diffusers

History

Anton Lozhkov d9316bf8bc Fix mutable proj_out weight in the Attention layer (#73 ) * Catch unused params in DDP * Fix proj_out, add test		2022-07-04 12:36:37 +02:00
..
experimental	Add the model card template (#43 )	2022-06-29 15:37:23 +02:00
README.md	Update README.md	2022-06-22 13:52:45 +02:00
train_unconditional.py	Fix mutable proj_out weight in the Attention layer (#73 )	2022-07-04 12:36:37 +02:00

README.md

Training examples

Unconditional Flowers

The command to train a DDPM UNet model on the Oxford Flowers dataset:

python -m torch.distributed.launch \
  --nproc_per_node 4 \
  train_unconditional.py \
  --dataset="huggan/flowers-102-categories" \
  --resolution=64 \
  --output_dir="flowers-ddpm" \
  --batch_size=16 \
  --num_epochs=100 \
  --gradient_accumulation_steps=1 \
  --lr=1e-4 \
  --warmup_steps=500 \
  --mixed_precision=no

A full training run takes 2 hours on 4xV100 GPUs.

Unconditional Pokemon

The command to train a DDPM UNet model on the Pokemon dataset:

python -m torch.distributed.launch \
  --nproc_per_node 4 \
  train_unconditional.py \
  --dataset="huggan/pokemon" \
  --resolution=64 \
  --output_dir="pokemon-ddpm" \
  --batch_size=16 \
  --num_epochs=100 \
  --gradient_accumulation_steps=1 \
  --lr=1e-4 \
  --warmup_steps=500 \
  --mixed_precision=no

A full training run takes 2 hours on 4xV100 GPUs.