latte

models/utils.py

+# adopted from
+# https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py
+# and
+# https://github.com/lucidrains/denoising-diffusion-pytorch/blob/7706bdfc6f527f58d33f84b7b522e61e6e3164b3/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py
+# and
+# https://github.com/openai/guided-diffusion/blob/0ba878e517b276c45d1195eb29f6f5f72659a05b/guided_diffusion/nn.py
+#
+# thanks!
+
+
+import os
+import math
+import torch
+
+import numpy as np
+import torch.nn as nn
+
+from einops import repeat
+
+
+#################################################################################
+#                                  Unet Utils                                   #
+#################################################################################
+
+def checkpoint(func, inputs, params, flag):
+    """
+    Evaluate a function without caching intermediate activations, allowing for
+    reduced memory at the expense of extra compute in the backward pass.
+    :param func: the function to evaluate.
+    :param inputs: the argument sequence to pass to `func`.
+    :param params: a sequence of parameters `func` depends on but does not
+                   explicitly take as arguments.
+    :param flag: if False, disable gradient checkpointing.
+    """
+    if flag:
+        args = tuple(inputs) + tuple(params)
+        return CheckpointFunction.apply(func, len(inputs), *args)
+    else:
+        return func(*inputs)
+
+
+class CheckpointFunction(torch.autograd.Function):
+    @staticmethod
+    def forward(ctx, run_function, length, *args):
+        ctx.run_function = run_function
+        ctx.input_tensors = list(args[:length])
+        ctx.input_params = list(args[length:])
+
+        with torch.no_grad():
+            output_tensors = ctx.run_function(*ctx.input_tensors)
+        return output_tensors
+
+    @staticmethod
+    def backward(ctx, *output_grads):
+        ctx.input_tensors = [x.detach().requires_grad_(True) for x in ctx.input_tensors]
+        with torch.enable_grad():
+            # Fixes a bug where the first op in run_function modifies the
+            # Tensor storage in place, which is not allowed for detach()'d
+            # Tensors.
+            shallow_copies = [x.view_as(x) for x in ctx.input_tensors]
+            output_tensors = ctx.run_function(*shallow_copies)
+        input_grads = torch.autograd.grad(
+            output_tensors,
+            ctx.input_tensors + ctx.input_params,
+            output_grads,
+            allow_unused=True,
+        )
+        del ctx.input_tensors
+        del ctx.input_params
+        del output_tensors
+        return (None, None) + input_grads
+
+
+def timestep_embedding(timesteps, dim, max_period=10000, repeat_only=False):
+    """
+    Create sinusoidal timestep embeddings.
+    :param timesteps: a 1-D Tensor of N indices, one per batch element.
+                      These may be fractional.
+    :param dim: the dimension of the output.
+    :param max_period: controls the minimum frequency of the embeddings.
+    :return: an [N x dim] Tensor of positional embeddings.
+    """
+    if not repeat_only:
+        half = dim // 2
+        freqs = torch.exp(
+            -math.log(max_period) * torch.arange(start=0, end=half, dtype=torch.float32) / half
+        ).to(device=timesteps.device)
+        args = timesteps[:, None].float() * freqs[None]
+        embedding = torch.cat([torch.cos(args), torch.sin(args)], dim=-1)
+        if dim % 2:
+            embedding = torch.cat([embedding, torch.zeros_like(embedding[:, :1])], dim=-1)
+    else:
+        embedding = repeat(timesteps, 'b -> b d', d=dim).contiguous()
+    return embedding
+
+
+def zero_module(module):
+    """
+    Zero out the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().zero_()
+    return module
+
+
+def scale_module(module, scale):
+    """
+    Scale the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().mul_(scale)
+    return module
+
+
+def mean_flat(tensor):
+    """
+    Take the mean over all non-batch dimensions.
+    """
+    return tensor.mean(dim=list(range(1, len(tensor.shape))))
+
+
+def normalization(channels):
+    """
+    Make a standard normalization layer.
+    :param channels: number of input channels.
+    :return: an nn.Module for normalization.
+    """
+    return GroupNorm32(32, channels)
+
+
+# PyTorch 1.7 has SiLU, but we support PyTorch 1.5.
+class SiLU(nn.Module):
+    def forward(self, x):
+        return x * torch.sigmoid(x)
+
+
+class GroupNorm32(nn.GroupNorm):
+    def forward(self, x):
+        return super().forward(x.float()).type(x.dtype)
+
+def conv_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D convolution module.
+    """
+    if dims == 1:
+        return nn.Conv1d(*args, **kwargs)
+    elif dims == 2:
+        return nn.Conv2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.Conv3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+def linear(*args, **kwargs):
+    """
+    Create a linear module.
