rct_model / train_model.py

Commit History

Use vae for encoding and decoding for training
9bde8da

frutiemax commited on

Use float16 for inference and float32 for training
42f8b67

frutiemax commited on

Use SGD and text encoder/tokenizer
6fa0b52

frutiemax commited on

Fix other things...
a150f0f

frutiemax commited on

Switch to stabilityai/sd-vae-ft-mse
2104644

frutiemax commited on

Use VAE to speed up inference
054faf7

frutiemax commited on

Revert to not using accelerated
21640c8

frutiemax commited on

Push unet and scheduler outside of pipeline
a4c8091

frutiemax commited on

Use accelerate
f6f5f48

frutiemax commited on

Fix batch size
1618027

frutiemax commited on

Fix training
04d70cd

frutiemax commited on

Use ConditionalUnetModel
88deab4

frutiemax commited on

Training code
d751051

frutiemax commited on