Commit d2ac95fa authored by AUTOMATIC's avatar AUTOMATIC

remove the need to place configs near models

parent 7a14c8ab
# File modified by authors of InstructPix2Pix from original (https://github.com/CompVis/stable-diffusion).
# See more details in LICENSE.
model:
base_learning_rate: 1.0e-04
target: modules.models.diffusion.ddpm_edit.LatentDiffusion
params:
linear_start: 0.00085
linear_end: 0.0120
num_timesteps_cond: 1
log_every_t: 200
timesteps: 1000
first_stage_key: edited
cond_stage_key: edit
# image_size: 64
# image_size: 32
image_size: 16
channels: 4
cond_stage_trainable: false # Note: different from the one we trained before
conditioning_key: hybrid
monitor: val/loss_simple_ema
scale_factor: 0.18215
use_ema: true
load_ema: true
scheduler_config: # 10000 warmup steps
target: ldm.lr_scheduler.LambdaLinearScheduler
params:
warm_up_steps: [ 0 ]
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:
image_size: 32 # unused
in_channels: 8
out_channels: 4
model_channels: 320
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: 4
monitor: val/rec_loss
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
cond_stage_config:
target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
data:
target: main.DataModuleFromConfig
params:
batch_size: 128
num_workers: 1
wrap: false
validation:
target: edit_dataset.EditDataset
params:
path: data/clip-filtered-dataset
cache_dir: data/
cache_name: data_10k
split: val
min_text_sim: 0.2
min_image_sim: 0.75
min_direction_sim: 0.2
max_samples_per_prompt: 1
min_resize_res: 512
max_resize_res: 512
crop_res: 512
output_as_edit: False
real_input: True
model:
base_learning_rate: 1.0e-4
target: ldm.models.diffusion.ddpm.LatentDiffusion
base_learning_rate: 7.5e-05
target: ldm.models.diffusion.ddpm.LatentInpaintDiffusion
params:
parameterization: "v"
linear_start: 0.00085
linear_end: 0.0120
num_timesteps_cond: 1
......@@ -12,29 +11,36 @@ model:
cond_stage_key: "txt"
image_size: 64
channels: 4
cond_stage_trainable: false
conditioning_key: crossattn
cond_stage_trainable: false # Note: different from the one we trained before
conditioning_key: hybrid # important
monitor: val/loss_simple_ema
scale_factor: 0.18215
use_ema: False # we set this to false because this is an inference only config
finetune_keys: null
scheduler_config: # 10000 warmup steps
target: ldm.lr_scheduler.LambdaLinearScheduler
params:
warm_up_steps: [ 2500 ] # NOTE for resuming. use 10000 if starting from scratch
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:
use_checkpoint: True
use_fp16: True
image_size: 32 # unused
in_channels: 4
in_channels: 9 # 4 data + 4 downscaled image + 1 mask
out_channels: 4
model_channels: 320
attention_resolutions: [ 4, 2, 1 ]
num_res_blocks: 2
channel_mult: [ 1, 2, 4, 4 ]
num_head_channels: 64 # need to fix for flash-attn
num_heads: 8
use_spatial_transformer: True
use_linear_in_transformer: True
transformer_depth: 1
context_dim: 1024
context_dim: 768
use_checkpoint: True
legacy: False
first_stage_config:
......@@ -43,7 +49,6 @@ model:
embed_dim: 4
monitor: val/rec_loss
ddconfig:
#attn_type: "vanilla-xformers"
double_z: true
z_channels: 4
resolution: 256
......@@ -62,7 +67,4 @@ model:
target: torch.nn.Identity
cond_stage_config:
target: ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
params:
freeze: True
layer: "penultimate"
\ No newline at end of file
target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
......@@ -18,7 +18,8 @@ from modules.textual_inversion.textual_inversion import create_embedding, train_
from modules.textual_inversion.preprocess import preprocess
from modules.hypernetworks.hypernetwork import create_hypernetwork, train_hypernetwork
from PIL import PngImagePlugin,Image
from modules.sd_models import checkpoints_list, find_checkpoint_config
from modules.sd_models import checkpoints_list
from modules.sd_models_config import find_checkpoint_config_near_filename
from modules.realesrgan_model import get_realesrgan_models
from modules import devices
from typing import List
......@@ -387,7 +388,7 @@ class Api:
]
def get_sd_models(self):
return [{"title": x.title, "model_name": x.model_name, "hash": x.shorthash, "sha256": x.sha256, "filename": x.filename, "config": find_checkpoint_config(x)} for x in checkpoints_list.values()]
return [{"title": x.title, "model_name": x.model_name, "hash": x.shorthash, "sha256": x.sha256, "filename": x.filename, "config": find_checkpoint_config_near_filename(x)} for x in checkpoints_list.values()]
def get_hypernetworks(self):
return [{"name": name, "path": shared.hypernetworks[name]} for name in shared.hypernetworks]
......
......@@ -34,14 +34,18 @@ def get_cuda_device_string():
return "cuda"
def get_optimal_device():
def get_optimal_device_name():
if torch.cuda.is_available():
return torch.device(get_cuda_device_string())
return get_cuda_device_string()
if has_mps():
return torch.device("mps")
return "mps"
return "cpu"
return cpu
def get_optimal_device():
return torch.device(get_optimal_device_name())
def get_device_for(task):
......
......@@ -96,15 +96,6 @@ def p_sample_plms(self, x, c, t, index, repeat_noise=False, use_original_steps=F
return x_prev, pred_x0, e_t
def should_hijack_inpainting(checkpoint_info):
from modules import sd_models
ckpt_basename = os.path.basename(checkpoint_info.filename).lower()
cfg_basename = os.path.basename(sd_models.find_checkpoint_config(checkpoint_info)).lower()
return "inpainting" in ckpt_basename and not "inpainting" in cfg_basename
def do_inpainting_hijack():
# p_sample_plms is needed because PLMS can't work with dicts as conditionings
......
