aboutsummaryrefslogtreecommitdiffstats
path: root/modules
diff options
context:
space:
mode:
Diffstat (limited to 'modules')
-rw-r--r--modules/api/api.py17
-rw-r--r--modules/api/models.py5
-rw-r--r--modules/extras.py2
-rw-r--r--modules/generation_parameters_copypaste.py9
-rw-r--r--modules/hypernetworks/hypernetwork.py23
-rw-r--r--modules/processing.py158
-rw-r--r--modules/scripts.py10
-rw-r--r--modules/sd_hijack.py12
-rw-r--r--modules/sd_models.py8
-rw-r--r--modules/sd_samplers.py14
-rw-r--r--modules/shared.py13
-rw-r--r--modules/textual_inversion/learn_schedule.py11
-rw-r--r--modules/textual_inversion/textual_inversion.py33
-rw-r--r--modules/txt2img.py5
-rw-r--r--modules/ui.py125
15 files changed, 333 insertions, 112 deletions
diff --git a/modules/api/api.py b/modules/api/api.py
index 6267afdc..2103709b 100644
--- a/modules/api/api.py
+++ b/modules/api/api.py
@@ -11,10 +11,10 @@ from fastapi.security import HTTPBasic, HTTPBasicCredentials
from secrets import compare_digest
import modules.shared as shared
-from modules import sd_samplers, deepbooru, sd_hijack
+from modules import sd_samplers, deepbooru, sd_hijack, images
from modules.api.models import *
from modules.processing import StableDiffusionProcessingTxt2Img, StableDiffusionProcessingImg2Img, process_images
-from modules.extras import run_extras, run_pnginfo
+from modules.extras import run_extras
from modules.textual_inversion.textual_inversion import create_embedding, train_embedding
from modules.textual_inversion.preprocess import preprocess
from modules.hypernetworks.hypernetwork import create_hypernetwork, train_hypernetwork
@@ -101,6 +101,7 @@ class Api:
self.router = APIRouter()
self.app = app
self.queue_lock = queue_lock
+ api_middleware(self.app)
self.add_api_route("/sdapi/v1/txt2img", self.text2imgapi, methods=["POST"], response_model=TextToImageResponse)
self.add_api_route("/sdapi/v1/img2img", self.img2imgapi, methods=["POST"], response_model=ImageToImageResponse)
self.add_api_route("/sdapi/v1/extra-single-image", self.extras_single_image_api, methods=["POST"], response_model=ExtrasSingleImageResponse)
@@ -232,9 +233,17 @@ class Api:
if(not req.image.strip()):
return PNGInfoResponse(info="")
- result = run_pnginfo(decode_base64_to_image(req.image.strip()))
+ image = decode_base64_to_image(req.image.strip())
+ if image is None:
+ return PNGInfoResponse(info="")
+
+ geninfo, items = images.read_info_from_image(image)
+ if geninfo is None:
+ geninfo = ""
+
+ items = {**{'parameters': geninfo}, **items}
- return PNGInfoResponse(info=result[1])
+ return PNGInfoResponse(info=geninfo, items=items)
def progressapi(self, req: ProgressRequest = Depends()):
# copy from check_progress_call of ui.py
diff --git a/modules/api/models.py b/modules/api/models.py
index 4a632c68..d8198a27 100644
--- a/modules/api/models.py
+++ b/modules/api/models.py
@@ -157,7 +157,8 @@ class PNGInfoRequest(BaseModel):
image: str = Field(title="Image", description="The base64 encoded PNG image")
class PNGInfoResponse(BaseModel):
- info: str = Field(title="Image info", description="A string with all the info the image had")
+ info: str = Field(title="Image info", description="A string with the parameters used to generate the image")
+ items: dict = Field(title="Items", description="An object containing all the info the image had")
class ProgressRequest(BaseModel):
skip_current_image: bool = Field(default=False, title="Skip current image", description="Skip current image serialization")
@@ -258,4 +259,4 @@ class EmbeddingItem(BaseModel):
class EmbeddingsResponse(BaseModel):
loaded: Dict[str, EmbeddingItem] = Field(title="Loaded", description="Embeddings loaded for the current model")
- skipped: Dict[str, EmbeddingItem] = Field(title="Skipped", description="Embeddings skipped for the current model (likely due to architecture incompatibility)") \ No newline at end of file
+ skipped: Dict[str, EmbeddingItem] = Field(title="Skipped", description="Embeddings skipped for the current model (likely due to architecture incompatibility)")
diff --git a/modules/extras.py b/modules/extras.py
index d665440a..7407bfe3 100644
--- a/modules/extras.py
+++ b/modules/extras.py
@@ -19,8 +19,6 @@ from modules.shared import opts
import modules.gfpgan_model
from modules.ui import plaintext_to_html
import modules.codeformer_model
-import piexif
-import piexif.helper
import gradio as gr
import safetensors.torch
diff --git a/modules/generation_parameters_copypaste.py b/modules/generation_parameters_copypaste.py
index 4baf4d9a..12a9de3d 100644
--- a/modules/generation_parameters_copypaste.py
+++ b/modules/generation_parameters_copypaste.py
@@ -212,11 +212,10 @@ def restore_old_hires_fix_params(res):
firstpass_width = math.ceil(scale * width / 64) * 64
firstpass_height = math.ceil(scale * height / 64) * 64
- hr_scale = width / firstpass_width if firstpass_width > 0 else height / firstpass_height
-
res['Size-1'] = firstpass_width
res['Size-2'] = firstpass_height
- res['Hires upscale'] = hr_scale
+ res['Hires resize-1'] = width
+ res['Hires resize-2'] = height
def parse_generation_parameters(x: str):
@@ -276,6 +275,10 @@ Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 965400086, Size: 512x512, Model
hypernet_hash = res.get("Hypernet hash", None)
res["Hypernet"] = find_hypernetwork_key(hypernet_name, hypernet_hash)
+ if "Hires resize-1" not in res:
+ res["Hires resize-1"] = 0
+ res["Hires resize-2"] = 0
+
restore_old_hires_fix_params(res)
return res
diff --git a/modules/hypernetworks/hypernetwork.py b/modules/hypernetworks/hypernetwork.py
index 450fecac..6a9b1398 100644
--- a/modules/hypernetworks/hypernetwork.py
+++ b/modules/hypernetworks/hypernetwork.py
@@ -402,10 +402,8 @@ def create_hypernetwork(name, enable_sizes, overwrite_old, layer_structure=None,
shared.reload_hypernetworks()
- return fn
-
-def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, gradient_step, data_root, log_directory, training_width, training_height, steps, shuffle_tags, tag_drop_out, latent_sampling_method, create_image_every, save_hypernetwork_every, template_file, preview_from_txt2img, preview_prompt, preview_negative_prompt, preview_steps, preview_sampler_index, preview_cfg_scale, preview_seed, preview_width, preview_height):
+def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, gradient_step, data_root, log_directory, training_width, training_height, steps, clip_grad_mode, clip_grad_value, shuffle_tags, tag_drop_out, latent_sampling_method, create_image_every, save_hypernetwork_every, template_file, preview_from_txt2img, preview_prompt, preview_negative_prompt, preview_steps, preview_sampler_index, preview_cfg_scale, preview_seed, preview_width, preview_height):
