4 days ago
-
Add LayoutLM and LayoutLMv3 trained on the original labels
8 days ago
-
Add linear SVMs on ImageNet ResNet-50 features (image-only, fusion, fusion + handwriting tokens)
-
Add SigLIP SVMs (image-only, fusion, fusion + handwriting tokens) and original-label BERT / RoBERTa
-
Add LayoutLMv3 (corrected labels, codes kept / removed, seed 42, 10 epochs)
12 days ago
-
Add LayoutLM v1, BERT and RoBERTa (corrected labels, seed 42)
-
Add Tesseract-text TF-IDF SVM and TF-IDF + CLIP fusion SVMs
13 days ago
-
Card: American spelling
16 days ago
-
Set the licence to MIT
-
Add linear SVM baselines (TF-IDF, TF-IDF + handwriting tokens, CLIP embeddings)
-
Remove the citation block for now
-
RVL-CDIP classifiers: seven architectures, four training configurations
-
initial commit
