stable-diffusion-finetune/configs/stable-diffusion/txt2img-upscale-clip-encoder-f16-1024-dev.yaml
2022-06-13 00:39:48 +02:00

175 lines
4.8 KiB
YAML

model:
base_learning_rate: 1.0e-04
target: ldm.models.diffusion.ddpm.LatentUpscaleDiffusion
params:
low_scale_key: "LR_image" # TODO: adapt
linear_start: 0.001
linear_end: 0.015
num_timesteps_cond: 1
log_every_t: 200
timesteps: 1000
first_stage_key: "image"
#first_stage_key: "jpg" # TODO: use this later
cond_stage_key: "caption"
#cond_stage_key: "txt" # TODO: use this later
image_size: 64
channels: 16
cond_stage_trainable: false
conditioning_key: "hybrid-adm"
monitor: val/loss_simple_ema
scale_factor: 0.22765929 # magic number
low_scale_config:
target: ldm.modules.encoders.modules.LowScaleEncoder
params:
linear_start: 0.00085
linear_end: 0.0120
timesteps: 1000
max_noise_level: 250
output_size: 64
model_config:
target: ldm.models.autoencoder.AutoencoderKL
params:
embed_dim: 4
monitor: val/rec_loss
ckpt_path: "models/first_stage_models/kl-f8/model.ckpt"
ddconfig:
double_z: true
z_channels: 4
resolution: 256
in_channels: 3
out_ch: 3
ch: 128
ch_mult:
- 1
- 2
- 4
- 4
num_res_blocks: 2
attn_resolutions: [ ]
dropout: 0.0
lossconfig:
target: torch.nn.Identity
scheduler_config: # 10000 warmup steps
target: ldm.lr_scheduler.LambdaLinearScheduler
params:
warm_up_steps: [ 10000 ]
cycle_lengths: [ 10000000000000 ] # incredibly large number to prevent corner cases
f_start: [ 1.e-6 ]
f_max: [ 1. ]
f_min: [ 1. ]
unet_config:
target: ldm.modules.diffusionmodules.openaimodel.UNetModel
params:
num_classes: 1000 # timesteps for noise conditoining
image_size: 64 # not really needed
in_channels: 20
out_channels: 16
model_channels: 32 # TODO: more
attention_resolutions: [ 4, 2, 1 ]
num_res_blocks: 2
channel_mult: [ 1, 2, 4, 4 ]
num_heads: 8
use_spatial_transformer: True
transformer_depth: 1
context_dim: 768
use_checkpoint: True
legacy: False
first_stage_config:
target: ldm.models.autoencoder.AutoencoderKL
params:
embed_dim: 16
monitor: val/rec_loss
ckpt_path: "models/first_stage_models/kl-f16/model.ckpt"
ddconfig:
double_z: True
z_channels: 16
resolution: 256
in_channels: 3
out_ch: 3
ch: 128
ch_mult: [ 1,1,2,2,4 ] # num_down = len(ch_mult)-1
num_res_blocks: 2
attn_resolutions: [ 16 ]
dropout: 0.0
lossconfig:
target: torch.nn.Identity
cond_stage_config:
target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
#data:
# target: ldm.data.laion.WebDataModuleFromConfig
# params:
# tar_base: "pipe:aws s3 cp s3://s-datasets/laion5b/laion2B-data/"
# batch_size: 4
# num_workers: 4
# multinode: True
# min_size: 256 # TODO: experiment. Note: for 2B, images are stored at max 384 resolution
# train:
# shards: '{000000..231317}.tar -'
# shuffle: 10000
# image_key: jpg
# image_transforms:
# - target: torchvision.transforms.Resize
# params:
# size: 1024
# interpolation: 3
# - target: torchvision.transforms.RandomCrop
# params:
# size: 1024
#
# # NOTE use enough shards to avoid empty validation loops in workers
# validation:
# shards: '{231318..231349}.tar -'
# shuffle: 0
# image_key: jpg
# image_transforms:
# - target: torchvision.transforms.Resize
# params:
# size: 1024
# interpolation: 3
# - target: torchvision.transforms.CenterCrop
# params:
# size: 1024
data:
target: main.DataModuleFromConfig
params:
batch_size: 8
num_workers: 7
wrap: false
train:
target: ldm.data.imagenet.ImageNetSRTrain
params:
size: 1024
downscale_f: 4
degradation: "cv_nearest"
lightning:
callbacks:
image_logger:
target: main.ImageLogger
params:
batch_frequency: 10
max_images: 4
increase_log_steps: False
log_first_step: False
log_images_kwargs:
use_ema_scope: False
inpaint: False
plot_progressive_rows: False
plot_diffusion_rows: False
N: 4
unconditional_guidance_scale: 3.0
unconditional_guidance_label: [""]
trainer:
benchmark: True
# val_check_interval: 5000000 # really sorry # TODO: bring back in
num_sanity_val_steps: 0
accumulate_grad_batches: 1