Error with Tokenizer parallelism when using gradio and mlflow

I have written a script using gradio and sometimes (I emphasize this- only sometimes) when I run it I get