fix vae type error during training sdxl

library/sdxl_train_util.py

@@ -17,7 +17,6 @@ TOKENIZER2_PATH = "laion/CLIP-ViT-bigG-14-laion2B-39B-b160k"
 
 
 def load_target_model(args, accelerator, model_version: str, weight_dtype):
-    # load models for each process
     model_dtype = match_mixed_precision(args, weight_dtype)  # prepare fp16/bf16
     for pi in range(accelerator.state.num_processes):
         if pi == accelerator.state.local_process_index:

library/train_util.py

@@ -4042,28 +4042,23 @@ def _load_target_model(args: argparse.Namespace, weight_dtype, device="cpu", une
 
 
 def load_target_model(args, weight_dtype, accelerator, unet_use_linear_projection_in_v2=False):
-    # load models for each process
     for pi in range(accelerator.state.num_processes):
         if pi == accelerator.state.local_process_index:
             print(f"loading model for process {accelerator.state.local_process_index}/{accelerator.state.num_processes}")
-
             text_encoder, vae, unet, load_stable_diffusion_format = _load_target_model(
                 args,
                 weight_dtype,
                 accelerator.device if args.lowram else "cpu",
                 unet_use_linear_projection_in_v2=unet_use_linear_projection_in_v2,
             )
-
             # work on low-ram device
             if args.lowram:
                 text_encoder.to(accelerator.device)
                 unet.to(accelerator.device)
                 vae.to(accelerator.device)
-
             gc.collect()
             torch.cuda.empty_cache()
         accelerator.wait_for_everyone()
-
     return text_encoder, vae, unet, load_stable_diffusion_format
 
 

sdxl_train.py

@@ -392,23 +392,20 @@ def train(args):
     if args.deepspeed:
         # Wrapping model for DeepSpeed
         class DeepSpeedModel(torch.nn.Module): 
-            def __init__(self, unet, text_encoder, vae) -> None:
+            def __init__(self, unet, text_encoder) -> None:
                 super().__init__()
                 self.unet = unet
                 self.text_encoders = self.text_encoder = torch.nn.ModuleList(text_encoder)
-                self.vae = vae
                 
             def get_models(self):
-                return self.unet, self.text_encoders, self.vae
+                return self.unet, self.text_encoders
         text_encoders = [text_encoder1, text_encoder2]
-        unet.to(accelerator.device, dtype=weight_dtype)
+        ds_model = DeepSpeedModel(unet, text_encoders)
-        [t_enc.to(accelerator.device, dtype=weight_dtype) for t_enc in text_encoders]
-        ds_model = DeepSpeedModel(unet, text_encoders, vae)
         ds_model, optimizer, train_dataloader, lr_scheduler = accelerator.prepare(ds_model, optimizer, train_dataloader, lr_scheduler)
         # Now, ds_model is an instance of DeepSpeedEngine. 
-        unet, text_encoders, vae = ds_model.get_models() # for compatiblility
+        unet, text_encoders = ds_model.get_models() # for compatiblility
-        vae.to(vae_dtype) # to avoid explicitly half-vae
+        text_encoder1, text_encoder2 = text_encoder = text_encoders
-        text_encoder1, text_encoder2 = text_encoders[0], text_encoders[1]
+        training_models = [unet, text_encoder1, text_encoder2]
     else: # acceleratorがなんかよろしくやってくれるらしい
         if train_unet:
             unet = accelerator.prepare(unet)
@@ -493,10 +490,10 @@ def train(args):
         for step, batch in enumerate(train_dataloader):
             current_step.value = global_step
             with accelerator.accumulate(*training_models):
-                if "latents" in batch and batch["latents"] is not None:
+                with torch.no_grad(): # why this block differ within train_network.py?
-                    latents = batch["latents"].to(accelerator.device).to(dtype=weight_dtype)
+                    if "latents" in batch and batch["latents"] is not None:
-                else:
+                        latents = batch["latents"].to(accelerator.device).to(dtype=weight_dtype)
-                    with torch.no_grad():
+                    else:
                         # latentに変換
                         latents = vae.encode(batch["images"].to(vae_dtype)).latent_dist.sample().to(weight_dtype)
 
@@ -504,7 +501,7 @@ def train(args):
                         if torch.any(torch.isnan(latents)):
                             accelerator.print("NaN found in latents, replacing with zeros")
                             latents = torch.nan_to_num(latents, 0, out=latents)
-                latents = latents * sdxl_model_util.VAE_SCALE_FACTOR
+                    latents = latents * sdxl_model_util.VAE_SCALE_FACTOR
 
                 if "text_encoder_outputs1_list" not in batch or batch["text_encoder_outputs1_list"] is None:
                     input_ids1 = batch["input_ids"]