Open Masterchenyong opened 1 year ago
Describe Model I am using (UniLM, MiniLM, LayoutLM ...):
DIT and BEIT encoders are the same, except the image tokenizer is fine-tuned for the specified dataset. I have tried it but the model is not converging. you can check the issue #1268
https://github.com/microsoft/unilm/issues/1268
Describe Model I am using (UniLM, MiniLM, LayoutLM ...):