Save checkpoint before terminating the training run

huggingface / nanotron

Minimalistic large language model 3D-parallelism training

Apache License 2.0

1.23k stars 122 forks source link

Closed xrsrke closed 10 months ago

xrsrke commented 10 months ago

Why don't we save a model checkpoint before terminating the training run? [link]