michael/stable-diffusion-webui
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

- do not use ckpt cache, if disabled

Browse source 
- cache model after is has been loaded from file
This commit is contained in:
cluder 2022-11-09 04:54:21 +01:00
parent 2f47724b73
commit 3b51d239ac
1 changed files with 17 additions and 10 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

27
modules/sd_models.py
View file

@ -163,13 +163,21 @@ def load_model_weights(model, checkpoint_info, vae_file="auto"):
checkpoint_file = checkpoint_info.filename
sd_model_hash = checkpoint_info.hash
if shared.opts.sd_checkpoint_cache > 0 and hasattr(model, "sd_checkpoint_info"):
cache_enabled = shared.opts.sd_checkpoint_cache > 0
if cache_enabled:
sd_vae.restore_base_vae(model)
checkpoints_loaded[model.sd_checkpoint_info] = model.state_dict().copy()
vae_file = sd_vae.resolve_vae(checkpoint_file, vae_file=vae_file)
if checkpoint_info not in checkpoints_loaded:
if cache_enabled and checkpoint_info in checkpoints_loaded:
# use checkpoint cache
vae_name = sd_vae.get_filename(vae_file) if vae_file else None
vae_message = f" with {vae_name} VAE" if vae_name else ""
print(f"Loading weights [{sd_model_hash}]{vae_message} from cache")
model.load_state_dict(checkpoints_loaded[checkpoint_info])
else:
# load from file
print(f"Loading weights [{sd_model_hash}] from {checkpoint_file}")
pl_sd = torch.load(checkpoint_file, map_location=shared.weight_load_location)
@ -180,6 +188,10 @@ def load_model_weights(model, checkpoint_info, vae_file="auto"):
del pl_sd
model.load_state_dict(sd, strict=False)
del sd
if cache_enabled:
# cache newly loaded model
checkpoints_loaded[checkpoint_info] = model.state_dict().copy()
if shared.cmd_opts.opt_channelslast:
model.to(memory_format=torch.channels_last)
@ -199,13 +211,8 @@ def load_model_weights(model, checkpoint_info, vae_file="auto"):
model.first_stage_model.to(devices.dtype_vae)
else:
vae_name = sd_vae.get_filename(vae_file) if vae_file else None
vae_message = f" with {vae_name} VAE" if vae_name else ""
print(f"Loading weights [{sd_model_hash}]{vae_message} from cache")
model.load_state_dict(checkpoints_loaded[checkpoint_info])
if shared.opts.sd_checkpoint_cache > 0:
# clean up cache if limit is reached
if cache_enabled:
while len(checkpoints_loaded) > shared.opts.sd_checkpoint_cache:
checkpoints_loaded.popitem(last=False) # LRU