train.py

### Modified from the text to image training script in Diffusers library
### Original file: https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py

import argparse
import logging
import math
import os
import json
import random
import shutil
from pathlib import Path
from PIL import Image

import accelerate
import datasets
import numpy as np
import torch
import torch.nn.functional as F
import torch.utils.checkpoint
import transformers
from accelerate import Accelerator
from accelerate.logging import get_logger
from accelerate.utils import ProjectConfiguration, set_seed
from packaging import version
from torchvision import transforms
from tqdm.auto import tqdm
from transformers import CLIPTextModel, CLIPTokenizer
from torch.utils.data import DataLoader, Dataset

import diffusers
from diffusers import AutoencoderKL, DDPMScheduler, StableDiffusionPipeline, UNet2DConditionModel
from diffusers.optimization import get_scheduler
from diffusers.training_utils import compute_snr
from diffusers.utils import is_wandb_available
from diffusers.utils.import_utils import is_xformers_available


if is_wandb_available():
    import wandb

logger = get_logger(__name__, log_level="INFO")


def log_validation(vae, text_encoder, value_encoder, tokenizer, unet, args, accelerator, weight_dtype, epoch):
    logger.info("Running validation... ")

    value_encoder = accelerator.unwrap_model(value_encoder)
    pipeline = StableDiffusionPipeline.from_pretrained(
        args.pretrained_model_name_or_path,
        vae=accelerator.unwrap_model(vae),
        text_encoder=accelerator.unwrap_model(text_encoder),
        tokenizer=tokenizer,
        unet=accelerator.unwrap_model(unet),
        safety_checker=None,
        revision=args.revision,
        torch_dtype=weight_dtype,
    )

    pipeline = pipeline.to(accelerator.device)
    pipeline.set_progress_bar_config(disable=True)

    if args.enable_xformers_memory_efficient_attention:
        pipeline.enable_xformers_memory_efficient_attention()

    if args.seed is None:
        generator = None
    else:
        generator = torch.Generator(device=accelerator.device).manual_seed(args.seed)

    images = []
    for i in range(len(args.validation_prompts)):
        with torch.autocast("cuda"):
            _, prompt = args.validation_prompts[i].split(', ', 1)
            input_ids = pipeline.tokenizer(prompt, return_tensors="pt").input_ids
            prompt_embeds = pipeline.text_encoder(input_ids.to(accelerator.device))[0]
            valued_input_ids = pipeline.tokenizer(args.validation_prompts[i], return_tensors="pt").input_ids
            value_embeds = value_encoder(valued_input_ids.to(accelerator.device))[1]
            concat_embeds = torch.cat([value_embeds.unsqueeze(1), prompt_embeds], dim=1)
            images += pipeline(prompt_embeds=concat_embeds, num_images_per_prompt=args.num_images_per_val_prompt, generator=generator).images

    for tracker in accelerator.trackers:
        if tracker.name == "tensorboard":
            np_images = np.stack([np.asarray(img) for img in images])
            tracker.writer.add_images("validation", np_images, epoch, dataformats="NHWC")
        elif tracker.name == "wandb":
            tracker.log(
                {
                    "validation": [
                        wandb.Image(image, caption=f"{i // args.num_images_per_val_prompt}: {args.validation_prompts[i // args.num_images_per_val_prompt]}")
                        for i, image in enumerate(images)
                    ]
                }
            )
        else:
            logger.warn(f"image logging not implemented for {tracker.name}")

    del pipeline
    torch.cuda.empty_cache()

