File size: 3,208 Bytes
25a40d3 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 |
from typing import List, Optional, Union
import torch
import torch.nn as nn
from omegaconf import ListConfig
from taming.modules.losses.lpips import LPIPS
from ...util import append_dims, instantiate_from_config
class StandardDiffusionLoss(nn.Module):
def __init__(
self,
sigma_sampler_config,
type="l2",
offset_noise_level=0.0,
offset_noise_varying_dim = 1,
batch2model_keys: Optional[Union[str, List[str], ListConfig]] = None,
):
super().__init__()
assert type in ["l2", "l1", "lpips"]
self.sigma_sampler = instantiate_from_config(sigma_sampler_config)
self.type = type
self.offset_noise_level = offset_noise_level
self.offset_noise_varying_dim = offset_noise_varying_dim
if type == "lpips":
self.lpips = LPIPS().eval()
if not batch2model_keys:
batch2model_keys = []
if isinstance(batch2model_keys, str):
batch2model_keys = [batch2model_keys]
self.batch2model_keys = set(batch2model_keys)
def __call__(self, network, denoiser, conditioner, input, batch):
cond = conditioner(batch)
additional_model_inputs = {
key: batch[key] for key in self.batch2model_keys.intersection(batch)
}
sigmas = self.sigma_sampler(input.shape[0]).to(input.device)
noise = torch.randn_like(input)
if self.offset_noise_level > 0.0:
# noise = noise + self.offset_noise_level * append_dims(
# torch.randn(input.shape[0], device=input.device), input.ndim
# )
assert input.ndim > self.offset_noise_varying_dim, 'input.ndim should be larger than self.offset_noise_varying_dim'
noise = noise + self.offset_noise_level * append_dims(
torch.randn(input.shape[:self.offset_noise_varying_dim], device=input.device), input.ndim
)
noised_input = input + noise * append_dims(sigmas, input.ndim)
# noised_input: [1, 4, 9, 40, 40]
# cond['crossattn']: [1, 77, 1024]
# sigmas: the coefficient of the corresponding t.
# import torchvision, einops
# vis = einops.rearrange(input, '1 c t h w -> t c h w')[:,:3]
# torchvision.utils.save_image(vis, 'input.png', normalize=True)
# import pdb; pdb.set_trace()
# model_output = denoiser(
# network, noised_input, sigmas, cond, **additional_model_inputs
# )
model_output = denoiser(network, noised_input, sigmas, cond)
w = append_dims(denoiser.w(sigmas), input.ndim)
return self.get_loss(model_output, input, w)
def get_loss(self, model_output, target, w):
if self.type == "l2":
return torch.mean(
(w * (model_output - target) ** 2).reshape(target.shape[0], -1), 1
)
elif self.type == "l1":
return torch.mean(
(w * (model_output - target).abs()).reshape(target.shape[0], -1), 1
)
elif self.type == "lpips":
loss = self.lpips(model_output, target).reshape(-1)
return loss |