Merge pull request #9256 from papuSpartan/tomesd

Integrate optional speed and memory improvements by token merging (via dbolya/tomesd)
This commit is contained in:
AUTOMATIC1111 2023-05-14 08:21:02 +03:00 committed by GitHub
commit 7f6ef764b9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
5 changed files with 63 additions and 1 deletions

View File

@ -308,8 +308,10 @@ infotext_to_setting_name_mapping = [
('UniPC skip type', 'uni_pc_skip_type'), ('UniPC skip type', 'uni_pc_skip_type'),
('UniPC order', 'uni_pc_order'), ('UniPC order', 'uni_pc_order'),
('UniPC lower order final', 'uni_pc_lower_order_final'), ('UniPC lower order final', 'uni_pc_lower_order_final'),
('Token merging ratio', 'token_merging_ratio'),
('Token merging ratio hr', 'token_merging_ratio_hr'),
('RNG', 'randn_source'), ('RNG', 'randn_source'),
('NGMS', 's_min_uncond'), ('NGMS', 's_min_uncond')
] ]

View File

@ -29,6 +29,13 @@ from ldm.models.diffusion.ddpm import LatentDepth2ImageDiffusion
from einops import repeat, rearrange from einops import repeat, rearrange
from blendmodes.blend import blendLayers, BlendType from blendmodes.blend import blendLayers, BlendType
import tomesd
# add a logger for the processing module
logger = logging.getLogger(__name__)
# manually set output level here since there is no option to do so yet through launch options
# logging.basicConfig(level=logging.DEBUG, format='%(asctime)s %(levelname)s %(name)s %(message)s')
# some of those options should not be changed at all because they would break the model, so I removed them from options. # some of those options should not be changed at all because they would break the model, so I removed them from options.
opt_C = 4 opt_C = 4
@ -471,6 +478,7 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments=None, iter
index = position_in_batch + iteration * p.batch_size index = position_in_batch + iteration * p.batch_size
clip_skip = getattr(p, 'clip_skip', opts.CLIP_stop_at_last_layers) clip_skip = getattr(p, 'clip_skip', opts.CLIP_stop_at_last_layers)
enable_hr = getattr(p, 'enable_hr', False)
generation_params = { generation_params = {
"Steps": p.steps, "Steps": p.steps,
@ -489,6 +497,8 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments=None, iter
"Conditional mask weight": getattr(p, "inpainting_mask_weight", shared.opts.inpainting_mask_weight) if p.is_using_inpainting_conditioning else None, "Conditional mask weight": getattr(p, "inpainting_mask_weight", shared.opts.inpainting_mask_weight) if p.is_using_inpainting_conditioning else None,
"Clip skip": None if clip_skip <= 1 else clip_skip, "Clip skip": None if clip_skip <= 1 else clip_skip,
"ENSD": None if opts.eta_noise_seed_delta == 0 else opts.eta_noise_seed_delta, "ENSD": None if opts.eta_noise_seed_delta == 0 else opts.eta_noise_seed_delta,
"Token merging ratio": None if opts.token_merging_ratio == 0 else opts.token_merging_ratio,
"Token merging ratio hr": None if not enable_hr or opts.token_merging_ratio_hr == 0 else opts.token_merging_ratio_hr,
"Init image hash": getattr(p, 'init_img_hash', None), "Init image hash": getattr(p, 'init_img_hash', None),
"RNG": opts.randn_source if opts.randn_source != "GPU" else None, "RNG": opts.randn_source if opts.randn_source != "GPU" else None,
"NGMS": None if p.s_min_uncond == 0 else p.s_min_uncond, "NGMS": None if p.s_min_uncond == 0 else p.s_min_uncond,
@ -522,9 +532,18 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
if k == 'sd_vae': if k == 'sd_vae':
sd_vae.reload_vae_weights() sd_vae.reload_vae_weights()
if opts.token_merging_ratio > 0:
sd_models.apply_token_merging(sd_model=p.sd_model, hr=False)
logger.debug(f"Token merging applied to first pass. Ratio: '{opts.token_merging_ratio}'")
res = process_images_inner(p) res = process_images_inner(p)
finally: finally:
# undo model optimizations made by tomesd
if opts.token_merging_ratio > 0:
tomesd.remove_patch(p.sd_model)
logger.debug('Token merging model optimizations removed')
# restore opts to original state # restore opts to original state
if p.override_settings_restore_afterwards: if p.override_settings_restore_afterwards:
for k, v in stored_opts.items(): for k, v in stored_opts.items():
@ -977,8 +996,22 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing):
x = None x = None
devices.torch_gc() devices.torch_gc()
# apply token merging optimizations from tomesd for high-res pass
if opts.token_merging_ratio_hr > 0:
# in case the user has used separate merge ratios
if opts.token_merging_ratio > 0:
tomesd.remove_patch(self.sd_model)
logger.debug('Adjusting token merging ratio for high-res pass')
sd_models.apply_token_merging(sd_model=self.sd_model, hr=True)
logger.debug(f"Applied token merging for high-res pass. Ratio: '{opts.token_merging_ratio_hr}'")
samples = self.sampler.sample_img2img(self, samples, noise, conditioning, unconditional_conditioning, steps=self.hr_second_pass_steps or 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)
if opts.token_merging_ratio_hr > 0 or opts.token_merging_ratio > 0:
tomesd.remove_patch(self.sd_model)
logger.debug('Removed token merging optimizations from model')
self.is_hr_pass = False self.is_hr_pass = False
return samples return samples

