3.5, SLM Backbone=RoBERTa-Base2025.06 | 78.24 | |
| SuperICLLLM Annotator=GPT-3.5, SLM Backbone=RoBERTa-Base2025.06 | 76.05 | |
| DistillationLLM Annotator=GPT-3.5, SLM Backbone=RoBERTa-Base2025.06 | 76.04 | |
| AnnoLLMLLM Annotator=GPT-3.52025.06 | 73.73 | |
| CoTLLM Annotator=GPT-3.52025.06 | 71.88 | |
| Few-shotLLM Annotator=GPT-3.52025.06 | 71.07 | |
| SCLLM Annotator=GPT-3.5, Inference Strategy=Self-consistency2025.06 | 71.06 | |
| Zero-shotLLM Annotator=GPT-3.52025.06 | 62.84 |