PyTorch Frame: A Modular Framework for Multi-Modal Tabular Learning

Benchmark Model Rank Results
binary-classification-on-fakeTrompt + OpenAI embedding#1AUROC: 0.979
binary-classification-on-fakeLightGBM + OpenAI embedding#2AUROC: 0.966
binary-classification-on-fakeFTTransformer + RoBERTa fintune#3AUROC: 0.96
binary-classification-on-fakeLightGBM + RoBERTa embedding#4AUROC: 0.954
binary-classification-on-fakeFTTransformer + RoBERTa embedding#5AUROC: 0.936
binary-classification-on-fakeResNet + RoBERTa embedding#6AUROC: 0.934
binary-classification-on-fakeResNet + OpenAI embedding#7AUROC: 0.923
binary-classification-on-fakeFTTransformer + OpenAI embedding#8AUROC: 0.911
binary-classification-on-kickstarterTrompt + OpenAI embedding#1AUROC: 0.81
binary-classification-on-kickstarterResNet + RoBERTa finetune#3AUROC: 0.786
binary-classification-on-kickstarterLightGBM + RoBERTa embedding#4AUROC: 0.767
toxic-comment-classification-on-civilResNet + RoBERTa finetune#15AUROC: 0.97
toxic-comment-classification-on-civilTrompt + OpenAI embedding#16AUROC: 0.947
toxic-comment-classification-on-civilResNet + OpenAI embedding#17AUROC: 0.945
toxic-comment-classification-on-civilTrompt + RoBERTa embedding#18AUROC: 0.885
toxic-comment-classification-on-civilResNet + RoBERTa embedding#19AUROC: 0.882
toxic-comment-classification-on-civilLightGBM + RoBERTa embedding#20AUROC: 0.865