    return images

def parse_args():
    parser = argparse.ArgumentParser(description="Simple example of a training script.")
    parser.add_argument(
        "--input_perturbation", type=float, default=0, help="The scale of input perturbation. Recommended 0.1."
    )
    parser.add_argument(
        "--pretrained_model_name_or_path",
        type=str,
        default=None,
        required=True,
        help="Path to pretrained model or model identifier from huggingface.co/models.",
    )
    parser.add_argument(
        "--revision",
        type=str,
        default=None,
        required=False,
        help="Revision of pretrained model identifier from huggingface.co/models.",
    )
    parser.add_argument(
        "--dataset_name",
        type=str,
        default=None,
        help=(
            "The name of the Dataset (from the HuggingFace hub) to train on (could be your own, possibly private,"
            " dataset). It can also be a path pointing to a local copy of a dataset in your filesystem,"
            " or to a folder containing files that 🤗 Datasets can understand."
        ),
    )
    parser.add_argument(
        "--train_data_dir",
        type=str,
        default=None,
        help=(
            "A folder containing the training data. Folder contents must follow the structure described in"
            " https://huggingface.co/docs/datasets/image_dataset#imagefolder. In particular, a `metadata.jsonl` file"
            " must exist to provide the captions for the images. Ignored if `dataset_name` is specified."
        ),
    )
    parser.add_argument(
        "--max_train_proportion",
        type=float,
        default=None,
        help=(
            "For debugging purposes or quicker training, truncate the number of training examples to this "
            "value if set."
        ),
    )
    parser.add_argument(
        "--validation_prompts",
        type=str,
        default=None,
        nargs="+",
        help=("A set of prompts evaluated every `--validation_steps` and logged to `--report_to`."),
    )
    parser.add_argument(
        "--num_images_per_val_prompt",
        type=int,
        default=1,
        help="The number of images to generate per validation prompt.",
    )
    parser.add_argument(
        "--name",
        type=str,
        default=None,
        required=True,
        help="Training name",
    )
    parser.add_argument(
        "--output_dir",
        type=str,
        required=True,
        help="The output directory where the model predictions and checkpoints will be written.",
    )
    parser.add_argument(
        "--cache_dir",
        type=str,
        default=None,
        help="The directory where the downloaded models and datasets will be stored.",
    )
    parser.add_argument("--seed", type=int, default=None, help="A seed for reproducible training.")
    parser.add_argument(
        "--resolution",
        type=int,
        default=512,
        help=(
            "The resolution for input images, all the images in the train/validation dataset will be resized to this"
            " resolution"
        ),
    )
    parser.add_argument(
        "--center_crop",
        default=False,
        action="store_true",
        help=(
            "Whether to center crop the input images to the resolution. If not set, the images will be randomly"
            " cropped. The images will be resized to the resolution first before cropping."
        ),
    )
    parser.add_argument(
        "--random_flip",
        action="store_true",
        help="whether to randomly flip images horizontally",
    )
    parser.add_argument(
        "--train_batch_size", type=int, default=16, help="Batch size (per device) for the training dataloader."
    )
    parser.add_argument("--num_train_epochs", type=int, default=100)
    parser.add_argument(
        "--max_train_steps",
        type=int,
        default=None,
        help="Total number of training steps to perform.  If provided, overrides num_train_epochs.",
    )
    parser.add_argument(
        "--gradient_accumulation_steps",
        type=int,
        default=1,
        help="Number of updates steps to accumulate before performing a backward/update pass.",
    )
    parser.add_argument(
        "--gradient_checkpointing",
        action="store_true",
        help="Whether or not to use gradient checkpointing to save memory at the expense of slower backward pass.",
    )
    parser.add_argument(
        "--learning_rate",
        type=float,
        default=1e-4,
        help="Initial learning rate (after the potential warmup period) to use.",
    )
    parser.add_argument(
        "--scale_lr",
        action="store_true",
        default=False,
        help="Scale the learning rate by the number of GPUs, gradient accumulation steps, and batch size.",
    )
    parser.add_argument(
        "--lr_scheduler",
        type=str,
        default="constant",
        help=(
            'The scheduler type to use. Choose between ["linear", "cosine", "cosine_with_restarts", "polynomial",'
            ' "constant", "constant_with_warmup"]'
        ),
    )
    parser.add_argument(
        "--lr_warmup_steps", type=int, default=500, help="Number of steps for the warmup in the lr scheduler."
    )
    parser.add_argument(
        "--snr_gamma",
        type=float,
        default=None,
        help="SNR weighting gamma to be used if rebalancing the loss. Recommended value is 5.0. "
        "More details here: https://arxiv.org/abs/2303.09556.",
    )
    parser.add_argument(
        "--livo_beta",
        type=float,
        default=0.1,
        help="beta parameter for the LiVO loss."
    )
    parser.add_argument(
        "--livo_alpha",
        type=float,
        default=0.1,
        help="alpha parameter for the LiVO loss."
    )
    parser.add_argument(
        "--livo_gamma_1",
        type=float,
        default=0.1,
        help="beta parameter for the LiVO loss."
    )
    parser.add_argument(
        "--livo_gamma_2",
        type=float,
        default=0.1,
        help="beta parameter for the LiVO loss."
    )
    parser.add_argument(
        "--use_8bit_adam", action="store_true", help="Whether or not to use 8-bit Adam from bitsandbytes."
    )
    parser.add_argument(
        "--allow_tf32",
        action="store_true",
        help=(
            "Whether or not to allow TF32 on Ampere GPUs. Can be used to speed up training. For more information, see"
            " https://pytorch.org/docs/stable/notes/cuda.html#tensorfloat-32-tf32-on-ampere-devices"
        ),
    )
    parser.add_argument(
        "--dataloader_num_workers",
        type=int,
        default=0,
        help=(
            "Number of subprocesses to use for data loading. 0 means that the data will be loaded in the main process."
        ),
    )
    parser.add_argument("--adam_beta1", type=float, default=0.9, help="The beta1 parameter for the Adam optimizer.")
    parser.add_argument("--adam_beta2", type=float, default=0.999, help="The beta2 parameter for the Adam optimizer.")
    parser.add_argument("--adam_weight_decay", type=float, default=1e-2, help="Weight decay to use.")
    parser.add_argument("--adam_epsilon", type=float, default=1e-08, help="Epsilon value for the Adam optimizer")
    parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.")
    parser.add_argument(
        "--prediction_type",
        type=str,
        default=None,
        help="The prediction_type that shall be used for training. Choose between 'epsilon' or 'v_prediction' or leave `None`. If left to `None` the default prediction type of the scheduler: `noise_scheduler.config.prediciton_type` is chosen.",
    )
    parser.add_argument(
        "--logging_dir",
        type=str,
        default="logs",
        help=(
            "[TensorBoard](https://www.tensorflow.org/tensorboard) log directory. Will default to"
            " *output_dir/runs/**CURRENT_DATETIME_HOSTNAME***."
        ),
    )
    parser.add_argument(
        "--mixed_precision",
        type=str,
        default=None,
        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.  Default to the value of accelerate config of the current system or the"
            " flag passed with the `accelerate.launch` command. Use this argument to override the accelerate config."
        ),
    )
    parser.add_argument(
        "--report_to",
        type=str,
        default="tensorboard",
        help=(
            'The integration to report the results and logs to. Supported platforms are `"tensorboard"`'
            ' (default), `"wandb"` and `"comet_ml"`. Use `"all"` to report to all integrations.'
        ),
    )
    parser.add_argument("--local_rank", type=int, default=-1, help="For distributed training: local_rank")
    parser.add_argument(
        "--checkpointing_steps",
        type=int,
        default=500,
        help=(
            "Save a checkpoint of the training state every X updates. These checkpoints are only suitable for resuming"
            " training using `--resume_from_checkpoint`."
        ),
    )
    parser.add_argument(
        "--checkpoints_total_limit",
        type=int,
        default=None,
        help=("Max number of checkpoints to store."),
    )
    parser.add_argument(
        "--resume_from_checkpoint",
        type=str,
        default=None,
        help=(
            "Whether training should be resumed from a previous checkpoint. Use a path saved by"
            ' `--checkpointing_steps`, or `"latest"` to automatically select the last available checkpoint.'
        ),
    )
    parser.add_argument(
        "--enable_xformers_memory_efficient_attention", action="store_true", help="Whether or not to use xformers."
    )
    parser.add_argument("--noise_offset", type=float, default=0, help="The scale of noise offset.")
    parser.add_argument(
        "--validation_steps",
        type=int,
        default=100,
        help=(
            "Run validation every X steps. Validation consists of running the prompt"
            " `args.validation_prompt` multiple times: `args.num_validation_images`"
            " and logging the images."
        ),
    )
    parser.add_argument(
        "--tracker_project_name",
        type=str,
        default="text2image-fine-tune",
        help=(
            "The `project_name` argument passed to Accelerator.init_trackers for"
            " more information see https://huggingface.co/docs/accelerate/v0.17.0/en/package_reference/accelerator#accelerate.Accelerator"
        ),
    )