# images allows training previews to have infotext. Importing it at the top causes a circular import problem.
from modules import images
@@ -448,6 +446,10 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, gradient_step,
return hypernetwork, filename
scheduler = LearnRateScheduler(learn_rate, steps, initial_step)
+
+ clip_grad = torch.nn.utils.clip_grad_value_ if clip_grad_mode == "value" else torch.nn.utils.clip_grad_norm_ if clip_grad_mode == "norm" else None
+ if clip_grad:
+ clip_grad_sched = LearnRateScheduler(clip_grad_value, steps, initial_step, verbose=False)
# dataset loading may take a while, so input validations and early returns should be done before this
shared.state.textinfo = f"Preparing dataset from {html.escape(data_root)}..."
@@ -466,7 +468,7 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, gradient_step,
shared.parallel_processing_allowed = False
shared.sd_model.cond_stage_model.to(devices.cpu)
shared.sd_model.first_stage_model.to(devices.cpu)
-
+
weights = hypernetwork.weights()
hypernetwork.train_mode()
@@ -525,6 +527,9 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, gradient_step,
if shared.state.interrupted:
break
+ if clip_grad:
+ clip_grad_sched.step(hypernetwork.step)
+
with devices.autocast():
x = batch.latent_sample.to(devices.device, non_blocking=pin_memory)
if tag_drop_out != 0 or shuffle_tags:
@@ -539,14 +544,14 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, gradient_step,
_loss_step += loss.item()
scaler.scale(loss).backward()
+
# go back until we reach gradient accumulation steps
if (j + 1) % gradient_step != 0:
continue
- # print(f"grad:{weights[0].grad.detach().cpu().abs().mean().item():.7f}")
- # scaler.unscale_(optimizer)
- # print(f"grad:{weights[0].grad.detach().cpu().abs().mean().item():.15f}")
- # torch.nn.utils.clip_grad_norm_(weights, max_norm=1.0)
- # print(f"grad:{weights[0].grad.detach().cpu().abs().mean().item():.15f}")
+
+ if clip_grad:
+ clip_grad(weights, clip_grad_sched.learn_rate)
+
scaler.step(optimizer)
scaler.update()
hypernetwork.step += 1
diff --git a/modules/processing.py b/modules/processing.py
index fd7c7015..a408d622 100644
--- a/modules/processing.py
+++ b/modules/processing.py
@@ -50,9 +50,9 @@ def apply_color_correction(correction, original_image):
correction,
channel_axis=2
), cv2.COLOR_LAB2RGB).astype("uint8"))
-
+
image = blendLayers(image, original_image, BlendType.LUMINOSITY)
-
+
return image
@@ -76,6 +76,24 @@ def apply_overlay(image, paste_loc, index, overlays):
return image
+def txt2img_image_conditioning(sd_model, x, width, height):
+ if sd_model.model.conditioning_key not in {'hybrid', 'concat'}:
+ # Dummy zero conditioning if we're not using inpainting model.
+ # Still takes up a bit of memory, but no encoder call.
+ # Pretty sure we can just make this a 1x1 image since its not going to be used besides its batch size.
+ return x.new_zeros(x.shape[0], 5, 1, 1, dtype=x.dtype, device=x.device)
+
+ # The "masked-image" in this case will just be all zeros since the entire image is masked.
+ image_conditioning = torch.zeros(x.shape[0], 3, height, width, device=x.device)
+ image_conditioning = sd_model.get_first_stage_encoding(sd_model.encode_first_stage(image_conditioning))
+
+ # Add the fake full 1s mask to the first dimension.
+ image_conditioning = torch.nn.functional.pad(image_conditioning, (0, 0, 0, 0, 1, 0), value=1.0)
+ image_conditioning = image_conditioning.to(x.dtype)
+
+ return image_conditioning
+
+
class StableDiffusionProcessing():
"""
The first set of paramaters: sd_models -> do_not_reload_embeddings represent the minimum required to create a StableDiffusionProcessing
@@ -136,28 +154,12 @@ class StableDiffusionProcessing():
self.all_negative_prompts = None
self.all_seeds = None
self.all_subseeds = None
+ self.iteration = 0
def txt2img_image_conditioning(self, x, width=None, height=None):
- if self.sampler.conditioning_key not in {'hybrid', 'concat'}:
- # Dummy zero conditioning if we're not using inpainting model.
- # Still takes up a bit of memory, but no encoder call.
- # Pretty sure we can just make this a 1x1 image since its not going to be used besides its batch size.
- return x.new_zeros(x.shape[0], 5, 1, 1)
-
- self.is_using_inpainting_conditioning = True
-
- height = height or self.height
- width = width or self.width
-
- # The "masked-image" in this case will just be all zeros since the entire image is masked.
- image_conditioning = torch.zeros(x.shape[0], 3, height, width, device=x.device)
- image_conditioning = self.sd_model.get_first_stage_encoding(self.sd_model.encode_first_stage(image_conditioning))
-
- # Add the fake full 1s mask to the first dimension.
- image_conditioning = torch.nn.functional.pad(image_conditioning, (0, 0, 0, 0, 1, 0), value=1.0)
- image_conditioning = image_conditioning.to(x.dtype)
+ self.is_using_inpainting_conditioning = self.sd_model.model.conditioning_key in {'hybrid', 'concat'}
- return image_conditioning
+ return txt2img_image_conditioning(self.sd_model, x, width or self.width, height or self.height)
def depth2img_image_conditioning(self, source_image):
# Use the AddMiDaS helper to Format our source image to suit the MiDaS model
@@ -420,7 +422,7 @@ def fix_seed(p):
p.subseed = get_fixed_seed(p.subseed)
-def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments, iteration=0, position_in_batch=0):
+def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments=None, iteration=0, position_in_batch=0):
index = position_in_batch + iteration * p.batch_size
clip_skip = getattr(p, 'clip_skip', opts.CLIP_stop_at_last_layers)
@@ -464,9 +466,15 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
try:
for k, v in p.override_settings.items():
setattr(opts, k, v)
- if k == 'sd_hypernetwork': shared.reload_hypernetworks() # make onchange call for changing hypernet
- if k == 'sd_model_checkpoint': sd_models.reload_model_weights() # make onchange call for changing SD model
- if k == 'sd_vae': sd_vae.reload_vae_weights() # make onchange call for changing VAE
+ if k == 'sd_hypernetwork':
+ shared.reload_hypernetworks() # make onchange call for changing hypernet
+
+ if k == 'sd_model_checkpoint':
+ sd_models.reload_model_weights() # make onchange call for changing SD model
+ p.sd_model = shared.sd_model
+
+ if k == 'sd_vae':
+ sd_vae.reload_vae_weights() # make onchange call for changing VAE
res = process_images_inner(p)
@@ -536,6 +544,29 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed:
infotexts = []
output_images = []
+ cached_uc = [None, None]
+ cached_c = [None, None]
+
+ def get_conds_with_caching(function, required_prompts, steps, cache):
+ """
+ Returns the result of calling function(shared.sd_model, required_prompts, steps)
+ using a cache to store the result if the same arguments have been used before.
