List your company
ResearchFormerly WizWand

Prompt Robustness Evaluation on CRC-100K

0.105Parameter VarianceGAUC
0.09680.152150.20750.26285May 18, 2026
Updated 5mo ago

Evaluation results

MethodLinks
GAUCModel=Qwen, Shot=3-shot2026.050.1050.095
GAUCModel=Qwen, Shot=1-shot2026.050.1270.103
MIMICModel=Qwen, Shot=3-shot2026.050.1630.107
MIMICModel=Qwen, Shot=1-shot2026.050.1750.145
DRModel=Qwen, Shot=3-shot2026.050.190.12
GAUCModel=LLaVA, Shot=3-shot2026.050.190.13
kNNModel=Qwen, Shot=3-shot2026.050.1970.125
MIMICModel=LLaVA, Shot=3-shot2026.050.1970.147
DRModel=LLaVA, Shot=3-shot2026.050.20.19
kNNModel=Qwen, Shot=1-shot2026.050.2070.183
GAUCModel=LLaVA, Shot=1-shot2026.050.2070.163
DRModel=Qwen, Shot=1-shot2026.050.210.19
kNNModel=LLaVA, Shot=3-shot2026.050.2130.175
DRModel=LLaVA, Shot=1-shot2026.050.2150.223
MIMICModel=LLaVA, Shot=1-shot2026.050.220.187
kNNModel=LLaVA, Shot=1-shot2026.050.2330.23
RandomModel=Qwen, Shot=3-shot2026.050.2650.187
RandomModel=Qwen, Shot=1-shot2026.050.2830.227
RandomModel=LLaVA, Shot=3-shot2026.050.2970.213
RandomModel=LLaVA, Shot=1-shot2026.050.310.255