`use_wandb_tracking` was not stored in most Trainers when it is `False`

lucidrains / audiolm-pytorch

Implementation of AudioLM, a SOTA Language Modeling Approach to Audio Generation out of Google Research, in Pytorch

MIT License

2.32k stars 249 forks source link

`use_wandb_tracking` was not stored in most Trainers when it is `False` #258

Closed orrp closed 6 months ago

orrp commented 6 months ago

In the __init__s of SemanticTransformerTrainer, CoarseTransformerTrainer and FineTransformerTrainer, the attribute self.use_wandb_tracking was only set when the parameter use_wandb_tracking was True. This caused the Trainers to throw an AttributeError upon training when not using WandB.

This is now fixed.

lucidrains commented 6 months ago

@LukasNel @orrp oh nice, thank you!