+
+ cache is an array containing two elements. The first element is a tuple
+ representing the previously used arguments, or None if no arguments
+ have been used before. The second element is where the previously
+ computed result is stored.
+ """
+
+ if cache[0] is not None and (required_prompts, steps) == cache[0]:
+ return cache[1]
+
+ with devices.autocast():
+ cache[1] = function(shared.sd_model, required_prompts, steps)
+
+ cache[0] = (required_prompts, steps)
+ return cache[1]
+
with torch.no_grad(), p.sd_model.ema_scope():
with devices.autocast():
p.init(p.all_prompts, p.all_seeds, p.all_subseeds)
@@ -544,6 +575,8 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed:
state.job_count = p.n_iter
for n in range(p.n_iter):
+ p.iteration = n
+
if state.skipped:
state.skipped = False
@@ -561,9 +594,8 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed:
if p.scripts is not None:
p.scripts.process_batch(p, batch_number=n, prompts=prompts, seeds=seeds, subseeds=subseeds)
- with devices.autocast():
- uc = prompt_parser.get_learned_conditioning(shared.sd_model, negative_prompts, p.steps)
- c = prompt_parser.get_multicond_learned_conditioning(shared.sd_model, prompts, p.steps)
+ uc = get_conds_with_caching(prompt_parser.get_learned_conditioning, negative_prompts, p.steps, cached_uc)
+ c = get_conds_with_caching(prompt_parser.get_multicond_learned_conditioning, prompts, p.steps, cached_c)
if len(model_hijack.comments) > 0:
for comment in model_hijack.comments:
@@ -658,12 +690,17 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed:
class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
sampler = None
- def __init__(self, enable_hr: bool = False, denoising_strength: float = 0.75, firstphase_width: int = 0, firstphase_height: int = 0, hr_scale: float = 2.0, hr_upscaler: str = None, **kwargs):
+ def __init__(self, enable_hr: bool = False, denoising_strength: float = 0.75, firstphase_width: int = 0, firstphase_height: int = 0, hr_scale: float = 2.0, hr_upscaler: str = None, hr_second_pass_steps: int = 0, hr_resize_x: int = 0, hr_resize_y: int = 0, **kwargs):
super().__init__(**kwargs)
self.enable_hr = enable_hr
self.denoising_strength = denoising_strength
self.hr_scale = hr_scale
self.hr_upscaler = hr_upscaler
+ self.hr_second_pass_steps = hr_second_pass_steps
+ self.hr_resize_x = hr_resize_x
+ self.hr_resize_y = hr_resize_y
+ self.hr_upscale_to_x = hr_resize_x
+ self.hr_upscale_to_y = hr_resize_y
if firstphase_width != 0 or firstphase_height != 0:
print("firstphase_width/firstphase_height no longer supported; use hr_scale", file=sys.stderr)
@@ -671,14 +708,60 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
self.width = firstphase_width
self.height = firstphase_height
+ self.truncate_x = 0
+ self.truncate_y = 0
+
+
def init(self, all_prompts, all_seeds, all_subseeds):
if self.enable_hr:
- if state.job_count == -1:
- state.job_count = self.n_iter * 2
+ if self.hr_resize_x == 0 and self.hr_resize_y == 0:
+ self.extra_generation_params["Hires upscale"] = self.hr_scale
+ self.hr_upscale_to_x = int(self.width * self.hr_scale)
+ self.hr_upscale_to_y = int(self.height * self.hr_scale)
else:
+ self.extra_generation_params["Hires resize"] = f"{self.hr_resize_x}x{self.hr_resize_y}"
+
+ if self.hr_resize_y == 0:
+ self.hr_upscale_to_x = self.hr_resize_x
+ self.hr_upscale_to_y = self.hr_resize_x * self.height // self.width
+ elif self.hr_resize_x == 0:
+ self.hr_upscale_to_x = self.hr_resize_y * self.width // self.height
+ self.hr_upscale_to_y = self.hr_resize_y
+ else:
+ target_w = self.hr_resize_x
+ target_h = self.hr_resize_y
+ src_ratio = self.width / self.height
+ dst_ratio = self.hr_resize_x / self.hr_resize_y
+
+ if src_ratio < dst_ratio:
+ self.hr_upscale_to_x = self.hr_resize_x
+ self.hr_upscale_to_y = self.hr_resize_x * self.height // self.width
+ else:
+ self.hr_upscale_to_x = self.hr_resize_y * self.width // self.height
+ self.hr_upscale_to_y = self.hr_resize_y
+
+ self.truncate_x = (self.hr_upscale_to_x - target_w) // opt_f
+ self.truncate_y = (self.hr_upscale_to_y - target_h) // opt_f
+
+ # special case: the user has chosen to do nothing
+ if self.hr_upscale_to_x == self.width and self.hr_upscale_to_y == self.height:
+ self.enable_hr = False
+ self.denoising_strength = None
+ self.extra_generation_params.pop("Hires upscale", None)
+ self.extra_generation_params.pop("Hires resize", None)
+ return
+
+ if not state.processing_has_refined_job_count:
+ if state.job_count == -1:
+ state.job_count = self.n_iter
+
+ shared.total_tqdm.updateTotal((self.steps + (self.hr_second_pass_steps or self.steps)) * state.job_count)
state.job_count = state.job_count * 2
+ state.processing_has_refined_job_count = True
+
+ if self.hr_second_pass_steps:
+ self.extra_generation_params["Hires steps"] = self.hr_second_pass_steps
- self.extra_generation_params["Hires upscale"] = self.hr_scale
if self.hr_upscaler is not None:
self.extra_generation_params["Hires upscaler"] = self.hr_upscaler
@@ -695,8 +778,8 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
if not self.enable_hr:
return samples
- target_width = int(self.width * self.hr_scale)
- target_height = int(self.height * self.hr_scale)
+ target_width = self.hr_upscale_to_x
+ target_height = self.hr_upscale_to_y
def save_intermediate(image, index):
"""saves image before applying hires fix, if enabled in options; takes as an argument either an image or batch with latent space images"""
@@ -707,7 +790,8 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
if not isinstance(image, Image.Image):
image = sd_samplers.sample_to_image(image, index, approximation=0)
- images.save_image(image, self.outpath_samples, "", seeds[index], prompts[index], opts.samples_format, suffix="-before-highres-fix")
+ info = create_infotext(self, self.all_prompts, self.all_seeds, self.all_subseeds, [], iteration=self.iteration, position_in_batch=index)
