Science Question Answering on ScienceQA (Accuracy and Throughput)
95.19AccuracyUDS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| UDSBackbone=Qwen-2.5-7B, Zero-shot Evaluation=true2025.10 | 95.19 | 7.85 | |
| RegularBackbone=Qwen-2.5-7B, Zero-shot Evaluation=true2025.10 | 94.56 | 7.05 | |
| UDSBackbone=Llama-3.1-8B, Zero-shot Evaluation=true2025.10 | 94.33 | 7.46 | |
| GREATSBackbone=Qwen-2.5-7B, Zero-shot Evaluation=true2025.10 | 94.17 | 6.53 | |
| MaxGradBackbone=Qwen-2.5-7B, Zero-shot Evaluation=true2025.10 | 93.86 | 0.98 | |
| RHO-LossBackbone=Qwen-2.5-7B, Zero-shot Evaluation=true2025.10 | 93.8 | 3.89 | |
| GREATSBackbone=Llama-3.1-8B, Zero-shot Evaluation=true2025.10 | 93.68 | 6.04 | |
| RHO-LossBackbone=Llama-3.1-8B, Zero-shot Evaluation=true2025.10 | 93.42 | 3.83 | |
| RandomBackbone=Qwen-2.5-7B, Zero-shot Evaluation=true2025.10 | 93.28 | 10.27 | |
| RegularBackbone=Llama-3.1-8B, Zero-shot Evaluation=true2025.10 | 93.19 | 6.61 | |
| MaxLossBackbone=Qwen-2.5-7B, Zero-shot Evaluation=true2025.10 | 93.05 | 7.93 | |
| RandomBackbone=Llama-3.1-8B, Zero-shot Evaluation=true2025.10 | 92.86 | 10.06 | |
| MaxLossBackbone=Llama-3.1-8B, Zero-shot Evaluation=true2025.10 | 92.82 | 7.49 | |
| MaxGradBackbone=Llama-3.1-8B, Zero-shot Evaluation=true2025.10 | 92.77 | 0.94 |