Sentiment Analysis on SST2 (Accuracy and Standard Deviation)
89.9AccuracyROAST
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ROASTModel=Qwen3-8B, Training Size=100, Aggregation=group2026.02 | 89.9 | 0 | |
| ROASTModel=Qwen3-8B, Training Size=1000, Aggregation=group2026.02 | 88.47 | 0.01 | |
| Few-shotModel=Qwen3-8B, Training Size=1002026.02 | 86.81 | 0.11 | |
| BaselineModel=Qwen3-8B2026.02 | 86.68 | 0.21 | |
| ROASTModel=Qwen3-0.6B, Training Size=100, Aggregation=group2026.02 | 85.24 | 0.13 | |
| ROASTModel=Qwen3-0.6B, Training Size=100, Aggregation=no-group2026.02 | 85.03 | 0.32 | |
| ROASTModel=Qwen3-0.6B, Training Size=1000, Aggregation=no-group2026.02 | 84.81 | 0.21 | |
| ROASTModel=Qwen3-0.6B, Training Size=1000, Aggregation=group2026.02 | 84.6 | 0.38 | |
| ROASTModel=Gemma3-1B, Training Size=1000, Aggregation=group2026.02 | 84.31 | 5.58 | |
| Few-shotModel=Gemma3-1B, Training Size=1002026.02 | 84.05 | 0 | |
| Few-shotModel=Qwen3-0.6B, Training Size=1002026.02 | 84.03 | 0.38 | |
| SADIModel=Qwen3-0.6B, Training Size=10002026.02 | 82.81 | 0.03 | |
| BaselineModel=Gemma3-1B2026.02 | 81.08 | 0.11 | |
| SADIModel=Qwen3-0.6B, Training Size=1002026.02 | 80.52 | 0.11 | |
| CAAModel=Qwen3-0.6B, Training Size=1002026.02 | 79.08 | 1.1 | |
| BaselineModel=Qwen3-0.6B2026.02 | 78.8 | 0.29 | |
| CAAModel=Qwen3-0.6B, Training Size=10002026.02 | 78.55 | 0.25 |