integrate ort (#1110)

* integrate ort

* use return_dict=False

* revert unet return value change

* revert unet return value change

* add note to readme

* adjust readme

* add contact

* `make style`
Co-authored-by: Prathik Rao <prathikrao@microsoft.com>
Co-authored-by: Anton Lozhkov <anton@huggingface.co>

examples/unconditional_image_generation/README.md

@@ -127,3 +127,24 @@ dataset.push_to_hub("name_of_your_dataset", private=True)
 and that's it! You can now train your model by simply setting the `--dataset_name` argument to the name of your dataset on the hub.
 
 More on this can also be found in [this blog post](https://huggingface.co/blog/image-search-datasets).
+
+#### Use ONNXRuntime to accelerate training
+
+In order to leverage onnxruntime to accelerate training, please use train_unconditional_ort.py
+
+The command to train a DDPM UNet model on the Oxford Flowers dataset with onnxruntime:
+
+```bash
+accelerate launch train_unconditional_ort.py \
+  --dataset_name="huggan/flowers-102-categories" \
+  --resolution=64 \
+  --output_dir="ddpm-ema-flowers-64" \
+  --train_batch_size=16 \
+  --num_epochs=1 \
+  --gradient_accumulation_steps=1 \
+  --learning_rate=1e-4 \
+  --lr_warmup_steps=500 \
+  --mixed_precision=fp16
+  ```
+
+Please contact Prathik Rao (prathikr), Sunghoon Choi (hanbitmyths), Ashwini Khade (askhade), or Peng Wang (pengwa) on github with any questions.
\ No newline at end of file

examples/unconditional_image_generation/train_unconditional_ort.py

+import argparse
+import math
+import os
+
+import torch
+import torch.nn.functional as F
+
+from accelerate import Accelerator
+from accelerate.logging import get_logger
+from datasets import load_dataset
+from diffusers import DDPMPipeline, DDPMScheduler, UNet2DModel
+from diffusers.hub_utils import init_git_repo, push_to_hub
+from diffusers.optimization import get_scheduler
+from diffusers.training_utils import EMAModel
+from onnxruntime.training.ortmodule import ORTModule
+from torchvision.transforms import (
+    CenterCrop,
+    Compose,
+    InterpolationMode,
+    Normalize,
+    RandomHorizontalFlip,
+    Resize,
+    ToTensor,
+)
+from tqdm.auto import tqdm
+
+
+logger = get_logger(__name__)
+
+
+def main(args):
+    logging_dir = os.path.join(args.output_dir, args.logging_dir)
+    accelerator = Accelerator(
+        gradient_accumulation_steps=args.gradient_accumulation_steps,
+        mixed_precision=args.mixed_precision,
+        log_with="tensorboard",
+        logging_dir=logging_dir,
+    )
+
+    model = UNet2DModel(
+        sample_size=args.resolution,
+        in_channels=3,
+        out_channels=3,
+        layers_per_block=2,
+        block_out_channels=(128, 128, 256, 256, 512, 512),
+        down_block_types=(
+            "DownBlock2D",
+            "DownBlock2D",
+            "DownBlock2D",
+            "DownBlock2D",
+            "AttnDownBlock2D",
+            "DownBlock2D",
+        ),
+        up_block_types=(
+            "UpBlock2D",
+            "AttnUpBlock2D",
+            "UpBlock2D",
+            "UpBlock2D",
+            "UpBlock2D",
+            "UpBlock2D",
+        ),
+    )
+    model = ORTModule(model)
+    noise_scheduler = DDPMScheduler(num_train_timesteps=1000, tensor_format="pt")
+    optimizer = torch.optim.AdamW(
+        model.parameters(),
+        lr=args.learning_rate,
+        betas=(args.adam_beta1, args.adam_beta2),
+        weight_decay=args.adam_weight_decay,
+        eps=args.adam_epsilon,
+    )
+
+    augmentations = Compose(
+        [
+            Resize(args.resolution, interpolation=InterpolationMode.BILINEAR),
+            CenterCrop(args.resolution),
+            RandomHorizontalFlip(),
+            ToTensor(),
+            Normalize([0.5], [0.5]),
+        ]
+    )
+
+    if args.dataset_name is not None:
+        dataset = load_dataset(
+            args.dataset_name,
+            args.dataset_config_name,
+            cache_dir=args.cache_dir,
+            use_auth_token=True if args.use_auth_token else None,
+            split="train",
+        )
+    else:
+        dataset = load_dataset("imagefolder", data_dir=args.train_data_dir, cache_dir=args.cache_dir, split="train")
+
+    def transforms(examples):
+        images = [augmentations(image.convert("RGB")) for image in examples["image"]]
+        return {"input": images}
+
+    dataset.set_transform(transforms)
+    train_dataloader = torch.utils.data.DataLoader(dataset, batch_size=args.train_batch_size, shuffle=True)
+
+    lr_scheduler = get_scheduler(
+        args.lr_scheduler,
+        optimizer=optimizer,
+        num_warmup_steps=args.lr_warmup_steps,
+        num_training_steps=(len(train_dataloader) * args.num_epochs) // args.gradient_accumulation_steps,
+    )
+
+    model, optimizer, train_dataloader, lr_scheduler = accelerator.prepare(
+        model, optimizer, train_dataloader, lr_scheduler
+    )
+
+    num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
+
+    ema_model = EMAModel(model, inv_gamma=args.ema_inv_gamma, power=args.ema_power, max_value=args.ema_max_decay)
+
+    if args.push_to_hub:
+        repo = init_git_repo(args, at_init=True)
+
+    if accelerator.is_main_process:
+        run = os.path.split(__file__)[-1].split(".")[0]
+        accelerator.init_trackers(run)
+
+    global_step = 0
+    for epoch in range(args.num_epochs):
+        model.train()
+        progress_bar = tqdm(total=num_update_steps_per_epoch, disable=not accelerator.is_local_main_process)
