List your company

Multimodal Sarcasm Detection on redEval

84.8F1 ScoreProCrit
57.86464.85771.8578.843May 20, 2026

Evaluation results

MethodLinks
ProCritProtocol=Fine-tuned, Perspective=Adaptive, Generation Mode=Draft-only2026.0584.887.279.291.4
ProCritProtocol=Fine-tuned, Perspective=Adaptive2026.0583.785.775.793.7
Plain CoT ReasonerProtocol=Fine-tuned, Perspective=None2026.0582.684.874.992.2
Fixed-role ReasonerProtocol=Fine-tuned, Perspective=Predefined2026.0581.683.974.390.6
ProCritProtocol=Zero-shot, Perspective=Adaptive2026.0571.476.869.273.6
Zero-shot CoTProtocol=Zero-shot, Perspective=None2026.0570.678.275.166.6
Automatic Prompt EngineerProtocol=Zero-shot, Perspective=None2026.0570.278.275.965.3
Commander-GPTProtocol=Zero-shot, Perspective=Predefined2026.0570.172.360.982.5
Generated Knowledge PromptingProtocol=Zero-shot, Perspective=None2026.0561.575.480.149.9
S3AgentProtocol=Zero-shot, Perspective=Predefined2026.056173.873.652.2
IRONICProtocol=Zero-shot, Perspective=Predefined2026.0559.17689.744.1
Plan-and-SolveProtocol=Zero-shot, Perspective=None2026.0558.97277.447.6