+ images.save_image(image, self.outpath_samples, "", seeds[index], prompts[index], opts.samples_format, info=info, suffix="-before-highres-fix")
if latent_scale_mode is not None:
for i in range(samples.shape[0]):
@@ -750,13 +834,15 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
self.sampler = sd_samplers.create_sampler(self.sampler_name, self.sd_model)
+ samples = samples[:, :, self.truncate_y//2:samples.shape[2]-(self.truncate_y+1)//2, self.truncate_x//2:samples.shape[3]-(self.truncate_x+1)//2]
+
noise = create_random_tensors(samples.shape[1:], seeds=seeds, subseeds=subseeds, subseed_strength=subseed_strength, p=self)
# GC now before running the next img2img to prevent running out of memory
x = None
devices.torch_gc()
- samples = self.sampler.sample_img2img(self, samples, noise, conditioning, unconditional_conditioning, steps=self.steps, image_conditioning=image_conditioning)
+ samples = self.sampler.sample_img2img(self, samples, noise, conditioning, unconditional_conditioning, steps=self.hr_second_pass_steps or self.steps, image_conditioning=image_conditioning)
return samples
diff --git a/modules/scripts.py b/modules/scripts.py
index 722f8685..0c44f191 100644
--- a/modules/scripts.py
+++ b/modules/scripts.py
@@ -1,4 +1,5 @@
import os
+import re
import sys
import traceback
from collections import namedtuple
@@ -128,6 +129,15 @@ class Script:
"""unused"""
return ""
+ def elem_id(self, item_id):
+ """helper function to generate id for a HTML element, constructs final id out of script name, tab and user-supplied item_id"""
+
+ need_tabname = self.show(True) == self.show(False)
+ tabname = ('img2img' if self.is_img2img else 'txt2txt') + "_" if need_tabname else ""
+ title = re.sub(r'[^a-z_0-9]', '', re.sub(r'\s', '_', self.title().lower()))
+
+ return f'script_{tabname}{title}_{item_id}'
+
current_basedir = paths.script_path
diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py
index edcbaf52..fa2cd4bb 100644
--- a/modules/sd_hijack.py
+++ b/modules/sd_hijack.py
@@ -35,26 +35,35 @@ def apply_optimizations():
ldm.modules.diffusionmodules.model.nonlinearity = silu
ldm.modules.diffusionmodules.openaimodel.th = sd_hijack_unet.th
+
+ optimization_method = None
if cmd_opts.force_enable_xformers or (cmd_opts.xformers and shared.xformers_available and torch.version.cuda and (6, 0) <= torch.cuda.get_device_capability(shared.device) <= (9, 0)):
print("Applying xformers cross attention optimization.")
ldm.modules.attention.CrossAttention.forward = sd_hijack_optimizations.xformers_attention_forward
ldm.modules.diffusionmodules.model.AttnBlock.forward = sd_hijack_optimizations.xformers_attnblock_forward
+ optimization_method = 'xformers'
elif cmd_opts.opt_split_attention_v1:
print("Applying v1 cross attention optimization.")
ldm.modules.attention.CrossAttention.forward = sd_hijack_optimizations.split_cross_attention_forward_v1
+ optimization_method = 'V1'
elif not cmd_opts.disable_opt_split_attention and (cmd_opts.opt_split_attention_invokeai or not torch.cuda.is_available()):
if not invokeAI_mps_available and shared.device.type == 'mps':
print("The InvokeAI cross attention optimization for MPS requires the psutil package which is not installed.")
print("Applying v1 cross attention optimization.")
ldm.modules.attention.CrossAttention.forward = sd_hijack_optimizations.split_cross_attention_forward_v1
+ optimization_method = 'V1'
else:
print("Applying cross attention optimization (InvokeAI).")
ldm.modules.attention.CrossAttention.forward = sd_hijack_optimizations.split_cross_attention_forward_invokeAI
+ optimization_method = 'InvokeAI'
elif not cmd_opts.disable_opt_split_attention and (cmd_opts.opt_split_attention or torch.cuda.is_available()):
print("Applying cross attention optimization (Doggettx).")
ldm.modules.attention.CrossAttention.forward = sd_hijack_optimizations.split_cross_attention_forward
ldm.modules.diffusionmodules.model.AttnBlock.forward = sd_hijack_optimizations.cross_attention_attnblock_forward
+ optimization_method = 'Doggettx'
+
+ return optimization_method
def undo_optimizations():
@@ -75,6 +84,7 @@ class StableDiffusionModelHijack:
layers = None
circular_enabled = False
clip = None
+ optimization_method = None
embedding_db = modules.textual_inversion.textual_inversion.EmbeddingDatabase(cmd_opts.embeddings_dir)
@@ -94,7 +104,7 @@ class StableDiffusionModelHijack:
m.cond_stage_model.model.token_embedding = EmbeddingsWithFixes(m.cond_stage_model.model.token_embedding, self)
m.cond_stage_model = sd_hijack_open_clip.FrozenOpenCLIPEmbedderWithCustomWords(m.cond_stage_model, self)
- apply_optimizations()
+ self.optimization_method = apply_optimizations()
self.clip = m.cond_stage_model
diff --git a/modules/sd_models.py b/modules/sd_models.py
index 6dca4ddf..76a89e88 100644
--- a/modules/sd_models.py
+++ b/modules/sd_models.py
@@ -171,7 +171,10 @@ def get_state_dict_from_checkpoint(pl_sd):
def read_state_dict(checkpoint_file, print_global_state=False, map_location=None):
_, extension = os.path.splitext(checkpoint_file)
if extension.lower() == ".safetensors":
- pl_sd = safetensors.torch.load_file(checkpoint_file, device=map_location or shared.weight_load_location)
+ device = map_location or shared.weight_load_location
+ if device is None:
+ device = devices.get_cuda_device_string() if torch.cuda.is_available() else "cpu"
+ pl_sd = safetensors.torch.load_file(checkpoint_file, device=device)
else:
pl_sd = torch.load(checkpoint_file, map_location=map_location or shared.weight_load_location)
@@ -305,9 +308,6 @@ def load_model(checkpoint_info=None):
sd_config.model.params.unet_config.params.in_channels = 9
sd_config.model.params.finetune_keys = None
- # Create a "fake" config with a different name so that we know to unload it when switching models.