    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

    # Sanity checks
    if args.dataset_name is None and args.train_data_dir is None:
        raise ValueError("Need either a dataset name or a training folder.")

    return args

class ValueEncoderDataset(Dataset):
    def __init__(self, train_data_dir, tokenizer, resolution, center_crop, random_flip, max_train_proportion):
        self.tokenizer = tokenizer
        self.train_transforms = transforms.Compose(
            [
                transforms.Resize(resolution, interpolation=transforms.InterpolationMode.BILINEAR),
                transforms.CenterCrop(resolution) if center_crop else transforms.RandomCrop(resolution),
                transforms.RandomHorizontalFlip() if random_flip else transforms.Lambda(lambda x: x),
                transforms.ToTensor(),
                transforms.Normalize([0.5], [0.5]),
            ]
        )
        
        self.train_data_dir = Path(train_data_dir)
        type_dir_list = sorted(os.listdir(self.train_data_dir))
        self.scene_list = []
        self.scene_neg_list = []

        for type_dir in type_dir_list:
            scene_dir_list = sorted(os.listdir(self.train_data_dir / type_dir))
            for scene_dir in scene_dir_list:

                with open(self.train_data_dir / type_dir / scene_dir / 'metadata.jsonl', 'r') as f:
                    metadata_list = [json.loads(line) for line in f]
                
                positive_list, negative_list = [], []  
                for metadata in metadata_list:
                    if metadata['label'] == 1: # positive sample
                        positive_list.append({
                            'prompt': metadata['prompt'],
                            'value': metadata['value'],
                            'file_dir': self.train_data_dir / type_dir / scene_dir,
                            'file_name': metadata['file_name'],
                            'neg_list_idx': len(self.scene_neg_list)
                        })
                    else: # negative sample
                        negative_list.append(metadata['file_name'])
                
                if max_train_proportion is not None:
                    max_train_proportion = min(max_train_proportion, 1.0)
                    self.scene_list.extend(positive_list[:int(len(positive_list) * max_train_proportion)])
                    self.scene_neg_list.append(negative_list[:int(len(negative_list) * max_train_proportion)])
                else:
                    self.scene_list.extend(positive_list)
                    self.scene_neg_list.append(negative_list)