View File

@ -17,6 +17,7 @@ from ldm.util import instantiate_from_config
from modules import paths, shared, modelloader, devices, script_callbacks, sd_vae, sd_disable_initialization, errors, hashes, sd_models_config from modules import paths, shared, modelloader, devices, script_callbacks, sd_vae, sd_disable_initialization, errors, hashes, sd_models_config
from modules.sd_hijack_inpainting import do_inpainting_hijack from modules.sd_hijack_inpainting import do_inpainting_hijack
from modules.timer import Timer from modules.timer import Timer
import tomesd
model_dir = "Stable-diffusion" model_dir = "Stable-diffusion"
model_path = os.path.abspath(os.path.join(paths.models_path, model_dir)) model_path = os.path.abspath(os.path.join(paths.models_path, model_dir))
@ -578,3 +579,25 @@ def unload_model_weights(sd_model=None, info=None):
print(f"Unloaded weights {timer.summary()}.") print(f"Unloaded weights {timer.summary()}.")
return sd_model return sd_model
def apply_token_merging(sd_model, hr: bool):
"""
Applies speed and memory optimizations from tomesd.
Args:
hr (bool): True if called in the context of a high-res pass
"""
ratio = shared.opts.token_merging_ratio
if hr:
ratio = shared.opts.token_merging_ratio_hr
tomesd.apply_patch(
sd_model,
ratio=ratio,
use_rand=False, # can cause issues with some samplers
merge_attn=True,
merge_crossattn=False,
merge_mlp=False
)

View File

@ -350,6 +350,8 @@ options_templates.update(options_section(('sd', "Stable Diffusion"), {
"CLIP_stop_at_last_layers": OptionInfo(1, "Clip skip", gr.Slider, {"minimum": 1, "maximum": 12, "step": 1}), "CLIP_stop_at_last_layers": OptionInfo(1, "Clip skip", gr.Slider, {"minimum": 1, "maximum": 12, "step": 1}),
"upcast_attn": OptionInfo(False, "Upcast cross attention layer to float32"), "upcast_attn": OptionInfo(False, "Upcast cross attention layer to float32"),
"randn_source": OptionInfo("GPU", "Random number generator source. Changes seeds drastically. Use CPU to produce the same picture across different vidocard vendors.", gr.Radio, {"choices": ["GPU", "CPU"]}), "randn_source": OptionInfo("GPU", "Random number generator source. Changes seeds drastically. Use CPU to produce the same picture across different vidocard vendors.", gr.Radio, {"choices": ["GPU", "CPU"]}),
"token_merging_ratio_hr": OptionInfo(0, "Merging Ratio (high-res pass)", gr.Slider, {"minimum": 0, "maximum": 0.9, "step": 0.1}),
"token_merging_ratio": OptionInfo(0, "Merging Ratio", gr.Slider, {"minimum": 0, "maximum": 0.9, "step": 0.1})
})) }))
options_templates.update(options_section(('compatibility', "Compatibility"), { options_templates.update(options_section(('compatibility', "Compatibility"), {
@ -458,6 +460,7 @@ options_templates.update(options_section((None, "Hidden options"), {
"sd_checkpoint_hash": OptionInfo("", "SHA256 hash of the current checkpoint"), "sd_checkpoint_hash": OptionInfo("", "SHA256 hash of the current checkpoint"),
})) }))
options_templates.update() options_templates.update()

View File

@ -26,3 +26,4 @@ torchsde==0.2.5
safetensors==0.3.1 safetensors==0.3.1
httpcore<=0.15 httpcore<=0.15
fastapi==0.94.0 fastapi==0.94.0
tomesd>=0.1.2