- checkpoint_info = checkpoint_info._replace(config=checkpoint_config.replace(".yaml", "-inpainting.yaml"))
-
if not hasattr(sd_config.model.params, "use_ema"):
sd_config.model.params.use_ema = False
diff --git a/modules/sd_samplers.py b/modules/sd_samplers.py
index e904d860..01221b89 100644
--- a/modules/sd_samplers.py
+++ b/modules/sd_samplers.py
@@ -97,8 +97,9 @@ sampler_extra_params = {
def setup_img2img_steps(p, steps=None):
if opts.img2img_fix_steps or steps is not None:
- steps = int((steps or p.steps) / min(p.denoising_strength, 0.999)) if p.denoising_strength > 0 else 0
- t_enc = p.steps - 1
+ requested_steps = (steps or p.steps)
+ steps = int(requested_steps / min(p.denoising_strength, 0.999)) if p.denoising_strength > 0 else 0
+ t_enc = requested_steps - 1
else:
steps = p.steps
t_enc = int(min(p.denoising_strength, 0.999) * steps)
@@ -462,6 +463,13 @@ class KDiffusionSampler:
return extra_params_kwargs
def get_sigmas(self, p, steps):
+ discard_next_to_last_sigma = self.config is not None and self.config.options.get('discard_next_to_last_sigma', False)
+ if opts.always_discard_next_to_last_sigma and not discard_next_to_last_sigma:
+ discard_next_to_last_sigma = True
+ p.extra_generation_params["Discard penultimate sigma"] = True
+
+ steps += 1 if discard_next_to_last_sigma else 0
+
if p.sampler_noise_scheduler_override:
sigmas = p.sampler_noise_scheduler_override(steps)
elif self.config is not None and self.config.options.get('scheduler', None) == 'karras':
@@ -471,7 +479,7 @@ class KDiffusionSampler:
else:
sigmas = self.model_wrap.get_sigmas(steps)
- if self.config is not None and self.config.options.get('discard_next_to_last_sigma', False):
+ if discard_next_to_last_sigma:
sigmas = torch.cat([sigmas[:-2], sigmas[-1:]])
return sigmas
diff --git a/modules/shared.py b/modules/shared.py
index 9c9fd857..b7a3ce5c 100644
--- a/modules/shared.py
+++ b/modules/shared.py
@@ -153,6 +153,7 @@ class State:
job = ""
job_no = 0
job_count = 0
+ processing_has_refined_job_count = False
job_timestamp = '0'
sampling_step = 0
sampling_steps = 0
@@ -183,6 +184,7 @@ class State:
"interrupted": self.interrupted,
"job": self.job,
"job_count": self.job_count,
+ "job_timestamp": self.job_timestamp,
"job_no": self.job_no,
"sampling_step": self.sampling_step,
"sampling_steps": self.sampling_steps,
@@ -193,6 +195,7 @@ class State:
def begin(self):
self.sampling_step = 0
self.job_count = -1
+ self.processing_has_refined_job_count = False
self.job_no = 0
self.job_timestamp = datetime.datetime.now().strftime("%Y%m%d%H%M%S")
self.current_latent = None
@@ -213,12 +216,13 @@ class State:
"""sets self.current_image from self.current_latent if enough sampling steps have been made after the last call to this"""
def set_current_image(self):
+ if not parallel_processing_allowed:
+ return
+
if self.sampling_step - self.current_image_sampling_step >= opts.show_progress_every_n_steps and opts.show_progress_every_n_steps > 0:
self.do_set_current_image()
def do_set_current_image(self):
- if not parallel_processing_allowed:
- return
if self.current_latent is None:
return
@@ -230,6 +234,7 @@ class State:
self.current_image_sampling_step = self.sampling_step
+
state = State()
artist_db = modules.artists.ArtistsDatabase(os.path.join(script_path, 'artists.csv'))
@@ -437,6 +442,7 @@ options_templates.update(options_section(('sampler-params', "Sampler parameters"
's_tmin': OptionInfo(0.0, "sigma tmin", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}),
's_noise': OptionInfo(1.0, "sigma noise", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}),
'eta_noise_seed_delta': OptionInfo(0, "Eta noise seed delta", gr.Number, {"precision": 0}),
+ 'always_discard_next_to_last_sigma': OptionInfo(False, "Always discard next-to-last sigma"),
}))
options_templates.update(options_section((None, "Hidden options"), {
@@ -570,6 +576,7 @@ latent_upscale_modes = {
"Latent (bicubic)": {"mode": "bicubic", "antialias": False},
"Latent (bicubic antialiased)": {"mode": "bicubic", "antialias": True},
"Latent (nearest)": {"mode": "nearest", "antialias": False},
+ "Latent (nearest-exact)": {"mode": "nearest-exact", "antialias": False},
}
sd_upscalers = []
@@ -605,7 +612,7 @@ class TotalTQDM:
return
if self._tqdm is None:
self.reset()
- self._tqdm.total=new_total
+ self._tqdm.total = new_total
def clear(self):
if self._tqdm is not None:
diff --git a/modules/textual_inversion/learn_schedule.py b/modules/textual_inversion/learn_schedule.py
index dd0c0ad1..f63fc72f 100644
--- a/modules/textual_inversion/learn_schedule.py
+++ b/modules/textual_inversion/learn_schedule.py
@@ -58,14 +58,19 @@ class LearnRateScheduler:
self.finished = False
- def apply(self, optimizer, step_number):
+ def step(self, step_number):
if step_number < self.end_step:
- return
+ return False
try:
(self.learn_rate, self.end_step) = next(self.schedules)
- except Exception:
+ except StopIteration:
self.finished = True
+ return False
+ return True
+
+ def apply(self, optimizer, step_number):
+ if not self.step(step_number):
return
if self.verbose:
diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py
index 214db01c..24b43045 100644
--- a/modules/textual_inversion/textual_inversion.py
+++ b/modules/textual_inversion/textual_inversion.py
@@ -251,7 +251,7 @@ def validate_train_inputs(model_name, learn_rate, batch_size, gradient_step, dat
if save_model_every or create_image_every:
assert log_directory, "Log directory is empty"
-def train_embedding(embedding_name, learn_rate, batch_size, gradient_step, data_root, log_directory, training_width, training_height, steps, shuffle_tags, tag_drop_out, latent_sampling_method, create_image_every, save_embedding_every, template_file, save_image_with_stored_embedding, preview_from_txt2img, preview_prompt, preview_negative_prompt, preview_steps, preview_sampler_index, preview_cfg_scale, preview_seed, preview_width, preview_height):
+def train_embedding(embedding_name, learn_rate, batch_size, gradient_step, data_root, log_directory, training_width, training_height, steps, clip_grad_mode, clip_grad_value, shuffle_tags, tag_drop_out, latent_sampling_method, create_image_every, save_embedding_every, template_file, save_image_with_stored_embedding, preview_from_txt2img, preview_prompt, preview_negative_prompt, preview_steps, preview_sampler_index, preview_cfg_scale, preview_seed, preview_width, preview_height):
save_embedding_every = save_embedding_every or 0
create_image_every = create_image_every or 0
validate_train_inputs(embedding_name, learn_rate, batch_size, gradient_step, data_root, template_file, steps, save_embedding_every, create_image_every, log_directory, name="embedding")
@@ -294,6 +294,11 @@ def train_embedding(embedding_name, learn_rate, batch_size, gradient_step, data_
return embedding, filename
scheduler = LearnRateScheduler(learn_rate, steps, initial_step)
+ clip_grad = torch.nn.utils.clip_grad_value_ if clip_grad_mode == "value" else \
+ torch.nn.utils.clip_grad_norm_ if clip_grad_mode == "norm" else \
+ None
+ if clip_grad:
+ clip_grad_sched = LearnRateScheduler(clip_grad_value, steps, initial_step, verbose=False)
# dataset loading may take a while, so input validations and early returns should be done before this
shared.state.textinfo = f"Preparing dataset from {html.escape(data_root)}..."
