Update the training examples (#102)

* New unet, gradient accumulation

* Save every n epochs

* Remove find_unused_params, hooray!

* Update examples

* Switch to DDPM completely

examples/README.md

@@ -5,18 +5,17 @@
 The command to train a DDPM UNet model on the Oxford Flowers dataset:
 
 ```bash
-python -m torch.distributed.launch \
+accelerate launch train_unconditional.py \
-  --nproc_per_node 4 \
-  train_unconditional.py \
   --dataset="huggan/flowers-102-categories" \
   --resolution=64 \
-  --output_dir="flowers-ddpm" \
+  --output_dir="ddpm-ema-flowers-64" \
-  --batch_size=16 \
+  --train_batch_size=16 \
   --num_epochs=100 \
   --gradient_accumulation_steps=1 \
-  --lr=1e-4 \
+  --learning_rate=1e-4 \
-  --warmup_steps=500 \
+  --lr_warmup_steps=500 \
-  --mixed_precision=no
+  --mixed_precision=no \
+  --push_to_hub
 ```
 
 A full training run takes 2 hours on 4xV100 GPUs.
@@ -29,18 +28,17 @@ A full training run takes 2 hours on 4xV100 GPUs.
 The command to train a DDPM UNet model on the Pokemon dataset:
 
 ```bash
-python -m torch.distributed.launch \
+accelerate launch train_unconditional.py \
-  --nproc_per_node 4 \
-  train_unconditional.py \
   --dataset="huggan/pokemon" \
   --resolution=64 \
-  --output_dir="pokemon-ddpm" \
+  --output_dir="ddpm-ema-pokemon-64" \
-  --batch_size=16 \
+  --train_batch_size=16 \
   --num_epochs=100 \
   --gradient_accumulation_steps=1 \
-  --lr=1e-4 \
+  --learning_rate=1e-4 \
-  --warmup_steps=500 \
+  --lr_warmup_steps=500 \
-  --mixed_precision=no
+  --mixed_precision=no \
+  --push_to_hub
 ```
 
 A full training run takes 2 hours on 4xV100 GPUs.

examples/train_unconditional.py

@@ -4,10 +4,10 @@ import os
 import torch
 import torch.nn.functional as F
 
-from accelerate import Accelerator, DistributedDataParallelKwargs
+from accelerate import Accelerator
 from accelerate.logging import get_logger
 from datasets import load_dataset
-from diffusers import DDIMPipeline, DDIMScheduler, UNetModel
+from diffusers import DDPMPipeline, DDPMScheduler, UNetUnconditionalModel
 from diffusers.hub_utils import init_git_repo, push_to_hub
 from diffusers.optimization import get_scheduler
 from diffusers.training_utils import EMAModel
@@ -27,25 +27,37 @@ logger = get_logger(__name__)
 
 
 def main(args):
-    ddp_unused_params = DistributedDataParallelKwargs(find_unused_parameters=True)
     logging_dir = os.path.join(args.output_dir, args.logging_dir)
     accelerator = Accelerator(
         mixed_precision=args.mixed_precision,
         log_with="tensorboard",
         logging_dir=logging_dir,
-        kwargs_handlers=[ddp_unused_params],
     )
 
-    model = UNetModel(
+    model = UNetUnconditionalModel(
-        attn_resolutions=(16,),
+        image_size=args.resolution,
-        ch=128,
+        in_channels=3,
-        ch_mult=(1, 2, 4, 8),
+        out_channels=3,
-        dropout=0.0,
         num_res_blocks=2,
-        resamp_with_conv=True,
+        block_channels=(128, 128, 256, 256, 512, 512),
-        resolution=args.resolution,
+        down_blocks=(
+            "UNetResDownBlock2D",
+            "UNetResDownBlock2D",
+            "UNetResDownBlock2D",
+            "UNetResDownBlock2D",
+            "UNetResAttnDownBlock2D",
+            "UNetResDownBlock2D",
+        ),
+        up_blocks=(
+            "UNetResUpBlock2D",
+            "UNetResAttnUpBlock2D",
+            "UNetResUpBlock2D",
+            "UNetResUpBlock2D",
+            "UNetResUpBlock2D",
+            "UNetResUpBlock2D",
+        ),
     )
-    noise_scheduler = DDIMScheduler(timesteps=1000, tensor_format="pt")
+    noise_scheduler = DDPMScheduler(num_train_timesteps=1000, tensor_format="pt")
     optimizer = torch.optim.AdamW(
         model.parameters(),
         lr=args.learning_rate,
@@ -92,19 +104,6 @@ def main(args):
         run = os.path.split(__file__)[-1].split(".")[0]
         accelerator.init_trackers(run)
 
