Only embedding layer is frozen. 10 epochs. 0.00001 learning rate. 8 batch size. 512 max tokens. AllQuAD dataset.

5 first layers are frozen. 10 epochs. 0.0001 learning rate. 8 batch size. 512 max tokens. AllQuAD dataset. 1st run. 1st commit. 1st push. 1st push to hub. 1st push to hub with commit message. 1st push to hub with commit message and tokenizer. 1st push to hub with commit message and tokenizer and config. 1st push to hub with commit message and tokenizer and config and model. 1st push to hub with commit message and tokenizer and config and model and repo name. 1st push to hub with commit message and tokenizer and config and model and repo name and push to hub. 1st push to hub with commit message and tokenizer and config and model and repo name and push to hub with commit message. 1st push to hub with commit message and tokenizer and config and model and repo name and push to hub with commit message and push to hub with commit message. 1st push to hub with commit message and tokenizer and config and model and repo name and push to hub with commit message and push to hub with commit message and push to hub with commit message. 1st push to hub with commit message and tokenizer and config and model and repo name and push to hub with commit message and push to hub with commit message and push to hub with commit message and push to hub with commit message. 1st push to hub with commit message and tokenizer and config and model and repo name and push to hub with commit message and push to hub with commit message and push to hub with commit message and push to hub with commit message and push to hub with commit message. 1st push to hub with commit message and tokenizer and config and model and repo name

Best model on parsQuAD

Create README.md

Initial release

initial commit