turboderp / exllamav2

A fast inference library for running LLMs locally on modern consumer-class GPUs
MIT License
3.19k stars 234 forks source link

[question] how to make generation determinsitic? #438

Closed yshui closed 2 weeks ago

yshui commented 1 month ago

I tried using random.seed() to set a fixed seed, but the output still changes.

DocShotgun commented 1 month ago

https://github.com/turboderp/exllamav2/issues/232#issuecomment-1860896496