minimaxir / textgenrnn

Easily train your own text-generating neural network of any size and complexity on any text dataset with a few lines of code.
Other
4.95k stars 751 forks source link

Slow generation of text even after fixing #231 #234

Open peterfarted opened 3 years ago

peterfarted commented 3 years ago

After the fix mentioned here: https://github.com/minimaxir/textgenrnn/issues/231

the model training is back to normal speed, but the actual generation of modeled text appears to have slowed down tremendously, both in the 0.2-0.5-1.0 samples generated after each "gen_epochs," and in the export step. What can I check to troubleshoot/fix this?