From 8ab999e501ce56fc333a7b85cd08a4f70e437a88 Mon Sep 17 00:00:00 2001
From: Vincent <phamvinh257@gmail.com>
Date: Thu, 11 Jan 2024 23:33:53 +0700
Subject: [PATCH] support compile

---
 examples/dreambooth/train_dreambooth.py | 32 +++++++++++++++----------
 1 file changed, 19 insertions(+), 13 deletions(-)

diff --git a/examples/dreambooth/train_dreambooth.py b/examples/dreambooth/train_dreambooth.py
index f652b1e79b..8b02db9ba9 100644
--- a/examples/dreambooth/train_dreambooth.py
+++ b/examples/dreambooth/train_dreambooth.py
@@ -55,6 +55,7 @@ from diffusers.optimization import get_scheduler
 from diffusers.training_utils import compute_snr
 from diffusers.utils import check_min_version, is_wandb_available
 from diffusers.utils.import_utils import is_xformers_available
+from diffusers.utils.torch_utils import is_compiled_module
 
 
 if is_wandb_available():
@@ -106,6 +107,10 @@ DreamBooth for the text encoder was enabled: {train_text_encoder}.
     with open(os.path.join(repo_folder, "README.md"), "w") as f:
         f.write(yaml + model_card)
 
+def unwrap_model(accelerator, model):
+    model = accelerator.unwrap_model(model)
+    model = model._orig_mod if is_compiled_module(model) else model
+    return model
 
 def log_validation(
     text_encoder,
@@ -130,14 +135,14 @@ def log_validation(
         pipeline_args["vae"] = vae
 
     if text_encoder is not None:
-        text_encoder = accelerator.unwrap_model(text_encoder)
+        text_encoder = unwrap_model(accelerator, text_encoder)
 
     # create pipeline (note: unet and vae are loaded again in float32)
     pipeline = DiffusionPipeline.from_pretrained(
         args.pretrained_model_name_or_path,
         tokenizer=tokenizer,
         text_encoder=text_encoder,
-        unet=accelerator.unwrap_model(unet),
+        unet=unwrap_model(accelerator, unet),
         revision=args.revision,
         variant=args.variant,
         torch_dtype=weight_dtype,
@@ -794,6 +799,7 @@ def encode_prompt(text_encoder, input_ids, attention_mask, text_encoder_use_atte
     prompt_embeds = text_encoder(
         text_input_ids,
         attention_mask=attention_mask,
+        return_dict=False,
     )
     prompt_embeds = prompt_embeds[0]
 
@@ -935,7 +941,7 @@ def main(args):
     def save_model_hook(models, weights, output_dir):
         if accelerator.is_main_process:
             for model in models:
-                sub_dir = "unet" if isinstance(model, type(accelerator.unwrap_model(unet))) else "text_encoder"
+                sub_dir = "unet" if isinstance(model, type(unwrap_model(accelerator, unet))) else "text_encoder"
                 model.save_pretrained(os.path.join(output_dir, sub_dir))
 
                 # make sure to pop weight so that corresponding model is not saved again
@@ -946,7 +952,7 @@ def main(args):
             # pop models so that they are not loaded again
             model = models.pop()
 
-            if isinstance(model, type(accelerator.unwrap_model(text_encoder))):
+            if isinstance(model, type(unwrap_model(accelerator, text_encoder))):
                 # load transformers style into model
                 load_model = text_encoder_cls.from_pretrained(input_dir, subfolder="text_encoder")
                 model.config = load_model.config
@@ -991,14 +997,14 @@ def main(args):
         " doing mixed precision training. copy of the weights should still be float32."
     )
 
-    if accelerator.unwrap_model(unet).dtype != torch.float32:
+    if unwrap_model(accelerator, unet).dtype != torch.float32:
         raise ValueError(
-            f"Unet loaded as datatype {accelerator.unwrap_model(unet).dtype}. {low_precision_error_string}"
+            f"Unet loaded as datatype {unwrap_model(accelerator, unet).dtype}. {low_precision_error_string}"
         )
 
-    if args.train_text_encoder and accelerator.unwrap_model(text_encoder).dtype != torch.float32:
+    if args.train_text_encoder and unwrap_model(accelerator, text_encoder).dtype != torch.float32:
         raise ValueError(
-            f"Text encoder loaded as datatype {accelerator.unwrap_model(text_encoder).dtype}."
+            f"Text encoder loaded as datatype {unwrap_model(accelerator, text_encoder).dtype}."
             f" {low_precision_error_string}"
         )
 
@@ -1246,7 +1252,7 @@ def main(args):
                         text_encoder_use_attention_mask=args.text_encoder_use_attention_mask,
                     )
 
-                if accelerator.unwrap_model(unet).config.in_channels == channels * 2:
+                if unwrap_model(accelerator, unet).config.in_channels == channels * 2:
                     noisy_model_input = torch.cat([noisy_model_input, noisy_model_input], dim=1)
 
                 if args.class_labels_conditioning == "timesteps":
@@ -1256,8 +1262,8 @@ def main(args):
 
                 # Predict the noise residual
                 model_pred = unet(
-                    noisy_model_input, timesteps, encoder_hidden_states, class_labels=class_labels
-                ).sample
+                    noisy_model_input, timesteps, encoder_hidden_states, class_labels=class_labels, return_dict=False
+                )[0]
 
                 if model_pred.shape[1] == 6:
                     model_pred, _ = torch.chunk(model_pred, 2, dim=1)
@@ -1375,14 +1381,14 @@ def main(args):
         pipeline_args = {}
 
         if text_encoder is not None:
-            pipeline_args["text_encoder"] = accelerator.unwrap_model(text_encoder)
+            pipeline_args["text_encoder"] = unwrap_model(accelerator, text_encoder)
 
         if args.skip_save_text_encoder:
             pipeline_args["text_encoder"] = None
 
         pipeline = DiffusionPipeline.from_pretrained(
             args.pretrained_model_name_or_path,
-            unet=accelerator.unwrap_model(unet),
+            unet=unwrap_model(accelerator, unet),
             revision=args.revision,
             variant=args.variant,
             **pipeline_args,