recreate tokenizer.json

delete tokenizer.json

fixes to tokenizer

fixing tokenizer to match model vocab size

Saving weights and logs of epoch 1

Add tokenizer file for oscar dataset - unshuffled_deduplicated_pl