+        progress_bar.set_description(f"Epoch {epoch}")
+        for step, batch in enumerate(train_dataloader):
+            clean_images = batch["input"]
+            # Sample noise that we'll add to the images
+            noise = torch.randn(clean_images.shape).to(clean_images.device)
+            bsz = clean_images.shape[0]
+            # Sample a random timestep for each image
+            timesteps = torch.randint(
+                0, noise_scheduler.config.num_train_timesteps, (bsz,), device=clean_images.device
+            ).long()
+
+            # 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, timesteps)
+
+            with accelerator.accumulate(model):
+                # Predict the noise residual
+                noise_pred = model(noisy_images, timesteps, return_dict=True)[0]
+                loss = F.mse_loss(noise_pred, noise)
+                accelerator.backward(loss)
+
+                accelerator.clip_grad_norm_(model.parameters(), 1.0)
+                optimizer.step()
+                lr_scheduler.step()
+                if args.use_ema:
+                    ema_model.step(model)
+                optimizer.zero_grad()
+
+            # Checks if the accelerator has performed an optimization step behind the scenes
+            if accelerator.sync_gradients:
+                progress_bar.update(1)
+                global_step += 1
+
+            logs = {"loss": loss.detach().item(), "lr": lr_scheduler.get_last_lr()[0], "step": global_step}
+            if args.use_ema:
+                logs["ema_decay"] = ema_model.decay
+            progress_bar.set_postfix(**logs)
+            accelerator.log(logs, step=global_step)
+        progress_bar.close()
+
+        accelerator.wait_for_everyone()
+
+        # Generate sample images for visual inspection
+        if accelerator.is_main_process:
+            if epoch % args.save_images_epochs == 0 or epoch == args.num_epochs - 1:
+                pipeline = DDPMPipeline(
+                    unet=accelerator.unwrap_model(ema_model.averaged_model if args.use_ema else model),
+                    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, output_type="numpy").images
+
+                # denormalize the images and save to tensorboard
+                images_processed = (images * 255).round().astype("uint8")
+                accelerator.trackers[0].writer.add_images(
+                    "test_samples", images_processed.transpose(0, 3, 1, 2), epoch
+                )
+
+            if epoch % args.save_model_epochs == 0 or epoch == args.num_epochs - 1:
+                # save the model
+                if args.push_to_hub:
+                    push_to_hub(args, pipeline, repo, commit_message=f"Epoch {epoch}", blocking=False)
+                else:
+                    pipeline.save_pretrained(args.output_dir)
+        accelerator.wait_for_everyone()
+
+    accelerator.end_training()
+
+
+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_name", type=str, default=None)
+    parser.add_argument("--dataset_config_name", type=str, default=None)
+    parser.add_argument("--train_data_dir", type=str, default=None, help="A folder containing the training data.")
+    parser.add_argument("--output_dir", type=str, default="ddpm-model-64")
+    parser.add_argument("--overwrite_output_dir", action="store_true")
+    parser.add_argument("--cache_dir", type=str, default=None)
+    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_images_epochs", type=int, default=10)
+    parser.add_argument("--save_model_epochs", type=int, default=10)
+    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")
+    parser.add_argument("--lr_warmup_steps", type=int, default=500)
+    parser.add_argument("--adam_beta1", type=float, default=0.95)
+    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-08)
+    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)
+    parser.add_argument("--push_to_hub", action="store_true")
+    parser.add_argument("--use_auth_token", action="store_true")
+    parser.add_argument("--hub_token", type=str, default=None)
+    parser.add_argument("--hub_model_id", type=str, default=None)
+    parser.add_argument("--hub_private_repo", action="store_true")
+    parser.add_argument("--logging_dir", type=str, default="logs")
+    parser.add_argument(
+        "--mixed_precision",
+        type=str,
+        default="no",
+        choices=["no", "fp16", "bf16"],
+        help=(
+            "Whether to use mixed precision. Choose"
+            "between fp16 and bf16 (bfloat16). Bf16 requires PyTorch >= 1.10."
+            "and an Nvidia Ampere GPU."
+        ),
+    )
+
+    args = parser.parse_args()
+    env_local_rank = int(os.environ.get("LOCAL_RANK", -1))
+    if env_local_rank != -1 and env_local_rank != args.local_rank:
+        args.local_rank = env_local_rank
+
+    if args.dataset_name is None and args.train_data_dir is None:
+        raise ValueError("You must specify either a dataset name from the hub or a train data directory.")
+
+    main(args)

src/diffusers/pipeline_utils.py

@@ -78,6 +78,9 @@ LOADABLE_CLASSES = {
         "ProcessorMixin": ["save_pretrained", "from_pretrained"],
         "ImageProcessingMixin": ["save_pretrained", "from_pretrained"],
     },
+    "onnxruntime.training": {
+        "ORTModule": ["save_pretrained", "from_pretrained"],
+    },
 }
 
 ALL_IMPORTABLE_CLASSES = {}