+    """
+    return nn.Linear(*args, **kwargs)
+
+
+def avg_pool_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D average pooling module.
+    """
+    if dims == 1:
+        return nn.AvgPool1d(*args, **kwargs)
+    elif dims == 2:
+        return nn.AvgPool2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.AvgPool3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+# class HybridConditioner(nn.Module):
+
+#     def __init__(self, c_concat_config, c_crossattn_config):
+#         super().__init__()
+#         self.concat_conditioner = instantiate_from_config(c_concat_config)
+#         self.crossattn_conditioner = instantiate_from_config(c_crossattn_config)
+
+#     def forward(self, c_concat, c_crossattn):
+#         c_concat = self.concat_conditioner(c_concat)
+#         c_crossattn = self.crossattn_conditioner(c_crossattn)
+#         return {'c_concat': [c_concat], 'c_crossattn': [c_crossattn]}
+
+
+def noise_like(shape, device, repeat=False):
+    repeat_noise = lambda: torch.randn((1, *shape[1:]), device=device).repeat(shape[0], *((1,) * (len(shape) - 1)))
+    noise = lambda: torch.randn(shape, device=device)
+    return repeat_noise() if repeat else noise()
+
+def count_flops_attn(model, _x, y):
+    """
+    A counter for the `thop` package to count the operations in an
+    attention operation.
+    Meant to be used like:
+        macs, params = thop.profile(
+            model,
+            inputs=(inputs, timestamps),
+            custom_ops={QKVAttention: QKVAttention.count_flops},
+        )
+    """
+    b, c, *spatial = y[0].shape
+    num_spatial = int(np.prod(spatial))
+    # We perform two matmuls with the same number of ops.
+    # The first computes the weight matrix, the second computes
+    # the combination of the value vectors.
+    matmul_ops = 2 * b * (num_spatial ** 2) * c
+    model.total_ops += torch.DoubleTensor([matmul_ops])
+
+def count_params(model, verbose=False):
+    total_params = sum(p.numel() for p in model.parameters())
+    if verbose:
+        print(f"{model.__class__.__name__} has {total_params * 1.e-6:.2f} M params.")
+    return total_params
\ No newline at end of file

requirements.txt

+#torch > 2.0.0
+# torchvision == 0.16 --no-deps
+# timm  --no-deps
+diffusers==0.24.0
+accelerate
+tensorboard
+einops
+transformers
+av
+scikit-image
+decord
+pandas
+omegaconf
+imageio==2.27.0
+imageio-ffmpeg==0.4.9
+pillow

sample/ffs.sh

+#!/bin/bash
+# export CUDA_VISIBLE_DEVICES=1
+
+python sample/sample.py \
+--config ./configs/ffs/ffs_sample.yaml \
+--ckpt ./share_ckpts/ffs.pt \
+--save_video_path ./test
\ No newline at end of file

sample/ffs_ddp.sh

+#!/bin/bash
+
+torchrun --nnodes=1 --nproc_per_node=2 sample/sample_ddp.py \
+--config ./configs/ffs/ffs_sample.yaml \
+--ckpt ./share_ckpts/ffs.pt \
+--save_video_path ./test

sample/pipeline_videogen.py

+# Copyright 2023 PixArt-Alpha Authors and The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import html
+import inspect
+import re
+import urllib.parse as ul
+from typing import Callable, List, Optional, Tuple, Union
+
+import torch
+import einops
+from transformers import T5EncoderModel, T5Tokenizer
+
+from diffusers.image_processor import VaeImageProcessor
+from diffusers.models import AutoencoderKL, Transformer2DModel
+from diffusers.schedulers import DPMSolverMultistepScheduler
+from diffusers.utils import (
+    BACKENDS_MAPPING,
+    is_bs4_available,
+    is_ftfy_available,
+    logging,
+    replace_example_docstring,
+)
+from diffusers.utils.torch_utils import randn_tensor
+from diffusers.pipelines.pipeline_utils import DiffusionPipeline
+from diffusers.utils import BaseOutput
+from dataclasses import dataclass
+
+
+logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
+
+if is_bs4_available():
+    from bs4 import BeautifulSoup
+
+if is_ftfy_available():
+    import ftfy
+
+
+EXAMPLE_DOC_STRING = """
+    Examples:
+        ```py
+        >>> import torch
+        >>> from diffusers import PixArtAlphaPipeline
+
+        >>> # You can replace the checkpoint id with "PixArt-alpha/PixArt-XL-2-512x512" too.
