Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-18 16:54:33 +02:00
Code Issues Packages Projects Releases Wiki Activity

Remove unnecessary ipex code

Browse Source
This commit is contained in:
Disty0
2023-06-12 04:45:29 +03:00
parent c9e58c9604
commit ab255b732b
2 changed files with 2 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/hypernetworks/hypernetwork.py
+1 -13
View File
@@ -8,10 +8,6 @@ from statistics import stdev, mean
from rich import progress
import tqdm
import torch
try:
import intel_extension_for_pytorch as ipex # pylint: disable=import-error, unused-import
except:
pass
from torch import einsum
from torch.nn.init import normal_, xavier_normal_, xavier_uniform_, kaiming_normal_, kaiming_uniform_, zeros_
from einops import rearrange, repeat
@@ -594,10 +590,7 @@ def train_hypernetwork(id_task, hypernetwork_name, learn_rate, batch_size, gradi
print(e)
if shared.cmd_opts.use_ipex:
#scaler = ipex.cpu.autocast._grad_scaler.GradScaler()
shared.sd_model = shared.sd_model.to(dtype=torch.float32)
shared.sd_model.train()
shared.sd_model, optimizer = ipex.optimize(shared.sd_model, optimizer=optimizer, dtype=devices.dtype)
pass
else:
scaler = torch.cuda.amp.GradScaler()
@@ -724,8 +717,6 @@ def train_hypernetwork(id_task, hypernetwork_name, learn_rate, batch_size, gradi
cuda_rng_state = torch.xpu.get_rng_state_all()
elif torch.cuda.is_available():
cuda_rng_state = torch.cuda.get_rng_state_all()
if shared.cmd_opts.use_ipex:
shared.sd_model = shared.sd_model.to(dtype=devices.dtype)
shared.sd_model.cond_stage_model.to(devices.device)
shared.sd_model.first_stage_model.to(devices.device)
@@ -757,9 +748,6 @@ def train_hypernetwork(id_task, hypernetwork_name, learn_rate, batch_size, gradi
processed = processing.process_images(p)
image = processed.images[0] if len(processed.images) > 0 else None
if shared.cmd_opts.use_ipex:
shared.sd_model = shared.sd_model.to(dtype=torch.float32)
if unload:
shared.sd_model.cond_stage_model.to(devices.cpu)
shared.sd_model.first_stage_model.to(devices.cpu)
modules/textual_inversion/textual_inversion.py
+1 -13
View File
@@ -3,10 +3,6 @@ import html
import csv
from collections import namedtuple
import torch
try:
import intel_extension_for_pytorch as ipex # pylint: disable=import-error, unused-import
except:
pass
from tqdm import tqdm
import safetensors.torch
import numpy as np
@@ -436,10 +432,7 @@ def train_embedding(id_task, embedding_name, learn_rate, batch_size, gradient_st
shared.log.info("No saved optimizer exists in checkpoint")
if shared.cmd_opts.use_ipex:
#scaler = ipex.cpu.autocast._grad_scaler.GradScaler()
shared.sd_model = shared.sd_model.to(dtype=torch.float32)
shared.sd_model.train()
shared.sd_model, optimizer = ipex.optimize(shared.sd_model, optimizer=optimizer, dtype=devices.dtype)
pass
else:
scaler = torch.cuda.amp.GradScaler()
@@ -533,8 +526,6 @@ def train_embedding(id_task, embedding_name, learn_rate, batch_size, gradient_st
if images_dir is not None and steps_done % create_image_every == 0:
forced_filename = f'{embedding_name}-{steps_done}'
last_saved_image = os.path.join(images_dir, forced_filename)
if shared.cmd_opts.use_ipex:
shared.sd_model = shared.sd_model.to(dtype=devices.dtype)
shared.sd_model.first_stage_model.to(devices.device)
p = processing.StableDiffusionProcessingTxt2Img(
@@ -563,9 +554,6 @@ def train_embedding(id_task, embedding_name, learn_rate, batch_size, gradient_st
processed = processing.process_images(p)
image = processed.images[0] if len(processed.images) > 0 else None
if shared.cmd_opts.use_ipex:
shared.sd_model = shared.sd_model.to(dtype=torch.float32)
if unload:
shared.sd_model.first_stage_model.to(devices.cpu)