old_parallel_processing_allowed = shared.parallel_processing_allowed
@@ -325,7 +330,6 @@ def train_embedding(embedding_name, learn_rate, batch_size, gradient_step, data_
else:
print("No saved optimizer exists in checkpoint")
-
scaler = torch.cuda.amp.GradScaler()
batch_size = ds.batch_size
@@ -341,6 +345,9 @@ def train_embedding(embedding_name, learn_rate, batch_size, gradient_step, data_
forced_filename = "<none>"
embedding_yet_to_be_embedded = False
+ is_training_inpainting_model = shared.sd_model.model.conditioning_key in {'hybrid', 'concat'}
+ img_c = None
+
pbar = tqdm.tqdm(total=steps - initial_step)
try:
for i in range((steps-initial_step) * gradient_step):
@@ -358,14 +365,22 @@ def train_embedding(embedding_name, learn_rate, batch_size, gradient_step, data_
if shared.state.interrupted:
break
+ if clip_grad:
+ clip_grad_sched.step(embedding.step)
+
with devices.autocast():
- # c = stack_conds(batch.cond).to(devices.device)
- # mask = torch.tensor(batch.emb_index).to(devices.device, non_blocking=pin_memory)
- # print(mask)
- # c[:, 1:1+embedding.vec.shape[0]] = embedding.vec.to(devices.device, non_blocking=pin_memory)
x = batch.latent_sample.to(devices.device, non_blocking=pin_memory)
c = shared.sd_model.cond_stage_model(batch.cond_text)
- loss = shared.sd_model(x, c)[0] / gradient_step
+
+ if is_training_inpainting_model:
+ if img_c is None:
+ img_c = processing.txt2img_image_conditioning(shared.sd_model, c, training_width, training_height)
+
+ cond = {"c_concat": [img_c], "c_crossattn": [c]}
+ else:
+ cond = c
+
+ loss = shared.sd_model(x, cond)[0] / gradient_step
del x
_loss_step += loss.item()
@@ -374,6 +389,10 @@ def train_embedding(embedding_name, learn_rate, batch_size, gradient_step, data_
# go back until we reach gradient accumulation steps
if (j + 1) % gradient_step != 0:
continue
+
+ if clip_grad:
+ clip_grad(embedding.vec, clip_grad_sched.learn_rate)
+
scaler.step(optimizer)
scaler.update()
embedding.step += 1
diff --git a/modules/txt2img.py b/modules/txt2img.py
index e189a899..38b5f591 100644
--- a/modules/txt2img.py
+++ b/modules/txt2img.py
@@ -8,7 +8,7 @@ import modules.processing as processing
from modules.ui import plaintext_to_html
-def txt2img(prompt: str, negative_prompt: str, prompt_style: str, prompt_style2: str, steps: int, sampler_index: int, restore_faces: bool, tiling: bool, n_iter: int, batch_size: int, cfg_scale: float, seed: int, subseed: int, subseed_strength: float, seed_resize_from_h: int, seed_resize_from_w: int, seed_enable_extras: bool, height: int, width: int, enable_hr: bool, denoising_strength: float, hr_scale: float, hr_upscaler: str, *args):
+def txt2img(prompt: str, negative_prompt: str, prompt_style: str, prompt_style2: str, steps: int, sampler_index: int, restore_faces: bool, tiling: bool, n_iter: int, batch_size: int, cfg_scale: float, seed: int, subseed: int, subseed_strength: float, seed_resize_from_h: int, seed_resize_from_w: int, seed_enable_extras: bool, height: int, width: int, enable_hr: bool, denoising_strength: float, hr_scale: float, hr_upscaler: str, hr_second_pass_steps: int, hr_resize_x: int, hr_resize_y: int, *args):
p = StableDiffusionProcessingTxt2Img(
sd_model=shared.sd_model,
outpath_samples=opts.outdir_samples or opts.outdir_txt2img_samples,
@@ -35,6 +35,9 @@ def txt2img(prompt: str, negative_prompt: str, prompt_style: str, prompt_style2:
denoising_strength=denoising_strength if enable_hr else None,
hr_scale=hr_scale,
hr_upscaler=hr_upscaler,
+ hr_second_pass_steps=hr_second_pass_steps,
+ hr_resize_x=hr_resize_x,
+ hr_resize_y=hr_resize_y,
)
p.scripts = modules.scripts.scripts_txt2img
diff --git a/modules/ui.py b/modules/ui.py
index bfc93634..81d96c5b 100644
--- a/modules/ui.py
+++ b/modules/ui.py
@@ -162,16 +162,14 @@ def save_files(js_data, images, do_make_zip, index):
return gr.File.update(value=fullfns, visible=True), plaintext_to_html(f"Saved: {filenames[0]}")
-
-
-def calc_time_left(progress, threshold, label, force_display):
+def calc_time_left(progress, threshold, label, force_display, show_eta):
if progress == 0:
return ""
else:
time_since_start = time.time() - shared.state.time_start
eta = (time_since_start/progress)
eta_relative = eta-time_since_start
- if (eta_relative > threshold and progress > 0.02) or force_display:
+ if (eta_relative > threshold and show_eta) or force_display:
if eta_relative > 3600:
return label + time.strftime('%H:%M:%S', time.gmtime(eta_relative))
elif eta_relative > 60:
@@ -193,7 +191,10 @@ def check_progress_call(id_part):
if shared.state.sampling_steps > 0:
progress += 1 / shared.state.job_count * shared.state.sampling_step / shared.state.sampling_steps
- time_left = calc_time_left( progress, 1, " ETA: ", shared.state.time_left_force_display )
+ # Show progress percentage and time left at the same moment, and base it also on steps done
+ show_eta = progress >= 0.01 or shared.state.sampling_step >= 10
+
+ time_left = calc_time_left(progress, 1, " ETA: ", shared.state.time_left_force_display, show_eta)
if time_left != "":
shared.state.time_left_force_display = True
@@ -201,7 +202,7 @@ def check_progress_call(id_part):
progressbar = ""
if opts.show_progressbar:
- progressbar = f"""<div class='progressDiv'><div class='progress' style="overflow:visible;width:{progress * 100}%;white-space:nowrap;">{"&nbsp;" * 2 + str(int(progress*100))+"%" + time_left if progress > 0.01 else ""}</div></div>"""
+ progressbar = f"""<div class='progressDiv'><div class='progress' style="overflow:visible;width:{progress * 100}%;white-space:nowrap;">{"&nbsp;" * 2 + str(int(progress*100))+"%" + time_left if show_eta else ""}</div></div>"""
image = gr_show(False)
preview_visibility = gr_show(False)
@@ -559,7 +560,7 @@ Requested path was: {f}
generation_info = None
with gr.Column():
with gr.Row(elem_id=f"image_buttons_{tabname}"):
- open_folder_button = gr.Button(folder_symbol, elem_id="hidden_element" if shared.cmd_opts.hide_ui_dir_config else 'open_folder')
+ open_folder_button = gr.Button(folder_symbol, elem_id="hidden_element" if shared.cmd_opts.hide_ui_dir_config else f'open_folder_{tabname}')
