Performance issues in the program

[DLPerf]

Hello,I found a performance issue in the definition of lambada_input ,
EleutherAI/gpt-neo/blob/master/tasks.py,
dataset.map was called without num_parallel_calls.
I think it will increase the efficiency of your program if you add this.

Here is the documemtation of tensorflow to support this thing.

Looking forward to your reply. Btw, I am very glad to create a PR to fix it if you are too busy.

[StellaAthena]

Please do open a PR, ideally with plots showing the performance gain

[DLPerf]

Ok,I am willing to offer you a PR.But do you have some benchmark tools?It'd be more convient if you provide me your benchmark tools.