-    # Train!
-    is_distributed = torch.distributed.is_available() and torch.distributed.is_initialized()
-    world_size = torch.distributed.get_world_size() if is_distributed else 1
-    total_train_batch_size = args.train_batch_size * args.gradient_accumulation_steps * world_size
-    max_steps = len(train_dataloader) // args.gradient_accumulation_steps * args.num_epochs
-    logger.info("***** Running training *****")
-    logger.info(f"  Num examples = {len(train_dataloader.dataset)}")
-    logger.info(f"  Num Epochs = {args.num_epochs}")
-    logger.info(f"  Instantaneous batch size per device = {args.train_batch_size}")
-    logger.info(f"  Total train batch size (w. parallel, distributed & accumulation) = {total_train_batch_size}")
-    logger.info(f"  Gradient Accumulation steps = {args.gradient_accumulation_steps}")
-    logger.info(f"  Total optimization steps = {max_steps}")
-
     global_step = 0
     for epoch in range(args.num_epochs):
         model.train()
@@ -112,45 +111,37 @@ def main(args):
         progress_bar.set_description(f"Epoch {epoch}")
         for step, batch in enumerate(train_dataloader):
             clean_images = batch["input"]
-            noise_samples = torch.randn(clean_images.shape).to(clean_images.device)
+            # Sample noise that we'll add to the images
+            noise = torch.randn(clean_images.shape).to(clean_images.device)
             bsz = clean_images.shape[0]
-            timesteps = torch.randint(0, noise_scheduler.timesteps, (bsz,), device=clean_images.device).long()
+            # Sample a random timestep for each image
+            timesteps = torch.randint(
+                0, noise_scheduler.num_train_timesteps, (bsz,), device=clean_images.device
+            ).long()
 
-            # add noise onto the clean images according to the noise magnitude at each timestep
+            # Add noise to the clean images according to the noise magnitude at each timestep
             # (this is the forward diffusion process)
-            noisy_images = noise_scheduler.add_noise(clean_images, noise_samples, timesteps)
+            noisy_images = noise_scheduler.add_noise(clean_images, noise, timesteps)
 
-            if step % args.gradient_accumulation_steps != 0:
+            with accelerator.accumulate(model):
-                with accelerator.no_sync(model):
+                # Predict the noise residual
-                    output = model(noisy_images, timesteps)
+                noise_pred = model(noisy_images, timesteps)["sample"]
-                    # predict the noise residual
+                loss = F.mse_loss(noise_pred, noise)
-                    loss = F.mse_loss(output, noise_samples)
-                    loss = loss / args.gradient_accumulation_steps
-                    accelerator.backward(loss)
-            else:
-                output = model(noisy_images, timesteps)
-                # predict the noise residual
-                loss = F.mse_loss(output, noise_samples)
-                loss = loss / args.gradient_accumulation_steps
                 accelerator.backward(loss)
-                torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
+
+                accelerator.clip_grad_norm_(model.parameters(), 1.0)
                 optimizer.step()
                 lr_scheduler.step()
-                ema_model.step(model)
+                if args.use_ema:
+                    ema_model.step(model)
                 optimizer.zero_grad()
+
             progress_bar.update(1)
-            progress_bar.set_postfix(
+            logs = {"loss": loss.detach().item(), "lr": lr_scheduler.get_last_lr()[0], "step": global_step}
-                loss=loss.detach().item(), lr=optimizer.param_groups[0]["lr"], ema_decay=ema_model.decay
+            if args.use_ema:
-            )
+                logs["ema_decay"] = ema_model.decay
-            accelerator.log(
+            progress_bar.set_postfix(**logs)
-                {
+            accelerator.log(logs, step=global_step)
-                    "train_loss": loss.detach().item(),
-                    "epoch": epoch,
-                    "ema_decay": ema_model.decay,
-                    "step": global_step,
-                },
-                step=global_step,
-            )
             global_step += 1
         progress_bar.close()
 
