3 files changed, 41 insertions, 39 deletions

diff --git a/train_ti.py b/train_ti.py
index 928b721..8631892 100644
--- a/train_ti.py
+++ b/train_ti.py
@@ -397,7 +397,7 @@ def parse_args():
     )
     parser.add_argument(
         "--emb_decay_factor",
-        default=1,
+        default=0,
         type=float,
         help="Embedding decay factor."
     )
@@ -532,13 +532,17 @@ class Checkpointer(CheckpointerBase):
 
     @torch.no_grad()
     def save_samples(self, step):
+        unet = self.accelerator.unwrap_model(self.unet)
         text_encoder = self.accelerator.unwrap_model(self.text_encoder)
 
         ema_context = self.ema_embeddings.apply_temporary(
             text_encoder.text_model.embeddings.temp_token_embedding.parameters()) if self.ema_embeddings is not None else nullcontext()
 
         with ema_context:
-            orig_dtype = text_encoder.dtype
+            orig_unet_dtype = unet.dtype
+            orig_text_encoder_dtype = text_encoder.dtype
+
+            unet.to(dtype=self.weight_dtype)
             text_encoder.to(dtype=self.weight_dtype)
 
             pipeline = VlpnStableDiffusion(
@@ -552,7 +556,8 @@ class Checkpointer(CheckpointerBase):
 
             super().save_samples(pipeline, step)
 
-            text_encoder.to(dtype=orig_dtype)
+            unet.to(dtype=orig_unet_dtype)
+            text_encoder.to(dtype=orig_text_encoder_dtype)
 
         del text_encoder
         del pipeline
@@ -742,20 +747,17 @@ def main():
             warmup_epochs=args.lr_warmup_epochs,
         )
 
-    text_encoder, optimizer, train_dataloader, val_dataloader, lr_scheduler = accelerator.prepare(
-        text_encoder, optimizer, train_dataloader, val_dataloader, lr_scheduler
+    unet, text_encoder, optimizer, train_dataloader, val_dataloader, lr_scheduler = accelerator.prepare(
+        unet, text_encoder, optimizer, train_dataloader, val_dataloader, lr_scheduler
     )
 
     vae.to(accelerator.device, dtype=weight_dtype)
-    unet.to(accelerator.device, dtype=weight_dtype)
 
     if args.use_ema:
         ema_embeddings.to(accelerator.device)
 
     if args.gradient_checkpointing:
         unet.train()
-    else:
-        unet.eval()
 
     @contextmanager
     def on_train(epoch: int):
@@ -780,10 +782,11 @@ def main():
 
     @torch.no_grad()
     def on_after_optimize(lr: float):
-        text_encoder.text_model.embeddings.normalize(
-            args.emb_decay_target,
-            min(1.0, max(0.0, args.emb_decay_factor * ((lr - args.emb_decay_start) / (args.learning_rate - args.emb_decay_start))))
-        )
+        if args.emb_decay_factor != 0:
+            text_encoder.text_model.embeddings.normalize(
+                args.emb_decay_target,
+                min(1.0, max(0.0, args.emb_decay_factor * ((lr - args.emb_decay_start) / (args.learning_rate - args.emb_decay_start))))
+            )
 
         if args.use_ema:
             ema_embeddings.step(text_encoder.text_model.embeddings.temp_token_embedding.parameters())
diff --git a/training/common.py b/training/common.py
index 8083137..5d1e3f9 100644
--- a/training/common.py
+++ b/training/common.py
@@ -316,30 +316,29 @@ def train_loop(
             cur_loss_val = AverageMeter()
             cur_acc_val = AverageMeter()
 
-            with torch.inference_mode():
-                with on_eval():
-                    for step, batch in enumerate(val_dataloader):
-                        loss, acc, bsz = loss_step(step, batch, True)
+            with torch.inference_mode(), on_eval():
+                for step, batch in enumerate(val_dataloader):
+                    loss, acc, bsz = loss_step(step, batch, True)
 
-                        loss = loss.detach_()
-                        acc = acc.detach_()
+                    loss = loss.detach_()
+                    acc = acc.detach_()
 
-                        cur_loss_val.update(loss, bsz)
-                        cur_acc_val.update(acc, bsz)
+                    cur_loss_val.update(loss, bsz)
+                    cur_acc_val.update(acc, bsz)
 
-                        avg_loss_val.update(loss, bsz)
-                        avg_acc_val.update(acc, bsz)
+                    avg_loss_val.update(loss, bsz)
+                    avg_acc_val.update(acc, bsz)
 
-                        local_progress_bar.update(1)
-                        global_progress_bar.update(1)
+                    local_progress_bar.update(1)
+                    global_progress_bar.update(1)
 
-                        logs = {
-                            "val/loss": avg_loss_val.avg.item(),
-                            "val/acc": avg_acc_val.avg.item(),
-                            "val/cur_loss": loss.item(),
-                            "val/cur_acc": acc.item(),
-                        }
-                        local_progress_bar.set_postfix(**logs)
+                    logs = {
+                        "val/loss": avg_loss_val.avg.item(),
+                        "val/acc": avg_acc_val.avg.item(),
+                        "val/cur_loss": loss.item(),
+                        "val/cur_acc": acc.item(),
+                    }
+                    local_progress_bar.set_postfix(**logs)
 
             logs["val/cur_loss"] = cur_loss_val.avg.item()
             logs["val/cur_acc"] = cur_acc_val.avg.item()
diff --git a/training/util.py b/training/util.py
index 1008021..781cf04 100644
--- a/training/util.py
+++ b/training/util.py
@@ -134,11 +134,11 @@ class EMAModel:
     def __init__(
         self,
         parameters: Iterable[torch.nn.Parameter],
-        update_after_step=0,
-        inv_gamma=1.0,
-        power=2 / 3,
-        min_value=0.0,
-        max_value=0.9999,
+        update_after_step: int = 0,
+        inv_gamma: float = 1.0,
+        power: float = 2 / 3,
+        min_value: float = 0.0,
+        max_value: float = 0.9999,
     ):
         """
         @crowsonkb's notes on EMA Warmup:
@@ -165,7 +165,7 @@ class EMAModel:
         self.decay = 0.0
         self.optimization_step = 0
 
-    def get_decay(self, optimization_step):
+    def get_decay(self, optimization_step: int):
         """
         Compute the decay factor for the exponential moving average.
         """
@@ -276,5 +276,5 @@ class EMAModel:
             self.copy_to(parameters)
             yield
         finally:
-            for s_param, param in zip(original_params, parameters):
-                param.data.copy_(s_param.data)
+            for o_param, param in zip(original_params, parameters):
+                param.data.copy_(o_param.data)