Implement sampling

[paulnorman1]

Hello! Thank you for this repo!

Can you please implement sampling in the example file, so we'd be able to generate outputs from the model (as done in GPT-2)?

[lucidrains]

@paulnorman1 hello, the code in the example folder does just that! (but i haven't added the sampling) also to be done is proper caching during evaluation to make things fast

[paulnorman1]

Excuse me, I'm not proficient in ML (I mainly test public code on my datasets). I don't see any code for generating output from the model in train.py. Is it something that you're planning on implementing?

[lucidrains]

yup, i will implement it at some point, most possibly with nucleus sampling

[paulnorman1]

Great, thanks!

[lucidrains]

@paulnorman1 I've updated the example to generate text!

[paulnorman1]

Awesome, thank you so much! Can't wait to test.