hoelee/stable-diffusion-webui
1
0
Fork 0
mirror of https://github.com/AUTOMATIC1111/stable-diffusion-webui.git synced 2025-08-08 13:19:54 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge branch 'master' into test_resolve_conflicts

Browse Source
This commit is contained in:
MalumaDev
2022-10-15 16:20:17 +02:00
committed by GitHub
parent 37d7ffb415 d3ffc962dd
commit 7b7561f6e4
32 changed files with 1094 additions and 292 deletions
Download Patch File Download Diff File Expand all files Collapse all files

79
modules/textual_inversion/textual_inversion.py
View File

@@ -6,6 +6,7 @@ import torch
import tqdm
import html
import datetime
import csv
from PIL import Image, PngImagePlugin
@@ -172,15 +173,33 @@ def create_embedding(name, num_vectors_per_token, init_text='*'):
return fn
def batched(dataset, total, n=1):
for ndx in range(0, total, n):
yield [dataset.__getitem__(i) for i in range(ndx, min(ndx + n, total))]
def write_loss(log_directory, filename, step, epoch_len, values):
if shared.opts.training_write_csv_every == 0:
return
if step % shared.opts.training_write_csv_every != 0:
return
write_csv_header = False if os.path.exists(os.path.join(log_directory, filename)) else True
with open(os.path.join(log_directory, filename), "a+", newline='') as fout:
csv_writer = csv.DictWriter(fout, fieldnames=["step", "epoch", "epoch_step", *(values.keys())])
if write_csv_header:
csv_writer.writeheader()
epoch = step // epoch_len
epoch_step = step - epoch * epoch_len
csv_writer.writerow({
"step": step + 1,
"epoch": epoch + 1,
"epoch_step": epoch_step + 1,
**values,
})
def train_embedding(embedding_name, learn_rate, data_root, log_directory, training_width, training_height, steps,
create_image_every, save_embedding_every, template_file, save_image_with_stored_embedding,
preview_image_prompt, batch_size=1,
gradient_accumulation=1):
def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_directory, training_width, training_height, steps, 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):
assert embedding_name, 'embedding not selected'
shared.state.textinfo = "Initializing textual inversion training..."
@@ -212,11 +231,7 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini
shared.state.textinfo = f"Preparing dataset from {html.escape(data_root)}..."
with torch.autocast("cuda"):
ds = modules.textual_inversion.dataset.PersonalizedBase(data_root=data_root, width=training_width,
height=training_height,
repeats=shared.opts.training_image_repeats_per_epoch,
placeholder_token=embedding_name, model=shared.sd_model,
device=devices.device, template_file=template_file)
ds = modules.textual_inversion.dataset.PersonalizedBase(data_root=data_root, width=training_width, height=training_height, repeats=shared.opts.training_image_repeats_per_epoch, placeholder_token=embedding_name, model=shared.sd_model, device=devices.device, template_file=template_file, batch_size=batch_size)
hijack = sd_hijack.model_hijack
@@ -235,8 +250,8 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini
scheduler = LearnRateScheduler(learn_rate, steps, ititial_step)
optimizer = torch.optim.AdamW([embedding.vec], lr=scheduler.learn_rate)
pbar = tqdm.tqdm(enumerate(batched(ds, steps - ititial_step, batch_size)), total=steps - ititial_step)
for i, entry in pbar:
pbar = tqdm.tqdm(enumerate(ds), total=steps-ititial_step)
for i, entries in pbar:
embedding.step = i + ititial_step
scheduler.apply(optimizer, embedding.step)
@@ -247,11 +262,9 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini
break
with torch.autocast("cuda"):
c = cond_model([e.cond_text for e in entry])
x = torch.stack([e.latent for e in entry]).to(devices.device)
c = cond_model([entry.cond_text for entry in entries])
x = torch.stack([entry.latent for entry in entries]).to(devices.device)
loss = shared.sd_model(x, c)[0]
del x
losses[embedding.step % losses.shape[0]] = loss.item()
@@ -271,21 +284,37 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini
last_saved_file = os.path.join(embedding_dir, f'{embedding_name}-{embedding.step}.pt')
embedding.save(last_saved_file)
write_loss(log_directory, "textual_inversion_loss.csv", embedding.step, len(ds), {
"loss": f"{losses.mean():.7f}",
"learn_rate": scheduler.learn_rate
})
if embedding.step > 0 and images_dir is not None and embedding.step % create_image_every == 0:
last_saved_image = os.path.join(images_dir, f'{embedding_name}-{embedding.step}.png')
preview_text = entry[0].cond_text if preview_image_prompt == "" else preview_image_prompt
p = processing.StableDiffusionProcessingTxt2Img(
sd_model=shared.sd_model,
prompt=preview_text,
steps=20,
height=training_height,
width=training_width,
do_not_save_grid=True,
do_not_save_samples=True,
)
if preview_from_txt2img:
p.prompt = preview_prompt
p.negative_prompt = preview_negative_prompt
p.steps = preview_steps
p.sampler_index = preview_sampler_index
p.cfg_scale = preview_cfg_scale
p.seed = preview_seed
p.width = preview_width
p.height = preview_height
else:
p.prompt = entries[0].cond_text
p.steps = 20
p.width = training_width
p.height = training_height
preview_text = p.prompt
processed = processing.process_images(p)
image = processed.images[0]
@@ -320,7 +349,7 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini
<p>
Loss: {losses.mean():.7f}<br/>
Step: {embedding.step}<br/>
Last prompt: {html.escape(entry[-1].cond_text)}<br/>
Last prompt: {html.escape(entries[0].cond_text)}<br/>
Last saved embedding: {html.escape(last_saved_file)}<br/>
Last saved image: {html.escape(last_saved_image)}<br/>
</p>