if tabname != "extras":
save = gr.Button('Save', elem_id=f'save_{tabname}')
@@ -575,13 +576,13 @@ Requested path was: {f}
if tabname != "extras":
with gr.Row():
- download_files = gr.File(None, file_count="multiple", interactive=False, show_label=False, visible=False)
+ download_files = gr.File(None, file_count="multiple", interactive=False, show_label=False, visible=False, elem_id=f'download_files_{tabname}')
with gr.Group():
- html_info = gr.HTML()
- html_log = gr.HTML()
+ html_info = gr.HTML(elem_id=f'html_info_{tabname}')
+ html_log = gr.HTML(elem_id=f'html_log_{tabname}')
- generation_info = gr.Textbox(visible=False)
+ generation_info = gr.Textbox(visible=False, elem_id=f'generation_info_{tabname}')
if tabname == 'txt2img' or tabname == 'img2img':
generation_info_button = gr.Button(visible=False, elem_id=f"{tabname}_generation_info_button")
generation_info_button.click(
@@ -623,9 +624,9 @@ Requested path was: {f}
)
else:
- html_info_x = gr.HTML()
- html_info = gr.HTML()
- html_log = gr.HTML()
+ html_info_x = gr.HTML(elem_id=f'html_info_x_{tabname}')
+ html_info = gr.HTML(elem_id=f'html_info_{tabname}')
+ html_log = gr.HTML(elem_id=f'html_log_{tabname}')
parameters_copypaste.bind_buttons(buttons, result_gallery, "txt2img" if tabname == "txt2img" else None)
return result_gallery, generation_info if tabname != "extras" else html_info_x, html_info, html_log
@@ -636,10 +637,10 @@ def create_sampler_and_steps_selection(choices, tabname):
with FormRow(elem_id=f"sampler_selection_{tabname}"):
sampler_index = gr.Dropdown(label='Sampling method', elem_id=f"{tabname}_sampling", choices=[x.name for x in choices], value=choices[0].name, type="index")
sampler_index.save_to_config = True
- steps = gr.Slider(minimum=1, maximum=150, step=1, elem_id=f"{tabname}_steps", label="Sampling Steps", value=20)
+ steps = gr.Slider(minimum=1, maximum=150, step=1, elem_id=f"{tabname}_steps", label="Sampling steps", value=20)
else:
with FormGroup(elem_id=f"sampler_selection_{tabname}"):
- steps = gr.Slider(minimum=1, maximum=150, step=1, elem_id=f"{tabname}_steps", label="Sampling Steps", value=20)
+ steps = gr.Slider(minimum=1, maximum=150, step=1, elem_id=f"{tabname}_steps", label="Sampling steps", value=20)
sampler_index = gr.Radio(label='Sampling method', elem_id=f"{tabname}_sampling", choices=[x.name for x in choices], value=choices[0].name, type="index")
return steps, sampler_index
@@ -708,10 +709,16 @@ def create_ui():
enable_hr = gr.Checkbox(label='Hires. fix', value=False, elem_id="txt2img_enable_hr")
elif category == "hires_fix":
- with FormRow(visible=False, elem_id="txt2img_hires_fix") as hr_options:
- hr_upscaler = gr.Dropdown(label="Upscaler", elem_id="txt2img_hr_upscaler", choices=[*shared.latent_upscale_modes, *[x.name for x in shared.sd_upscalers]], value=shared.latent_upscale_default_mode)
- hr_scale = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Upscale by", value=2.0, elem_id="txt2img_hr_scale")
- denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.7, elem_id="txt2img_denoising_strength")
+ with FormGroup(visible=False, elem_id="txt2img_hires_fix") as hr_options:
+ with FormRow(elem_id="txt2img_hires_fix_row1"):
+ hr_upscaler = gr.Dropdown(label="Upscaler", elem_id="txt2img_hr_upscaler", choices=[*shared.latent_upscale_modes, *[x.name for x in shared.sd_upscalers]], value=shared.latent_upscale_default_mode)
+ hr_second_pass_steps = gr.Slider(minimum=0, maximum=150, step=1, label='Hires steps', value=0, elem_id="txt2img_hires_steps")
+ denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.7, elem_id="txt2img_denoising_strength")
+
+ with FormRow(elem_id="txt2img_hires_fix_row2"):
+ hr_scale = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Upscale by", value=2.0, elem_id="txt2img_hr_scale")
+ hr_resize_x = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize width to", value=0, elem_id="txt2img_hr_resize_x")
+ hr_resize_y = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize height to", value=0, elem_id="txt2img_hr_resize_y")
elif category == "batch":
if not opts.dimensions_and_batch_together:
@@ -752,6 +759,9 @@ def create_ui():
denoising_strength,
hr_scale,
hr_upscaler,
+ hr_second_pass_steps,
+ hr_resize_x,
+ hr_resize_y,
] + custom_inputs,
outputs=[
@@ -803,6 +813,9 @@ def create_ui():
(hr_options, lambda d: gr.Row.update(visible="Denoising strength" in d)),
(hr_scale, "Hires upscale"),
(hr_upscaler, "Hires upscaler"),
+ (hr_second_pass_steps, "Hires steps"),
+ (hr_resize_x, "Hires resize-1"),
+ (hr_resize_y, "Hires resize-2"),
*modules.scripts.scripts_txt2img.infotext_fields
]
parameters_copypaste.add_paste_fields("txt2img", None, txt2img_paste_fields)
@@ -1280,38 +1293,48 @@ def create_ui():
with gr.Tab(label="Train"):
gr.HTML(value="<p style='margin-bottom: 0.7em'>Train an embedding or Hypernetwork; you must specify a directory with a set of 1:1 ratio images <a href=\"https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Textual-Inversion\" style=\"font-weight:bold;\">[wiki]</a></p>")
- with gr.Row():
+ with FormRow():
train_embedding_name = gr.Dropdown(label='Embedding', elem_id="train_embedding", choices=sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys()))
create_refresh_button(train_embedding_name, sd_hijack.model_hijack.embedding_db.load_textual_inversion_embeddings, lambda: {"choices": sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())}, "refresh_train_embedding_name")
- with gr.Row():
+
train_hypernetwork_name = gr.Dropdown(label='Hypernetwork', elem_id="train_hypernetwork", choices=[x for x in shared.hypernetworks.keys()])
create_refresh_button(train_hypernetwork_name, shared.reload_hypernetworks, lambda: {"choices": sorted([x for x in shared.hypernetworks.keys()])}, "refresh_train_hypernetwork_name")
- with gr.Row():
+
+ with FormRow():
embedding_learn_rate = gr.Textbox(label='Embedding Learning rate', placeholder="Embedding Learning rate", value="0.005", elem_id="train_embedding_learn_rate")
