Text Classification Suite (SST2, FPB, COLA, DBpedia, EmoC, EmoS, ETHOS-SO, ETHOS-R)
88.8SST2 AccuracyDemonstration Selection Algorithm
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Demonstration Selection AlgorithmLLM=GPT3-c2023.01 | 88.8 | 64.1 | 69 | 73.6 | 50.3 | 43.1 | 86.2 | 78.2 | 69.2 | |
| Demonstration Selection AlgorithmLLM=GPT-J2023.01 | 87.8 | 56.7 | 69.1 | 60 | 32.5 | 33.2 | 85.3 | 77 | 62.7 | |
| Demonstration Selection AlgorithmLLM=GPT2-l2023.01 | 86.2 | 60.4 | 69.1 | 56.5 | 48.4 | 38.6 | 82.5 | 76.6 | 64.8 | |
| RandomLLM=GPT3-c2023.01 | 84.2 | 52.6 | 59.1 | 70.6 | 44.3 | 32.3 | 77.5 | 77.5 | 62.3 | |
| RandomLLM=GPT-J2023.01 | 78.5 | 53.1 | 58.3 | 55.6 | 38.5 | 33.3 | 76.6 | 76.6 | 58.8 | |
| RandomLLM=GPT2-l2023.01 | 77.1 | 51.3 | 62.7 | 54.4 | 38.7 | 34.5 | 67.6 | 72.9 | 57.4 | |
| Demonstration Selection AlgorithmLLM=OPT2023.01 | 74.2 | 34.1 | 35.7 | 28.8 | 76.7 | 39 | 86.2 | 78.2 | 56.6 | |
| RandomLLM=OPT2023.01 | 72.4 | 32.8 | 34.8 | 29.4 | 67.1 | 36.9 | 86.2 | 78.2 | 54.7 | |
| Demonstration Selection AlgorithmLLM=LLAMA2023.01 | 60.5 | 19.1 | 30.8 | 16.9 | 4.3 | 35.3 | 77.2 | 56.3 | 37.6 | |
| RandomLLM=LLAMA2023.01 | 57.7 | 23.7 | 30.8 | 15.8 | 4.4 | 35.2 | 66.2 | 57.2 | 36.4 |