This diff is collapsed.
import re
import os
from modules import shared, paths
sd_configs_path = shared.sd_configs_path
sd_repo_configs_path = os.path.join(paths.paths['Stable Diffusion'], "configs", "stable-diffusion")
config_default = shared.sd_default_config
config_sd2 = os.path.join(sd_repo_configs_path, "v2-inference.yaml")
config_sd2v = os.path.join(sd_repo_configs_path, "v2-inference-v.yaml")
config_inpainting = os.path.join(sd_configs_path, "v1-inpainting-inference.yaml")
config_instruct_pix2pix = os.path.join(sd_configs_path, "instruct-pix2pix.yaml")
config_alt_diffusion = os.path.join(sd_configs_path, "alt-diffusion-inference.yaml")
re_parametrization_v = re.compile(r'-v\b')
def guess_model_config_from_state_dict(sd, filename):
fn = os.path.basename(filename)
sd2_cond_proj_weight = sd.get('cond_stage_model.model.transformer.resblocks.0.attn.in_proj_weight', None)
diffusion_model_input = sd.get('model.diffusion_model.input_blocks.0.0.weight', None)
roberta_weight = sd.get('cond_stage_model.roberta.embeddings.word_embeddings.weight', None)
if sd2_cond_proj_weight is not None and sd2_cond_proj_weight.shape[1] == 1024:
if re.search(re_parametrization_v, fn) or "v2-1_768" in fn:
return config_sd2v
else:
return config_sd2
if diffusion_model_input is not None:
if diffusion_model_input.shape[1] == 9:
return config_inpainting
if diffusion_model_input.shape[1] == 8:
return config_instruct_pix2pix
if roberta_weight is not None:
return config_alt_diffusion
return config_default
def find_checkpoint_config(state_dict, info):
if info is None:
return guess_model_config_from_state_dict(state_dict, "")
config = find_checkpoint_config_near_filename(info)
if config is not None:
return config
return guess_model_config_from_state_dict(state_dict, info.filename)
def find_checkpoint_config_near_filename(info):
if info is None:
return None
config = os.path.splitext(info.filename)[0] + ".yaml"
if os.path.exists(config):
return config
return None
......@@ -13,13 +13,14 @@ import modules.interrogate
import modules.memmon
import modules.styles
import modules.devices as devices
from modules import localization, sd_vae, extensions, script_loading, errors, ui_components, shared_items
from modules import localization, extensions, script_loading, errors, ui_components, shared_items
from modules.paths import models_path, script_path
demo = None
sd_default_config = os.path.join(script_path, "configs/v1-inference.yaml")
sd_configs_path = os.path.join(script_path, "configs")
sd_default_config = os.path.join(sd_configs_path, "v1-inference.yaml")
sd_model_file = os.path.join(script_path, 'model.ckpt')
default_sd_model_file = sd_model_file
......@@ -391,7 +392,7 @@ options_templates.update(options_section(('sd', "Stable Diffusion"), {
"sd_model_checkpoint": OptionInfo(None, "Stable Diffusion checkpoint", gr.Dropdown, lambda: {"choices": list_checkpoint_tiles()}, refresh=refresh_checkpoints),
"sd_checkpoint_cache": OptionInfo(0, "Checkpoints to cache in RAM", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}),
"sd_vae_checkpoint_cache": OptionInfo(0, "VAE Checkpoints to cache in RAM", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}),
"sd_vae": OptionInfo("Automatic", "SD VAE", gr.Dropdown, lambda: {"choices": ["Automatic", "None"] + list(sd_vae.vae_dict)}, refresh=sd_vae.refresh_vae_list),
"sd_vae": OptionInfo("Automatic", "SD VAE", gr.Dropdown, lambda: {"choices": shared_items.sd_vae_items()}, refresh=shared_items.refresh_vae_list),
"sd_vae_as_default": OptionInfo(True, "Ignore selected VAE for stable diffusion checkpoints that have their own .vae.pt next to them"),
"inpainting_mask_weight": OptionInfo(1.0, "Inpainting conditioning mask strength", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}),
"initial_noise_multiplier": OptionInfo(1.0, "Noise multiplier for img2img", gr.Slider, {"minimum": 0.5, "maximum": 1.5, "step": 0.01}),
......
......@@ -4,7 +4,20 @@ def realesrgan_models_names():
import modules.realesrgan_model
return [x.name for x in modules.realesrgan_model.get_realesrgan_models(None)]
def postprocessing_scripts():
import modules.scripts
return modules.scripts.scripts_postproc.scripts
\ No newline at end of file
return modules.scripts.scripts_postproc.scripts
def sd_vae_items():
import modules.sd_vae
return ["Automatic", "None"] + list(modules.sd_vae.vae_dict)
def refresh_vae_list():
import modules.sd_vae
return modules.sd_vae.refresh_vae_list
import time
class Timer:
def __init__(self):
self.start = time.time()
self.records = {}
self.total = 0
def elapsed(self):
end = time.time()
res = end - self.start
self.start = end
return res
def record(self, category, extra_time=0):
e = self.elapsed()
if category not in self.records:
self.records[category] = 0
self.records[category] += e + extra_time
self.total += e + extra_time
def summary(self):
res = f"{self.total:.1f}s"
additions = [x for x in self.records.items() if x[1] >= 0.1]
if not additions:
return res
res += " ("
res += ", ".join([f"{category}: {time_taken:.1f}s" for category, time_taken in additions])
res += ")"
return res
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment