Sentiment Classification on Amazon Reviews (Accuracy)
85.7AccuracyGPT-2 kNN-LM
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-2 kNN-LM# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 85.7 | |
| RoBERTa# Params=1.0x, Evaluation Protocol=Zero-shot2022.12 | 83.5 | |
| NPM# Params=1.0x, Evaluation Protocol=Zero-shot, nonparametric=true2022.12 | 83.4 | |
| GPT-2 + PMI# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 76.2 | |
| GPT-3 + PMI# Params=500x, Evaluation Protocol=Zero-shot2022.12 | 75 | |
| GPT-3# Params=500x, Evaluation Protocol=Zero-shot2022.12 | 59.4 | |
| T5# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 59.3 | |
| GPT-2# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 57.6 | |
| GPT-2 kNN# Params=2.2x, Evaluation Protocol=Zero-shot2022.12 | 57.4 | |
| T5 3B# Params=8.5x, Evaluation Protocol=Zero-shot2022.12 | 57 |