thenativefox
Added split files and tables
939262b
raw
history blame contribute delete
471 Bytes
accelerate config
Then launch your training with:
accelerate launch train.py
Train with a notebook
πŸ€— Accelerate can also run in a notebook if you're planning on using Colaboratory's TPUs. Wrap all the code responsible for training in a function, and pass it to [~accelerate.notebook_launcher]:
from accelerate import notebook_launcher
notebook_launcher(training_function)
For more information about πŸ€— Accelerate and its rich features, refer to the documentation.