    def __len__(self):
        return len(self.scene_list)
    
    def __getitem__(self, index):
        metadata = self.scene_list[index]

        pos_img = Image.open(metadata['file_dir'] / metadata['file_name']).convert("RGB")
        pos_pixel_values = self.train_transforms(pos_img)

        neg_img_name = random.choice(self.scene_neg_list[metadata['neg_list_idx']])
        neg_img = Image.open(metadata['file_dir'] / neg_img_name).convert("RGB")
        neg_pixel_values = self.train_transforms(neg_img)

        caption, value = metadata['prompt'], metadata['value']
        valued_caption = value + ', ' + caption

        inputs = self.tokenizer(
            caption, max_length=self.tokenizer.model_max_length, padding="max_length", truncation=True, return_tensors="pt"
        )
        valued_inputs = self.tokenizer(
            valued_caption, max_length=self.tokenizer.model_max_length, padding="max_length", truncation=True, return_tensors="pt"
        )

        return {'input_ids': inputs.input_ids, 'valued_input_ids': valued_inputs.input_ids, 'pixel_values': pos_pixel_values, 'neg_pixel_values': neg_pixel_values}


def main():
    args = parse_args()

    logging_dir = os.path.join(args.output_dir, args.logging_dir)

    accelerator_project_config = ProjectConfiguration(project_dir=args.output_dir, logging_dir=logging_dir)

    accelerator = Accelerator(
        gradient_accumulation_steps=args.gradient_accumulation_steps,
        mixed_precision=args.mixed_precision,
        log_with=args.report_to,
        project_config=accelerator_project_config,
    )

    # Make one log on every process with the configuration for debugging.
    logging.basicConfig(
        format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
        datefmt="%m/%d/%Y %H:%M:%S",
        level=logging.INFO,
    )
    logger.info(accelerator.state, main_process_only=False)
    if accelerator.is_local_main_process:
        datasets.utils.logging.set_verbosity_warning()
        transformers.utils.logging.set_verbosity_warning()
        diffusers.utils.logging.set_verbosity_info()
    else:
        datasets.utils.logging.set_verbosity_error()
        transformers.utils.logging.set_verbosity_error()
        diffusers.utils.logging.set_verbosity_error()

    # If passed along, set the training seed now.
    if args.seed is not None:
        set_seed(args.seed)

    # Handle the repository creation
    if accelerator.is_main_process:
        if args.output_dir is not None:
            os.makedirs(args.output_dir, exist_ok=True)

    # Load scheduler, tokenizer and models.
    noise_scheduler = DDPMScheduler.from_pretrained(args.pretrained_model_name_or_path, subfolder="scheduler")
    tokenizer = CLIPTokenizer.from_pretrained(
        args.pretrained_model_name_or_path, subfolder="tokenizer", revision=args.revision
    )
    text_encoder = CLIPTextModel.from_pretrained(
        args.pretrained_model_name_or_path, subfolder="text_encoder", revision=args.revision
    )

    value_encoder = CLIPTextModel.from_pretrained(
        args.pretrained_model_name_or_path, subfolder="text_encoder", revision=args.revision
    )

    vae = AutoencoderKL.from_pretrained(
        args.pretrained_model_name_or_path, subfolder="vae", revision=args.revision
    )
    unet = UNet2DConditionModel.from_pretrained(
        args.pretrained_model_name_or_path, subfolder="unet", revision=args.revision
    )

    # Freeze original stable diffusion
    text_encoder.requires_grad_(False)
    vae.requires_grad_(False)
    unet.requires_grad_(False)

    if args.gradient_checkpointing:
        # Keep unet in train mode if we are using gradient checkpointing to save memory.
        # The dropout cannot be != 0 so it doesn't matter if we are in eval or train mode.
        unet.train()
        value_encoder.gradient_checkpointing_enable()
        unet.enable_gradient_checkpointing()

    if args.enable_xformers_memory_efficient_attention:
        if is_xformers_available():
            import xformers

            xformers_version = version.parse(xformers.__version__)
            if xformers_version == version.parse("0.0.16"):
                logger.warn(
                    "xFormers 0.0.16 cannot be used for training in some GPUs. If you observe problems during training, please update xFormers to at least 0.0.17. See https://huggingface.co/docs/diffusers/main/en/optimization/xformers for more details."
                )
            unet.enable_xformers_memory_efficient_attention()
        else:
            raise ValueError("xformers is not available. Make sure it is installed correctly")

    # `accelerate` 0.16.0 will have better support for customized saving
    if version.parse(accelerate.__version__) >= version.parse("0.16.0"):
        # create custom saving & loading hooks so that `accelerator.save_state(...)` serializes in a nice format
        def save_model_hook(models, weights, output_dir):
            if accelerator.is_main_process:
                for i, model in enumerate(models):
                    model.save_pretrained(os.path.join(output_dir, "value_encoder"))

