Few-Shot Text Classification on GLUE SuperGLUE SNLI subsets (test)
93SST-2 AccuracyGFLOWPO
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GFLOWPOMethod Category=Discrete Prompt Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 93 | 69.6 | 82 | 80.2 | 68.7 | 78.6 | 78.7 | |
| StablePromptMethod Category=Discrete Prompt Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 92.5 | 71.3 | 81.5 | 75.9 | 63.3 | 74.1 | 76.4 | |
| Manual PromptMethod Category=Fixed Prompt, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 89.1 | 51 | 64 | 73 | 67 | 47 | 65.2 | |
| GrIPSMethod Category=Discrete Prompt Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 84.7 | 55.6 | 60.9 | 28.9 | 44.4 | 63.5 | 59.4 | |
| APEMethod Category=Discrete Prompt Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 83.2 | 55.3 | 78.6 | 75 | 54.6 | 72.3 | 70.1 | |
| Soft Prompt TuningMethod Category=Fine-Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 78.3 | 57.1 | 51.6 | 89 | 34.9 | 55.8 | 61.1 | |
| Fine-TuningMethod Category=Fine-Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B, Fine-tuning Type=LoRA-based2026.02 | 71.9 | 59.6 | 55.7 | 63.1 | 41.1 | 64.8 | 59.3 | |
| RLPromptMethod Category=Discrete Prompt Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 70.8 | 56 | 67.3 | 62.6 | 54.6 | 56.6 | 61.3 | |
| ProTeGiMethod Category=Discrete Prompt Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 69.2 | 48.8 | 73.2 | 74.2 | 56.6 | 61.3 | 64 | |
| PromptBoostingMethod Category=Discrete Prompt Tuning, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 65.4 | 52.7 | 71.6 | 71.6 | 35.5 | 52.6 | 58.2 | |
| Zero-Shot CoTMethod Category=Fixed Prompt, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 57.9 | 38.4 | 81.6 | 75.2 | 71.1 | 66.3 | 65.1 | |
| Few-shot PromptMethod Category=Fixed Prompt, Prompt-LM=Gemma-7B, Target-LM=Gemma-7B2026.02 | 55 | 49 | 76 | 82 | 58 | 52.2 | 62 |