@@ -159,14 +150,14 @@ def main(args):
         # Generate a sample image for visual inspection
         if accelerator.is_main_process:
             with torch.no_grad():
-                pipeline = DDIMPipeline(
+                pipeline = DDPMPipeline(
-                    unet=accelerator.unwrap_model(ema_model.averaged_model),
+                    unet=accelerator.unwrap_model(ema_model.averaged_model if args.use_ema else model),
-                    noise_scheduler=noise_scheduler,
+                    scheduler=noise_scheduler,
                 )
 
                 generator = torch.manual_seed(0)
                 # run pipeline in inference (sample random noise and denoise)
-                images = pipeline(generator=generator, batch_size=args.eval_batch_size, num_inference_steps=50)
+                images = pipeline(generator=generator, batch_size=args.eval_batch_size)
 
             # denormalize the images and save to tensorboard
             images_processed = (images.cpu() + 1.0) * 127.5
@@ -174,11 +165,12 @@ def main(args):
 
             accelerator.trackers[0].writer.add_images("test_samples", images_processed, epoch)
 
-            # save the model
+            if epoch % args.save_model_epochs == 0 or epoch == args.num_epochs - 1:
-            if args.push_to_hub:
+                # save the model
-                push_to_hub(args, pipeline, repo, commit_message=f"Epoch {epoch}", blocking=False)
+                if args.push_to_hub:
-            else:
+                    push_to_hub(args, pipeline, repo, commit_message=f"Epoch {epoch}", blocking=False)
-                pipeline.save_pretrained(args.output_dir)
+                else:
+                    pipeline.save_pretrained(args.output_dir)
         accelerator.wait_for_everyone()
 
     accelerator.end_training()
@@ -188,12 +180,13 @@ if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Simple example of a training script.")
     parser.add_argument("--local_rank", type=int, default=-1)
     parser.add_argument("--dataset", type=str, default="huggan/flowers-102-categories")
-    parser.add_argument("--output_dir", type=str, default="ddpm-model")
+    parser.add_argument("--output_dir", type=str, default="ddpm-flowers-64")
     parser.add_argument("--overwrite_output_dir", action="store_true")
     parser.add_argument("--resolution", type=int, default=64)
     parser.add_argument("--train_batch_size", type=int, default=16)
     parser.add_argument("--eval_batch_size", type=int, default=16)
     parser.add_argument("--num_epochs", type=int, default=100)
+    parser.add_argument("--save_model_epochs", type=int, default=5)
     parser.add_argument("--gradient_accumulation_steps", type=int, default=1)
     parser.add_argument("--learning_rate", type=float, default=1e-4)
     parser.add_argument("--lr_scheduler", type=str, default="cosine")
@@ -202,6 +195,7 @@ if __name__ == "__main__":
     parser.add_argument("--adam_beta2", type=float, default=0.999)
     parser.add_argument("--adam_weight_decay", type=float, default=1e-6)
     parser.add_argument("--adam_epsilon", type=float, default=1e-3)
+    parser.add_argument("--use_ema", action="store_true", default=True)
     parser.add_argument("--ema_inv_gamma", type=float, default=1.0)
     parser.add_argument("--ema_power", type=float, default=3 / 4)
     parser.add_argument("--ema_max_decay", type=float, default=0.9999)