hypernetwork_learn_rate = gr.Textbox(label='Hypernetwork Learning rate', placeholder="Hypernetwork Learning rate", value="0.00001", elem_id="train_hypernetwork_learn_rate")
+
+ with FormRow():
+ clip_grad_mode = gr.Dropdown(value="disabled", label="Gradient Clipping", choices=["disabled", "value", "norm"])
+ clip_grad_value = gr.Textbox(placeholder="Gradient clip value", value="0.1", show_label=False)
+
+ with FormRow():
+ batch_size = gr.Number(label='Batch size', value=1, precision=0, elem_id="train_batch_size")
+ gradient_step = gr.Number(label='Gradient accumulation steps', value=1, precision=0, elem_id="train_gradient_step")
- batch_size = gr.Number(label='Batch size', value=1, precision=0, elem_id="train_batch_size")
- gradient_step = gr.Number(label='Gradient accumulation steps', value=1, precision=0, elem_id="train_gradient_step")
dataset_directory = gr.Textbox(label='Dataset directory', placeholder="Path to directory with input images", elem_id="train_dataset_directory")
log_directory = gr.Textbox(label='Log directory', placeholder="Path to directory where to write outputs", value="textual_inversion", elem_id="train_log_directory")
template_file = gr.Textbox(label='Prompt template file', value=os.path.join(script_path, "textual_inversion_templates", "style_filewords.txt"), elem_id="train_template_file")
training_width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="train_training_width")
training_height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="train_training_height")
steps = gr.Number(label='Max steps', value=100000, precision=0, elem_id="train_steps")
- create_image_every = gr.Number(label='Save an image to log directory every N steps, 0 to disable', value=500, precision=0, elem_id="train_create_image_every")
- save_embedding_every = gr.Number(label='Save a copy of embedding to log directory every N steps, 0 to disable', value=500, precision=0, elem_id="train_save_embedding_every")
+
+ with FormRow():
+ create_image_every = gr.Number(label='Save an image to log directory every N steps, 0 to disable', value=500, precision=0, elem_id="train_create_image_every")
+ save_embedding_every = gr.Number(label='Save a copy of embedding to log directory every N steps, 0 to disable', value=500, precision=0, elem_id="train_save_embedding_every")
+
save_image_with_stored_embedding = gr.Checkbox(label='Save images with embedding in PNG chunks', value=True, elem_id="train_save_image_with_stored_embedding")
preview_from_txt2img = gr.Checkbox(label='Read parameters (prompt, etc...) from txt2img tab when making previews', value=False, elem_id="train_preview_from_txt2img")
- with gr.Row():
- shuffle_tags = gr.Checkbox(label="Shuffle tags by ',' when creating prompts.", value=False, elem_id="train_shuffle_tags")
- tag_drop_out = gr.Slider(minimum=0, maximum=1, step=0.1, label="Drop out tags when creating prompts.", value=0, elem_id="train_tag_drop_out")
- with gr.Row():
- latent_sampling_method = gr.Radio(label='Choose latent sampling method', value="once", choices=['once', 'deterministic', 'random'], elem_id="train_latent_sampling_method")
+
+ shuffle_tags = gr.Checkbox(label="Shuffle tags by ',' when creating prompts.", value=False, elem_id="train_shuffle_tags")
+ tag_drop_out = gr.Slider(minimum=0, maximum=1, step=0.1, label="Drop out tags when creating prompts.", value=0, elem_id="train_tag_drop_out")
+
+ latent_sampling_method = gr.Radio(label='Choose latent sampling method', value="once", choices=['once', 'deterministic', 'random'], elem_id="train_latent_sampling_method")
with gr.Row():
+ train_embedding = gr.Button(value="Train Embedding", variant='primary', elem_id="train_train_embedding")
interrupt_training = gr.Button(value="Interrupt", elem_id="train_interrupt_training")
train_hypernetwork = gr.Button(value="Train Hypernetwork", variant='primary', elem_id="train_train_hypernetwork")
- train_embedding = gr.Button(value="Train Embedding", variant='primary', elem_id="train_train_embedding")
params = script_callbacks.UiTrainTabParams(txt2img_preview_params)
@@ -1401,6 +1424,8 @@ def create_ui():
training_width,
training_height,
steps,
+ clip_grad_mode,
+ clip_grad_value,
shuffle_tags,
tag_drop_out,
latent_sampling_method,
@@ -1430,6 +1455,8 @@ def create_ui():
training_width,
training_height,
steps,
+ clip_grad_mode,
+ clip_grad_value,
shuffle_tags,
tag_drop_out,
latent_sampling_method,
@@ -1669,7 +1696,9 @@ def create_ui():
if os.path.exists("html/footer.html"):
with open("html/footer.html", encoding="utf8") as file:
- gr.HTML(file.read(), elem_id="footer")
+ footer = file.read()
+ footer = footer.format(versions=versions_html())
+ gr.HTML(footer, elem_id="footer")
text_settings = gr.Textbox(elem_id="settings_json", value=lambda: opts.dumpjson(), visible=False)
settings_submit.click(
@@ -1794,6 +1823,7 @@ def create_ui():
visit(img2img_interface, loadsave, "img2img")
visit(extras_interface, loadsave, "extras")
visit(modelmerger_interface, loadsave, "modelmerger")
+ visit(train_interface, loadsave, "train")
if not error_loading and (not os.path.exists(ui_config_file) or settings_count != len(ui_settings)):
with open(ui_config_file, "w", encoding="utf8") as file:
@@ -1829,3 +1859,30 @@ def reload_javascript():
if not hasattr(shared, 'GradioTemplateResponseOriginal'):
shared.GradioTemplateResponseOriginal = gradio.routes.templates.TemplateResponse
+
+
+def versions_html():
+ import torch
+ import launch
+
+ python_version = ".".join([str(x) for x in sys.version_info[0:3]])
+ commit = launch.commit_hash()
+ short_commit = commit[0:8]
+
+ if shared.xformers_available:
+ import xformers
+ xformers_version = xformers.__version__
+ else:
+ xformers_version = "N/A"
+
+ return f"""
+python: <span title="{sys.version}">{python_version}</span>
+ • 
+torch: {torch.__version__}
+ • 
+xformers: {xformers_version}
+ • 
+gradio: {gr.__version__}
+ • 
+commit: <a href="https://github.com/AUTOMATIC1111/stable-diffusion-webui/commit/{commit}">{short_commit}</a>
+"""
generated by cgit v1.2.3 (git 2.25.1) at 2025-05-02 03:20:57 +0000