                    # make sure to pop weight so that corresponding model is not saved again
                    weights.pop()

        def load_model_hook(models, input_dir):
            for i in range(len(models)):
                # pop models so that they are not loaded again
                model = models.pop()

                # load diffusers style into model
                load_model = CLIPTextModel.from_pretrained(input_dir, subfolder="value_encoder")
                model.register_to_config(**load_model.config)

                model.load_state_dict(load_model.state_dict())
                del load_model

        accelerator.register_save_state_pre_hook(save_model_hook)
        accelerator.register_load_state_pre_hook(load_model_hook)

    # Enable TF32 for faster training on Ampere GPUs,
    # cf https://pytorch.org/docs/stable/notes/cuda.html#tensorfloat-32-tf32-on-ampere-devices
    if args.allow_tf32:
        torch.backends.cuda.matmul.allow_tf32 = True

    if args.scale_lr:
        args.learning_rate = (
            args.learning_rate * args.gradient_accumulation_steps * args.train_batch_size * accelerator.num_processes
        )

    # Initialize the optimizer
    if args.use_8bit_adam:
        try:
            import bitsandbytes as bnb
        except ImportError:
            raise ImportError(
                "Please install bitsandbytes to use 8-bit Adam. You can do so by running `pip install bitsandbytes`"
            )

        optimizer_cls = bnb.optim.AdamW8bit
    else:
        optimizer_cls = torch.optim.AdamW

    optimizer = optimizer_cls(
        value_encoder.parameters(), # optimize the value encoder parameters
        lr=args.learning_rate,
        betas=(args.adam_beta1, args.adam_beta2),
        weight_decay=args.adam_weight_decay,
        eps=args.adam_epsilon,
    )

    train_dataset = ValueEncoderDataset(
        train_data_dir=args.train_data_dir,
        tokenizer=tokenizer,
        resolution=args.resolution,
        center_crop=args.center_crop,
        random_flip=args.random_flip,
        max_train_proportion=args.max_train_proportion
    )

    # DataLoaders creation:
    train_dataloader = DataLoader(
        train_dataset,
        shuffle=True,
        batch_size=args.train_batch_size,
        num_workers=args.dataloader_num_workers,
    )

    # Scheduler and math around the number of training steps.
    overrode_max_train_steps = False
    num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
    if args.max_train_steps is None:
        args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
        overrode_max_train_steps = True

    lr_scheduler = get_scheduler(
        args.lr_scheduler,
        optimizer=optimizer,
        num_warmup_steps=args.lr_warmup_steps * accelerator.num_processes,
        num_training_steps=args.max_train_steps * accelerator.num_processes,
    )

    # Prepare everything with our `accelerator`.
    value_encoder, optimizer, train_dataloader, lr_scheduler = accelerator.prepare(
        value_encoder, optimizer, train_dataloader, lr_scheduler
    )

    # For mixed precision training we cast all non-trainable weigths (vae, non-lora text_encoder and non-lora unet) to half-precision
    # as these weights are only used for inference, keeping weights in full precision is not required.
    weight_dtype = torch.float32
    if accelerator.mixed_precision == "fp16":
        weight_dtype = torch.float16
        args.mixed_precision = accelerator.mixed_precision
    elif accelerator.mixed_precision == "bf16":
        weight_dtype = torch.bfloat16
        args.mixed_precision = accelerator.mixed_precision

    # Move text_encoder, vae and unet to gpu and cast to weight_dtype
    text_encoder.to(accelerator.device, dtype=weight_dtype)
    vae.to(accelerator.device, dtype=weight_dtype)
    unet.to(accelerator.device, dtype=weight_dtype)

    # We need to recalculate our total training steps as the size of the training dataloader may have changed.
    num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
    if overrode_max_train_steps:
        args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
    # Afterwards we recalculate our number of training epochs
    args.num_train_epochs = math.ceil(args.max_train_steps / num_update_steps_per_epoch)

    # We need to initialize the trackers we use, and also store our configuration.
    # The trackers initializes automatically on the main process.
    if accelerator.is_main_process:
        tracker_config = dict(vars(args))
        tracker_config.pop("validation_prompts")
        accelerator.init_trackers(args.tracker_project_name, tracker_config, init_kwargs={
                "wandb": {
                    "name": args.name,
                    "dir": args.output_dir
                }
            })

    # Train!
    total_batch_size = args.train_batch_size * accelerator.num_processes * args.gradient_accumulation_steps

    logger.info("***** Running training *****")
    logger.info(f"  Num examples = {len(train_dataset)}")
    logger.info(f"  Num Epochs = {args.num_train_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_batch_size}")
    logger.info(f"  Gradient Accumulation steps = {args.gradient_accumulation_steps}")
    logger.info(f"  Total optimization steps = {args.max_train_steps}")
    global_step = 0
    first_epoch = 0

