forked from mrq/DL-Art-School
f12f0200d6
parametric efficiency improvements and lets try feeding the timestep into the conditioning encoder |
||
---|---|---|
.. | ||
audio | ||
classifiers | ||
clip | ||
composable | ||
diffusion | ||
image_generation | ||
image_latents | ||
lucidrains | ||
optical_flow | ||
vqvae | ||
__init__.py | ||
arch_util.py |