Merge pull request #6017 from hitomi/master

Add memory cache for VAE weights

modules/sd_vae.py

@@ -1,5 +1,6 @@
 import torch
 import os
+import collections
 from collections import namedtuple
 from modules import shared, devices, script_callbacks
 from modules.paths import models_path
@@ -30,6 +31,7 @@ base_vae = None
 loaded_vae_file = None
 checkpoint_info = None
 
+checkpoints_loaded = collections.OrderedDict()
 
 def get_base_vae(model):
     if base_vae is not None and checkpoint_info == model.sd_checkpoint_info and model:
@@ -149,13 +151,30 @@ def load_vae(model, vae_file=None):
     global first_load, vae_dict, vae_list, loaded_vae_file
     # save_settings = False
 
+    cache_enabled = shared.opts.sd_vae_checkpoint_cache > 0
+
     if vae_file:
-        assert os.path.isfile(vae_file), f"VAE file doesn't exist: {vae_file}"
+        if cache_enabled and vae_file in checkpoints_loaded:
-        print(f"Loading VAE weights from: {vae_file}")
+            # use vae checkpoint cache
-        store_base_vae(model)
+            print(f"Loading VAE weights [{get_filename(vae_file)}] from cache")
-        vae_ckpt = torch.load(vae_file, map_location=shared.weight_load_location)
+            store_base_vae(model)
-        vae_dict_1 = {k: v for k, v in vae_ckpt["state_dict"].items() if k[0:4] != "loss" and k not in vae_ignore_keys}
+            _load_vae_dict(model, checkpoints_loaded[vae_file])
-        _load_vae_dict(model, vae_dict_1)
+        else:
+            assert os.path.isfile(vae_file), f"VAE file doesn't exist: {vae_file}"
+            print(f"Loading VAE weights from: {vae_file}")
+            store_base_vae(model)
+            vae_ckpt = torch.load(vae_file, map_location=shared.weight_load_location)
+            vae_dict_1 = {k: v for k, v in vae_ckpt["state_dict"].items() if k[0:4] != "loss" and k not in vae_ignore_keys}
+            _load_vae_dict(model, vae_dict_1)
+
+            if cache_enabled:
+                # cache newly loaded vae
+                checkpoints_loaded[vae_file] = vae_dict_1.copy()
+
+        # clean up cache if limit is reached
+        if cache_enabled:
+            while len(checkpoints_loaded) > shared.opts.sd_vae_checkpoint_cache + 1: # we need to count the current model
+                checkpoints_loaded.popitem(last=False)  # LRU
 
         # If vae used is not in dict, update it
         # It will be removed on refresh though

modules/shared.py

@@ -356,6 +356,7 @@ options_templates.update(options_section(('training', "Training"), {
 options_templates.update(options_section(('sd', "Stable Diffusion"), {
     "sd_model_checkpoint": OptionInfo(None, "Stable Diffusion checkpoint", gr.Dropdown, lambda: {"choices": list_checkpoint_tiles()}, refresh=refresh_checkpoints),
     "sd_checkpoint_cache": OptionInfo(0, "Checkpoints to cache in RAM", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}),
+    "sd_vae_checkpoint_cache": OptionInfo(0, "VAE Checkpoints to cache in RAM", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}),
     "sd_vae": OptionInfo("auto", "SD VAE", gr.Dropdown, lambda: {"choices": sd_vae.vae_list}, refresh=sd_vae.refresh_vae_list),
     "sd_vae_as_default": OptionInfo(False, "Ignore selected VAE for stable diffusion checkpoints that have their own .vae.pt next to them"),
     "sd_hypernetwork": OptionInfo("None", "Hypernetwork", gr.Dropdown, lambda: {"choices": ["None"] + [x for x in hypernetworks.keys()]}, refresh=reload_hypernetworks),