    # Potentially load in the weights and states from a previous save
    if args.resume_from_checkpoint:
        if args.resume_from_checkpoint != "latest":
            path = os.path.basename(args.resume_from_checkpoint)
        else:
            # Get the most recent checkpoint
            dirs = os.listdir(args.output_dir)
            dirs = [d for d in dirs if d.startswith("checkpoint")]
            dirs = sorted(dirs, key=lambda x: int(x.split("-")[1]))
            path = dirs[-1] if len(dirs) > 0 else None

        if path is None:
            accelerator.print(
                f"Checkpoint '{args.resume_from_checkpoint}' does not exist. Starting a new training run."
            )
            args.resume_from_checkpoint = None
        else:
            accelerator.print(f"Resuming from checkpoint {path}")
            accelerator.load_state(os.path.join(args.output_dir, path))
            global_step = int(path.split("-")[1])

            resume_global_step = global_step * args.gradient_accumulation_steps
            first_epoch = global_step // num_update_steps_per_epoch
            resume_step = resume_global_step % (num_update_steps_per_epoch * args.gradient_accumulation_steps)

    # Only show the progress bar once on each machine.
    progress_bar = tqdm(range(global_step, args.max_train_steps), disable=not accelerator.is_local_main_process)
    progress_bar.set_description("Steps")

    for epoch in range(first_epoch, args.num_train_epochs):
        value_encoder.train()
        train_loss = 0.0
        loss_theta_w, loss_ref_w, loss_theta_l, loss_ref_l = 0.0, 0.0, 0.0, 0.0
        for step, batch in enumerate(train_dataloader):
            # Skip steps until we reach the resumed step
            if args.resume_from_checkpoint and epoch == first_epoch and step < resume_step:
                if step % args.gradient_accumulation_steps == 0:
                    progress_bar.update(1)
                continue

            with accelerator.accumulate(value_encoder):
                # Convert images to latent space
                latents = vae.encode(batch["pixel_values"].to(weight_dtype)).latent_dist.sample()
                latents = latents * vae.config.scaling_factor

                neg_latents = vae.encode(batch["neg_pixel_values"].to(weight_dtype)).latent_dist.sample()
                neg_latents = neg_latents * vae.config.scaling_factor

                # Sample noise that we'll add to the latents
                noise = torch.randn_like(latents)
                if args.noise_offset:
                    # https://www.crosslabs.org//blog/diffusion-with-offset-noise
                    noise += args.noise_offset * torch.randn(
                        (latents.shape[0], latents.shape[1], 1, 1), device=latents.device
                    )
                if args.input_perturbation:
                    new_noise = noise + args.input_perturbation * torch.randn_like(noise)
                bsz = latents.shape[0]
                # Sample a random timestep for each image
                timesteps = torch.randint(0, noise_scheduler.config.num_train_timesteps, (bsz,), device=latents.device)
                timesteps = timesteps.long()

                # Add noise to the latents according to the noise magnitude at each timestep
                # (this is the forward diffusion process)
                if args.input_perturbation:
                    noisy_latents = noise_scheduler.add_noise(latents, new_noise, timesteps)
                    neg_noisy_latents = noise_scheduler.add_noise(neg_latents, new_noise, timesteps)
                else:
                    noisy_latents = noise_scheduler.add_noise(latents, noise, timesteps)
                    neg_noisy_latents = noise_scheduler.add_noise(neg_latents, noise, timesteps)

                # Get the text embedding for conditioning
                encoder_hidden_states = text_encoder(batch["input_ids"])[0]

                # Get the value embedding for conditioning
                value_encoder_pooled_outputs = value_encoder(batch["valued_input_ids"])[1]
                conditioning_embedding = torch.cat([value_encoder_pooled_outputs.unsqueeze(dim=1), encoder_hidden_states], dim=1).to(dtype=weight_dtype)

                # Get the target for loss depending on the prediction type
                if args.prediction_type is not None:
                    # set prediction_type of scheduler if defined
                    noise_scheduler.register_to_config(prediction_type=args.prediction_type)

                if noise_scheduler.config.prediction_type == "epsilon":
                    target = noise
                    neg_target = noise
                elif noise_scheduler.config.prediction_type == "v_prediction":
                    target = noise_scheduler.get_velocity(latents, noise, timesteps)
                    neg_target = noise_scheduler.get_velocity(neg_latents, noise, timesteps)
                else:
                    raise ValueError(f"Unknown prediction type {noise_scheduler.config.prediction_type}")
                
