Topic Classification on Yahoo Answers Topics
66AccuracyOriginal
Evaluation results
| Method | Links | ||
|---|---|---|---|
| OriginalBackbone=BERT-large-uncased2026.02 | 66 | 66 | |
| OriginalBackbone=BERT-base-uncased2026.02 | 65 | 65 | |
| OriginalBackbone=BERT-base-cased2026.02 | 65 | 65 | |
| OriginalBackbone=BERT-large-cased2026.02 | 64 | 64 | |
| PruningBackbone=BERT-large-uncased2026.02 | 61 | 61 | |
| PruningBackbone=BERT-base-uncased2026.02 | 60 | 59 | |
| GhostBackbone=BERT-base-uncased2026.02 | 60 | 60 | |
| GhostBackbone=BERT-base-cased2026.02 | 60 | 60 | |
| PruningBackbone=BERT-base-cased2026.02 | 59 | 58 | |
| PruningBackbone=BERT-large-cased2026.02 | 59 | 59 | |
| GhostBackbone=BERT-large-cased2026.02 | 59 | 59 | |
| NoiseBackbone=BERT-base-uncased2026.02 | 55 | 55 | |
| GPT-3 + PMI# Params=500x, Evaluation Protocol=Zero-shot2022.12 | 54.7 | — | |
| NoiseBackbone=BERT-base-cased2026.02 | 54 | 53 | |
| NPM# Params=1.0x, Evaluation Protocol=Zero-shot, nonparametric=true2022.12 | 53.9 | — | |
| T5 3B# Params=8.5x, Evaluation Protocol=Zero-shot2022.12 | 53.6 | — | |
| GPT-3# Params=500x, Evaluation Protocol=Zero-shot2022.12 | 53.1 | — | |
| T5# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 51.3 | — | |
| GPT-2 kNN-LM# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 51 | — | |
| GPT-2# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 49.7 | — | |
| NoiseBackbone=BERT-large-cased2026.02 | 49 | 48 | |
| GPT-2 + PMI# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 48.8 | — | |
| GhostBackbone=BERT-large-uncased2026.02 | 47 | 48 | |
| RoBERTa# Params=1.0x, Evaluation Protocol=Zero-shot2022.12 | 41.4 | — | |
| GPT-2 kNN# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 37 | — | |
| NoiseBackbone=BERT-large-uncased2026.02 | 19 | 16 |