Add evaluation results on the offensive config and train split of tweet_eval

Update README.md

Update README.md

First model version

Update README.md

initial commit