                # Predict the noise residual and compute loss
                model_pred = unet(noisy_latents, timesteps, conditioning_embedding).sample
                neg_model_pred = unet(neg_noisy_latents, timesteps, conditioning_embedding).sample
                ref_model_pred = unet(noisy_latents, timesteps, encoder_hidden_states).sample
                ref_neg_model_pred = unet(neg_noisy_latents, timesteps, encoder_hidden_states).sample
                
                if args.snr_gamma is None:
                    loss = F.mse_loss(model_pred.float(), target.float(), reduction="none")
                    loss = loss.mean(dim=list(range(1, len(loss.shape))))
                    # loss = loss.mean()

                    neg_loss = F.mse_loss(neg_model_pred.float(), neg_target.float(), reduction="none")
                    neg_loss = neg_loss.mean(dim=list(range(1, len(neg_loss.shape))))

                    ref_loss = F.mse_loss(ref_model_pred.float(), target.float(), reduction="none")
                    ref_loss = ref_loss.mean(dim=list(range(1, len(ref_loss.shape))))

                    ref_neg_loss = F.mse_loss(ref_neg_model_pred.float(), neg_target.float(), reduction="none")
                    ref_neg_loss = ref_neg_loss.mean(dim=list(range(1, len(ref_neg_loss.shape))))
                else:
                    # TODO: maybe could also change this, but not sure if snr_weight works for tuning text encoder
                    # Compute loss-weights as per Section 3.4 of https://arxiv.org/abs/2303.09556.
                    # Since we predict the noise instead of x_0, the original formulation is slightly changed.
                    # This is discussed in Section 4.2 of the same paper.
                    snr = compute_snr(noise_scheduler, timesteps)
                    base_weight = (
                        torch.stack([snr, args.snr_gamma * torch.ones_like(timesteps)], dim=1).min(dim=1)[0] / snr
                    )

                    if noise_scheduler.config.prediction_type == "v_prediction":
                        # Velocity objective needs to be floored to an SNR weight of one.
                        mse_loss_weights = base_weight + 1
                    else:
                        # Epsilon and sample both use the same loss weights.
                        mse_loss_weights = base_weight

                    # For zero-terminal SNR, we have to handle the case where a sigma of Zero results in a Inf value.
                    # When we run this, the MSE loss weights for this timestep is set unconditionally to 1.
                    # If we do not run this, the loss value will go to NaN almost immediately, usually within one step.
                    mse_loss_weights[snr == 0] = 1.0

                    # We first calculate the original loss. Then we mean over the non-batch dimensions and
                    # rebalance the sample-wise losses with their respective loss weights.
                    # Finally, we take the mean of the rebalanced loss.
                    loss = F.mse_loss(model_pred.float(), target.float(), reduction="none")
                    loss = loss.mean(dim=list(range(1, len(loss.shape)))) * mse_loss_weights
                    # loss = loss.mean()

                    neg_loss = F.mse_loss(neg_model_pred.float(), neg_target.float(), reduction="none")
                    neg_loss = neg_loss.mean(dim=list(range(1, len(neg_loss.shape)))) * mse_loss_weights

                    ref_loss = F.mse_loss(ref_model_pred.floaf(), target.float(), reduction="none")
                    ref_loss = ref_loss.mean(dim=list(range(1, len(ref_loss.shape)))) * mse_loss_weights

                    ref_neg_loss = F.mse_loss(ref_neg_model_pred.float(), neg_target.float(), reduction="none")
                    ref_neg_loss = ref_neg_loss.mean(dim=list(range(1, len(ref_neg_loss.shape)))) * mse_loss_weights

                # LiVO loss
                livo_loss =  (torch.clamp(args.livo_gamma_1 + args.livo_beta * (loss - ref_loss), min=0)
                            +  torch.clamp(args.livo_gamma_2 + args.livo_alpha * (ref_neg_loss - neg_loss), min=0)).mean()

                # Gather the losses across all processes for logging (if we use distributed training).
                avg_loss = accelerator.gather(livo_loss.repeat(args.train_batch_size)).mean()
                train_loss += avg_loss.item() / args.gradient_accumulation_steps

                avg_loss_theta_w = accelerator.gather(loss.repeat(args.train_batch_size)).mean()
                loss_theta_w += avg_loss_theta_w.item() / args.gradient_accumulation_steps

                avg_loss_ref_w = accelerator.gather(ref_loss.repeat(args.train_batch_size)).mean()
                loss_ref_w += avg_loss_ref_w.item() / args.gradient_accumulation_steps

                avg_loss_theta_l = accelerator.gather(neg_loss.repeat(args.train_batch_size)).mean()
                loss_theta_l += avg_loss_theta_l.item() / args.gradient_accumulation_steps

                avg_loss_ref_l = accelerator.gather(ref_neg_loss.repeat(args.train_batch_size)).mean()
                loss_ref_l += avg_loss_ref_l.item() / args.gradient_accumulation_steps

                # Backpropagate
                accelerator.backward(livo_loss)
                if accelerator.sync_gradients:
                    accelerator.clip_grad_norm_(value_encoder.parameters(), args.max_grad_norm)
                optimizer.step()
                lr_scheduler.step()
                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
                accelerator.log({
                        "train_loss": train_loss,
                        "loss_theta_w": loss_theta_w,
                        "loss_ref_w": loss_ref_w,
                        "loss_theta_l": loss_theta_l,
                        "loss_ref_l": loss_ref_l,
                        "theta_w - ref_w": loss_theta_w - loss_ref_w,
                        "ref_l - theta_l": loss_ref_l - loss_theta_l,
                        "loss_1": args.livo_beta * (loss_theta_w - loss_ref_w),
                        "loss_2": args.livo_alpha * (loss_ref_l - loss_theta_l),
                        "global_step": global_step,
                        "lr": lr_scheduler.get_last_lr()[0]
                    }, step=global_step)
                train_loss = 0.0
                loss_theta_w, loss_ref_w, loss_theta_l, loss_ref_l = 0.0, 0.0, 0.0, 0.0

                if accelerator.is_main_process:
                    if global_step % args.checkpointing_steps == 0:
                        # _before_ saving state, check if this save would set us over the `checkpoints_total_limit`
                        if args.checkpoints_total_limit is not None:
                            checkpoints = os.listdir(args.output_dir)
                            checkpoints = [d for d in checkpoints if d.startswith("checkpoint")]
                            checkpoints = sorted(checkpoints, key=lambda x: int(x.split("-")[1]))

                            # before we save the new checkpoint, we need to have at _most_ `checkpoints_total_limit - 1` checkpoints
                            if len(checkpoints) >= args.checkpoints_total_limit:
                                num_to_remove = len(checkpoints) - args.checkpoints_total_limit + 1
                                removing_checkpoints = checkpoints[0:num_to_remove]

                                logger.info(
                                    f"{len(checkpoints)} checkpoints already exist, removing {len(removing_checkpoints)} checkpoints"
                                )
                                logger.info(f"removing checkpoints: {', '.join(removing_checkpoints)}")

                                for removing_checkpoint in removing_checkpoints:
                                    removing_checkpoint = os.path.join(args.output_dir, removing_checkpoint)
                                    shutil.rmtree(removing_checkpoint)

                        save_path = os.path.join(args.output_dir, f"checkpoint-{global_step}")
                        accelerator.save_state(save_path)
                        logger.info(f"Saved state to {save_path}")

                    if args.validation_prompts is not None and global_step % args.validation_steps == 0:
                        log_validation(
                            vae,
                            text_encoder,
                            value_encoder,
                            tokenizer,
                            unet,
                            args,
                            accelerator,
                            weight_dtype,
                            global_step,
                        )

            logs = {"step_loss": livo_loss.detach().item(), "lr": lr_scheduler.get_last_lr()[0]}
            progress_bar.set_postfix(**logs)

            if global_step >= args.max_train_steps:
                break
            

    # Create the pipeline using the trained modules and save it.
    accelerator.wait_for_everyone()
    if accelerator.is_main_process:
        value_encoder: CLIPTextModel = accelerator.unwrap_model(value_encoder)
        pipeline = StableDiffusionPipeline.from_pretrained(
            args.pretrained_model_name_or_path,
            text_encoder=text_encoder,
            vae=vae,
            unet=unet,
            safety_checker=None,
            revision=args.revision,
        )

        value_encoder.save_pretrained(os.path.join(args.output_dir, "value_encoder"))

        # Run a final round of inference.
        if args.validation_prompts is not None:
            logger.info("Running inference for collecting generated images...")
            pipeline = pipeline.to(accelerator.device)
            pipeline.torch_dtype = weight_dtype
            pipeline.set_progress_bar_config(disable=True)

            if args.enable_xformers_memory_efficient_attention:
                pipeline.enable_xformers_memory_efficient_attention()

            if args.seed is None:
                generator = None
            else:
                generator = torch.Generator(device=accelerator.device).manual_seed(args.seed)

            for i in range(len(args.validation_prompts)):
                with torch.autocast("cuda"):
                    _, prompt = args.validation_prompts[i].split(', ', 1)
                    input_ids = pipeline.tokenizer(prompt, return_tensors="pt").input_ids
                    prompt_embeds = pipeline.text_encoder(input_ids.to(accelerator.device))[0]
                    valued_input_ids = pipeline.tokenizer(args.validation_prompts[i], return_tensors="pt").input_ids
                    value_embeds = value_encoder(valued_input_ids.to(accelerator.device))[1]
                    concat_embeds = torch.cat([value_embeds.unsqueeze(1), prompt_embeds], dim=1)
                    images = pipeline(prompt_embeds=concat_embeds, num_images_per_prompt=args.num_images_per_val_prompt, generator=generator).images
                    for idx, image in enumerate(images):
                        os.makedirs(os.path.join(args.output_dir, 'final_val_imgs'), exist_ok=True)
                        image.save(os.path.join(args.output_dir, 'final_val_imgs', f"image_{i}_{idx}.jpg"))

    accelerator.end_training()


if __name__ == '__main__':
    main()