+        >>> pipe = PixArtAlphaPipeline.from_pretrained("PixArt-alpha/PixArt-XL-2-1024-MS", torch_dtype=torch.float16)
+        >>> # Enable memory optimizations.
+        >>> pipe.enable_model_cpu_offload()
+
+        >>> prompt = "A small cactus with a happy face in the Sahara desert."
+        >>> image = pipe(prompt).images[0]
+        ```
+"""
+
+@dataclass
+class VideoPipelineOutput(BaseOutput):
+    video: torch.Tensor
+
+
+class VideoGenPipeline(DiffusionPipeline):
+    r"""
+    Pipeline for text-to-image generation using PixArt-Alpha.
+
+    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods the
+    library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
+
+    Args:
+        vae ([`AutoencoderKL`]):
+            Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
+        text_encoder ([`T5EncoderModel`]):
+            Frozen text-encoder. PixArt-Alpha uses
+            [T5](https://huggingface.co/docs/transformers/model_doc/t5#transformers.T5EncoderModel), specifically the
+            [t5-v1_1-xxl](https://huggingface.co/PixArt-alpha/PixArt-alpha/tree/main/t5-v1_1-xxl) variant.
+        tokenizer (`T5Tokenizer`):
+            Tokenizer of class
+            [T5Tokenizer](https://huggingface.co/docs/transformers/model_doc/t5#transformers.T5Tokenizer).
+        transformer ([`Transformer2DModel`]):
+            A text conditioned `Transformer2DModel` to denoise the encoded image latents.
+        scheduler ([`SchedulerMixin`]):
+            A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
+    """
+    bad_punct_regex = re.compile(
+        r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
+    )  # noqa
+
+    _optional_components = ["tokenizer", "text_encoder"]
+    model_cpu_offload_seq = "text_encoder->transformer->vae"
+
+    def __init__(
+        self,
+        tokenizer: T5Tokenizer,
+        text_encoder: T5EncoderModel,
+        vae: AutoencoderKL,
+        transformer: Transformer2DModel,
+        scheduler: DPMSolverMultistepScheduler,
+    ):
+        super().__init__()
+
+        self.register_modules(
+            tokenizer=tokenizer, text_encoder=text_encoder, vae=vae, transformer=transformer, scheduler=scheduler
+        )
+
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
+
+    # Adapted from https://github.com/PixArt-alpha/PixArt-alpha/blob/master/diffusion/model/utils.py
+    def mask_text_embeddings(self, emb, mask):
+        if emb.shape[0] == 1:
+            keep_index = mask.sum().item()
+            return emb[:, :, :keep_index, :], keep_index # 1, 120, 4096 -> 1 7 4096
+        else:
+            masked_feature = emb * mask[:, None, :, None] # 1 120 4096
+            return masked_feature, emb.shape[2]
+
+    # Adapted from diffusers.pipelines.deepfloyd_if.pipeline_if.encode_prompt
+    def encode_prompt(
+        self,
+        prompt: Union[str, List[str]],
+        do_classifier_free_guidance: bool = True,
+        negative_prompt: str = "",
+        num_images_per_prompt: int = 1,
+        device: Optional[torch.device] = None,
+        prompt_embeds: Optional[torch.FloatTensor] = None,
+        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        clean_caption: bool = False,
+        mask_feature: bool = True,
+    ):
+        r"""
+        Encodes the prompt into text encoder hidden states.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                prompt to be encoded
+            negative_prompt (`str` or `List[str]`, *optional*):
+                The prompt not to guide the image generation. If not defined, one has to pass `negative_prompt_embeds`
+                instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`). For
+                PixArt-Alpha, this should be "".
+            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
+                whether to use classifier free guidance or not
+            num_images_per_prompt (`int`, *optional*, defaults to 1):
+                number of images that should be generated per prompt
+            device: (`torch.device`, *optional*):
+                torch device to place the resulting embeddings on
+            prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
+                provided, text embeddings will be generated from `prompt` input argument.
+            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated negative text embeddings. For PixArt-Alpha, it's should be the embeddings of the ""
+                string.
+            clean_caption (bool, defaults to `False`):
+                If `True`, the function will preprocess and clean the provided caption before encoding.
+            mask_feature: (bool, defaults to `True`):
+                If `True`, the function will mask the text embeddings.
+        """
+        embeds_initially_provided = prompt_embeds is not None and negative_prompt_embeds is not None
+
+        if device is None:
+            device = self._execution_device
+
+        if prompt is not None and isinstance(prompt, str):
+            batch_size = 1
+        elif prompt is not None and isinstance(prompt, list):
+            batch_size = len(prompt)
+        else:
+            batch_size = prompt_embeds.shape[0]
+
+        # See Section 3.1. of the paper.
+        max_length = 120
+
+        if prompt_embeds is None:
+            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
+            text_inputs = self.tokenizer(
+                prompt,
+                padding="max_length",
+                max_length=max_length,
+                truncation=True,
+                return_attention_mask=True,
+                add_special_tokens=True,
+                return_tensors="pt",
+            )
+            text_input_ids = text_inputs.input_ids
+            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids
+
+            if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
+                text_input_ids, untruncated_ids
+            ):
+                removed_text = self.tokenizer.batch_decode(untruncated_ids[:, max_length - 1 : -1])
+                logger.warning(
+                    "The following part of your input was truncated because CLIP can only handle sequences up to"
+                    f" {max_length} tokens: {removed_text}"
+                )
+
+            attention_mask = text_inputs.attention_mask.to(device)
+            prompt_embeds_attention_mask = attention_mask
+
+            prompt_embeds = self.text_encoder(text_input_ids.to(device), attention_mask=attention_mask)
+            prompt_embeds = prompt_embeds[0]
+        else:
+            prompt_embeds_attention_mask = torch.ones_like(prompt_embeds)
+
+        if self.text_encoder is not None:
+            dtype = self.text_encoder.dtype
+        elif self.transformer is not None:
+            dtype = self.transformer.dtype
+        else:
+            dtype = None
+
+        prompt_embeds = prompt_embeds.to(dtype=dtype, device=device)
+
+        bs_embed, seq_len, _ = prompt_embeds.shape
+        # duplicate text embeddings and attention mask for each generation per prompt, using mps friendly method
+        prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
+        prompt_embeds_attention_mask = prompt_embeds_attention_mask.view(bs_embed, -1)
+        prompt_embeds_attention_mask = prompt_embeds_attention_mask.repeat(num_images_per_prompt, 1)
+
+        # get unconditional embeddings for classifier free guidance
+        if do_classifier_free_guidance and negative_prompt_embeds is None:
+            uncond_tokens = [negative_prompt] * batch_size
+            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
+            max_length = prompt_embeds.shape[1]
+            uncond_input = self.tokenizer(
+                uncond_tokens,
+                padding="max_length",
+                max_length=max_length,
+                truncation=True,
+                return_attention_mask=True,
+                add_special_tokens=True,
+                return_tensors="pt",
+            )
+            attention_mask = uncond_input.attention_mask.to(device)
+
+            negative_prompt_embeds = self.text_encoder(
+                uncond_input.input_ids.to(device),
+                attention_mask=attention_mask,
+            )
+            negative_prompt_embeds = negative_prompt_embeds[0]
+
+        if do_classifier_free_guidance:
+            # duplicate unconditional embeddings for each generation per prompt, using mps friendly method
+            seq_len = negative_prompt_embeds.shape[1]
+
+            negative_prompt_embeds = negative_prompt_embeds.to(dtype=dtype, device=device)
+
+            negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+            negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
+
+            # For classifier free guidance, we need to do two forward passes.
+            # Here we concatenate the unconditional and text embeddings into a single batch
+            # to avoid doing two forward passes
+        else:
+            negative_prompt_embeds = None
+
+        # print(prompt_embeds.shape) # 1 120 4096
+        # print(negative_prompt_embeds.shape) # 1 120 4096
+
+        # Perform additional masking.
+        if mask_feature and not embeds_initially_provided:
+            prompt_embeds = prompt_embeds.unsqueeze(1)
+            masked_prompt_embeds, keep_indices = self.mask_text_embeddings(prompt_embeds, prompt_embeds_attention_mask)
+            masked_prompt_embeds = masked_prompt_embeds.squeeze(1)
+            masked_negative_prompt_embeds = (
+                negative_prompt_embeds[:, :keep_indices, :] if negative_prompt_embeds is not None else None
+            )
+
+            print(masked_prompt_embeds.shape) # 1 120 4096 第二个维度会变
+            print(masked_negative_prompt_embeds.shape) # 1 120 4096 第二个维度会变
+
+            # import torch.nn.functional as F
+
+            # padding = (0, 0, 0, 113)  # (左, 右, 下, 上)
+            # masked_prompt_embeds_ = F.pad(masked_prompt_embeds, padding, "constant", 0)
+            # masked_negative_prompt_embeds_ = F.pad(masked_negative_prompt_embeds, padding, "constant", 0)
+
+            # print(masked_prompt_embeds == masked_prompt_embeds_[:, :masked_negative_prompt_embeds.shape[1], ...])
+
+            return masked_prompt_embeds, masked_negative_prompt_embeds
+            # return masked_prompt_embeds_, masked_negative_prompt_embeds_
+        
+        return prompt_embeds, negative_prompt_embeds
+
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs
+    def prepare_extra_step_kwargs(self, generator, eta):
+        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
+        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
+        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # and should be between [0, 1]
+
+        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
+        extra_step_kwargs = {}
+        if accepts_eta:
+            extra_step_kwargs["eta"] = eta
+
+        # check if the scheduler accepts generator
+        accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
+        if accepts_generator:
+            extra_step_kwargs["generator"] = generator
+        return extra_step_kwargs
+
+    def check_inputs(
+        self,
+        prompt,
+        height,
+        width,
+        negative_prompt,
+        callback_steps,
+        prompt_embeds=None,
+        negative_prompt_embeds=None,
+    ):
+        if height % 8 != 0 or width % 8 != 0:
+            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
+
+        if (callback_steps is None) or (
+            callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0)
+        ):
+            raise ValueError(
+                f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
+                f" {type(callback_steps)}."
+            )
+
+        if prompt is not None and prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
+                " only forward one of the two."
+            )
+        elif prompt is None and prompt_embeds is None:
+            raise ValueError(
+                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
+            )
+        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
+            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
+
+        if prompt is not None and negative_prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `prompt`: {prompt} and `negative_prompt_embeds`:"
+                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
+            )
+
+        if negative_prompt is not None and negative_prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:"
+                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
+            )
+
+        if prompt_embeds is not None and negative_prompt_embeds is not None:
+            if prompt_embeds.shape != negative_prompt_embeds.shape:
+                raise ValueError(
+                    "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but"
+                    f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
+                    f" {negative_prompt_embeds.shape}."
+                )
+
+    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
+    def _text_preprocessing(self, text, clean_caption=False):
+        if clean_caption and not is_bs4_available():
+            logger.warn(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
+            logger.warn("Setting `clean_caption` to False...")
+            clean_caption = False
+
+        if clean_caption and not is_ftfy_available():
+            logger.warn(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
+            logger.warn("Setting `clean_caption` to False...")
+            clean_caption = False
+
+        if not isinstance(text, (tuple, list)):
+            text = [text]
+
+        def process(text: str):
+            if clean_caption:
+                text = self._clean_caption(text)
+                text = self._clean_caption(text)
+            else:
+                text = text.lower().strip()
+            return text
+
+        return [process(t) for t in text]
+
+    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
+    def _clean_caption(self, caption):
+        caption = str(caption)
+        caption = ul.unquote_plus(caption)
+        caption = caption.strip().lower()
+        caption = re.sub("<person>", "person", caption)
+        # urls:
+        caption = re.sub(
+            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
+            "",
+            caption,
+        )  # regex for urls
+        caption = re.sub(
+            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
+            "",
+            caption,
+        )  # regex for urls
+        # html:
+        caption = BeautifulSoup(caption, features="html.parser").text
+
+        # @<nickname>
+        caption = re.sub(r"@[\w\d]+\b", "", caption)
+
+        # 31C0—31EF CJK Strokes
+        # 31F0—31FF Katakana Phonetic Extensions
+        # 3200—32FF Enclosed CJK Letters and Months
+        # 3300—33FF CJK Compatibility
+        # 3400—4DBF CJK Unified Ideographs Extension A
+        # 4DC0—4DFF Yijing Hexagram Symbols
+        # 4E00—9FFF CJK Unified Ideographs
+        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
+        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
+        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
+        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
+        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
+        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
+        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
+        #######################################################
+
+        # все виды тире / all types of dash --> "-"
+        caption = re.sub(
+            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
+            "-",
+            caption,
+        )
+
+        # кавычки к одному стандарту
+        caption = re.sub(r"[`´«»“”¨]", '"', caption)
+        caption = re.sub(r"[‘’]", "'", caption)
+
+        # &quot;
+        caption = re.sub(r"&quot;?", "", caption)
+        # &amp
+        caption = re.sub(r"&amp", "", caption)
+
+        # ip adresses:
+        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)
+
+        # article ids:
+        caption = re.sub(r"\d:\d\d\s+$", "", caption)
+
+        # \n
+        caption = re.sub(r"\\n", " ", caption)
+
+        # "#123"
+        caption = re.sub(r"#\d{1,3}\b", "", caption)
+        # "#12345.."
+        caption = re.sub(r"#\d{5,}\b", "", caption)
+        # "123456.."
+        caption = re.sub(r"\b\d{6,}\b", "", caption)
+        # filenames:
+        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)
+
+        #
+        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
+        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""
+
+        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
+        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "
+
+        # this-is-my-cute-cat / this_is_my_cute_cat
+        regex2 = re.compile(r"(?:\-|\_)")
+        if len(re.findall(regex2, caption)) > 3:
+            caption = re.sub(regex2, " ", caption)
+
+        caption = ftfy.fix_text(caption)
+        caption = html.unescape(html.unescape(caption))
+
+        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
+        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
+        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231
+
+        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
+        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
+        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
+        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
+        caption = re.sub(r"\bpage\s+\d+\b", "", caption)
+
+        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...
+
+        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)
+
+        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
+        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
+        caption = re.sub(r"\s+", " ", caption)
+
+        caption.strip()
+
+        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
+        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
+        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
+        caption = re.sub(r"^\.\S+$", "", caption)
+
+        return caption.strip()
+
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
+    def prepare_latents(self, batch_size, num_channels_latents, video_length, height, width, dtype, device, generator, latents=None):
+        shape = (batch_size, num_channels_latents, video_length, height // self.vae_scale_factor, width // self.vae_scale_factor)
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
+
+        if latents is None:
+            latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
+        else:
+            latents = latents.to(device)
+
+        # scale the initial noise by the standard deviation required by the scheduler
+        latents = latents * self.scheduler.init_noise_sigma
+        return latents
+
+    @torch.no_grad()
+    @replace_example_docstring(EXAMPLE_DOC_STRING)
+    def __call__(
+        self,
+        prompt: Union[str, List[str]] = None,
+        negative_prompt: str = "",
+        num_inference_steps: int = 20,
+        timesteps: List[int] = None,
+        guidance_scale: float = 4.5,
+        num_images_per_prompt: Optional[int] = 1,
+        video_length: Optional[int] = None,
+        height: Optional[int] = None,
+        width: Optional[int] = None,
+        eta: float = 0.0,
+        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
+        latents: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.FloatTensor] = None,
+        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        output_type: Optional[str] = "pil",
+        return_dict: bool = True,
+        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback_steps: int = 1,
+        clean_caption: bool = True,
+        mask_feature: bool = True,
+        enable_temporal_attentions: bool = True,
+    ) -> Union[VideoPipelineOutput, Tuple]:
+        """
+        Function invoked when calling the pipeline for generation.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
+                instead.
+            negative_prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts not to guide the image generation. If not defined, one has to pass
+                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
+                less than `1`).
+            num_inference_steps (`int`, *optional*, defaults to 100):
+                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
+                expense of slower inference.
+            timesteps (`List[int]`, *optional*):
+                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
+                timesteps are used. Must be in descending order.
+            guidance_scale (`float`, *optional*, defaults to 7.0):
+                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
+                `guidance_scale` is defined as `w` of equation 2. of [Imagen
+                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
+                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
+                usually at the expense of lower image quality.
+            num_images_per_prompt (`int`, *optional*, defaults to 1):
+                The number of images to generate per prompt.
+            height (`int`, *optional*, defaults to self.unet.config.sample_size):
+                The height in pixels of the generated image.
+            width (`int`, *optional*, defaults to self.unet.config.sample_size):
+                The width in pixels of the generated image.
+            eta (`float`, *optional*, defaults to 0.0):
+                Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
+                [`schedulers.DDIMScheduler`], will be ignored for others.
+            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
+                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
+                to make generation deterministic.
+            latents (`torch.FloatTensor`, *optional*):
+                Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
+                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
+                tensor will ge generated by sampling using the supplied random `generator`.
+            prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
+                provided, text embeddings will be generated from `prompt` input argument.
+            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated negative text embeddings. For PixArt-Alpha this negative prompt should be "". If not
+                provided, negative_prompt_embeds will be generated from `negative_prompt` input argument.
+            output_type (`str`, *optional*, defaults to `"pil"`):
+                The output format of the generate image. Choose between
+                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
+            return_dict (`bool`, *optional*, defaults to `True`):
+                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
+            callback (`Callable`, *optional*):
+                A function that will be called every `callback_steps` steps during inference. The function will be
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+            callback_steps (`int`, *optional*, defaults to 1):
+                The frequency at which the `callback` function will be called. If not specified, the callback will be
+                called at every step.
+            clean_caption (`bool`, *optional*, defaults to `True`):
+                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
+                be installed. If the dependencies are not installed, the embeddings will be created from the raw
+                prompt.
+            mask_feature (`bool` defaults to `True`): If set to `True`, the text embeddings will be masked.
+
+        Examples:
+
+        Returns:
+            [`~pipelines.ImagePipelineOutput`] or `tuple`:
+                If `return_dict` is `True`, [`~pipelines.ImagePipelineOutput`] is returned, otherwise a `tuple` is
+                returned where the first element is a list with the generated images
+        """
+        # 1. Check inputs. Raise error if not correct
+        height = height or self.transformer.config.sample_size * self.vae_scale_factor
+        width = width or self.transformer.config.sample_size * self.vae_scale_factor
+        self.check_inputs(
+            prompt, height, width, negative_prompt, callback_steps, prompt_embeds, negative_prompt_embeds
+        )
+
+        # 2. Default height and width to transformer
+        if prompt is not None and isinstance(prompt, str):
+            batch_size = 1
+        elif prompt is not None and isinstance(prompt, list):
+            batch_size = len(prompt)
+        else:
+            batch_size = prompt_embeds.shape[0]
+
+        device = self._execution_device
+
+        # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
+        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # corresponds to doing no classifier free guidance.
+        do_classifier_free_guidance = guidance_scale > 1.0
+
+        # 3. Encode input prompt
+        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
+            prompt,
+            do_classifier_free_guidance,
+            negative_prompt=negative_prompt,
+            num_images_per_prompt=num_images_per_prompt,
+            device=device,
+            prompt_embeds=prompt_embeds,
+            negative_prompt_embeds=negative_prompt_embeds,
+            clean_caption=clean_caption,
+            mask_feature=mask_feature,
+        )
+        if do_classifier_free_guidance:
+            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
+
+        # 4. Prepare timesteps
+        self.scheduler.set_timesteps(num_inference_steps, device=device)
+        timesteps = self.scheduler.timesteps
+
+        # 5. Prepare latents.
+        latent_channels = self.transformer.config.in_channels
+        latents = self.prepare_latents(
+            batch_size * num_images_per_prompt,
+            latent_channels,
+            video_length,
+            height,
+            width,
+            prompt_embeds.dtype,
+            device,
+            generator,
+            latents,
+        )
+
+        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
+        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)
+
+        # 6.1 Prepare micro-conditions.
+        added_cond_kwargs = {"resolution": None, "aspect_ratio": None}
+        if self.transformer.config.sample_size == 128:
+            resolution = torch.tensor([height, width]).repeat(batch_size * num_images_per_prompt, 1)
+            aspect_ratio = torch.tensor([float(height / width)]).repeat(batch_size * num_images_per_prompt, 1)
+            resolution = resolution.to(dtype=prompt_embeds.dtype, device=device)
+            aspect_ratio = aspect_ratio.to(dtype=prompt_embeds.dtype, device=device)
+            added_cond_kwargs = {"resolution": resolution, "aspect_ratio": aspect_ratio}
+
+        # 7. Denoising loop
+        num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
+
+        with self.progress_bar(total=num_inference_steps) as progress_bar:
+            for i, t in enumerate(timesteps):
+                latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
+                latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
+
+                current_timestep = t
+                if not torch.is_tensor(current_timestep):
+                    # TODO: this requires sync between CPU and GPU. So try to pass timesteps as tensors if you can
+                    # This would be a good case for the `match` statement (Python 3.10+)
+                    is_mps = latent_model_input.device.type == "mps"
+                    if isinstance(current_timestep, float):
+                        dtype = torch.float32 if is_mps else torch.float64
+                    else:
+                        dtype = torch.int32 if is_mps else torch.int64
+                    current_timestep = torch.tensor([current_timestep], dtype=dtype, device=latent_model_input.device)
+                elif len(current_timestep.shape) == 0:
+                    current_timestep = current_timestep[None].to(latent_model_input.device)
+                # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
+                current_timestep = current_timestep.expand(latent_model_input.shape[0])
+
+                # predict noise model_output
+                noise_pred = self.transformer(
+                    latent_model_input,
+                    encoder_hidden_states=prompt_embeds,
+                    timestep=current_timestep,
+                    added_cond_kwargs=added_cond_kwargs,
+                    enable_temporal_attentions=enable_temporal_attentions,
+                    return_dict=False,
+                )[0]
+
+                # perform guidance
+                if do_classifier_free_guidance:
+                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
+                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
+
+                # learned sigma
+                if self.transformer.config.out_channels // 2 == latent_channels:
+                    noise_pred = noise_pred.chunk(2, dim=1)[0]
+                else:
+                    noise_pred = noise_pred
+
+                # compute previous image: x_t -> x_t-1
+                latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
+
+                # call the callback, if provided
+                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
+                    progress_bar.update()
+                    if callback is not None and i % callback_steps == 0:
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
+
+        # if not output_type == "latent":
+        #     image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
+        # else:
+        #     image = latents
+
+        # if not output_type == "latent":
+        #     image = self.image_processor.postprocess(image, output_type=output_type)
+
+        if not output_type == 'latents':
+            video = self.decode_latents(latents)
+        else:
+            video = latents
+            return VideoPipelineOutput(video=video)
+
+        # Offload all models
+        self.maybe_free_model_hooks()
+
+        if not return_dict:
+            return (video,)
+
+        return VideoPipelineOutput(video=video)
+    
+    def decode_latents(self, latents):
+        video_length = latents.shape[2]
+        latents = 1 / self.vae.config.scaling_factor * latents
+        latents = einops.rearrange(latents, "b c f h w -> (b f) c h w")
+        video = []
+        for frame_idx in range(latents.shape[0]):
+            video.append(self.vae.decode(
+                latents[frame_idx:frame_idx+1]).sample)
+        video = torch.cat(video)
+        video = einops.rearrange(video, "(b f) c h w -> b f h w c", f=video_length)
+        video = ((video / 2.0 + 0.5).clamp(0, 1) * 255).to(dtype=torch.uint8).cpu().contiguous()
+        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloa16
+        return video
+

sample/sample.py

+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""
+Sample new images from a pre-trained Latte.
+"""
+import os
+import sys
+try:
+    import utils
+
+    from diffusion import create_diffusion
+    from download import find_model
+except:
+    sys.path.append(os.path.split(sys.path[0])[0])
+
+    import utils
+
+    from diffusion import create_diffusion
+    from download import find_model
+
+import torch
+import argparse
+import torchvision
+
+from einops import rearrange
+from models import get_models
+from torchvision.utils import save_image
+from diffusers.models import AutoencoderKL
+from models.clip import TextEmbedder
+import imageio
+from omegaconf import OmegaConf
+
+torch.backends.cuda.matmul.allow_tf32 = True
+torch.backends.cudnn.allow_tf32 = True
+
+def main(args):
+    # Setup PyTorch:
+    # torch.manual_seed(args.seed)
+    torch.set_grad_enabled(False)
+    device = torch.device("cuda" if torch.cuda.is_available()  else "cpu")
+    print(device)
+    # device = "cpu"
+
+    if args.ckpt is None:
+        assert args.model == "Latte-XL/2", "Only Latte-XL/2 models are available for auto-download."
+        assert args.image_size in [256, 512]
+        assert args.num_classes == 1000
+
+    using_cfg = args.cfg_scale > 1.0
+
+    # Load model:
+    latent_size = args.image_size // 8
+    args.latent_size = latent_size
+    model = get_models(args).to(device)
+
+    if args.use_compile:
+        model = torch.compile(model)
+
+    # a pre-trained model or load a custom Latte checkpoint from train.py:
+    ckpt_path = args.ckpt
+    state_dict = find_model(ckpt_path)
+    model.load_state_dict(state_dict)
+
+    model.eval()  # important!
+    diffusion = create_diffusion(str(args.num_sampling_steps))
+    # vae = AutoencoderKL.from_pretrained(f"stabilityai/sd-vae-ft-{args.vae}").to(device)
+    vae = AutoencoderKL.from_pretrained(args.pretrained_model_path, subfolder="vae").to(device)
+    # text_encoder = TextEmbedder().to(device)
+
+    if args.use_fp16:
+        print('WARNING: using half percision for inferencing!')
+        vae.to(dtype=torch.float16)
+        model.to(dtype=torch.float16)
+        # text_encoder.to(dtype=torch.float16)
+
+    # Labels to condition the model with (feel free to change):
+
+    # Create sampling noise:
+    if args.use_fp16:
+        z = torch.randn(1, args.num_frames, 4, latent_size, latent_size, dtype=torch.float16, device=device) # b c f h w
+    else:
+        z = torch.randn(1, args.num_frames, 4, latent_size, latent_size, device=device)
+
+    # Setup classifier-free guidance:
+    # z = torch.cat([z, z], 0)
+    if using_cfg:
+        z = torch.cat([z, z], 0)
+        y = torch.randint(0, args.num_classes, (1,), device=device)
+        y_null = torch.tensor([101] * 1, device=device)
+        y = torch.cat([y, y_null], dim=0)
+        model_kwargs = dict(y=y, cfg_scale=args.cfg_scale, use_fp16=args.use_fp16)
+        sample_fn = model.forward_with_cfg
+    else:
+        sample_fn = model.forward
+        model_kwargs = dict(y=None, use_fp16=args.use_fp16)
+
+    # Sample images:
+    if args.sample_method == 'ddim':
+        samples = diffusion.ddim_sample_loop(
+            sample_fn, z.shape, z, clip_denoised=False, model_kwargs=model_kwargs, progress=True, device=device
+        )
+    elif args.sample_method == 'ddpm':
+        samples = diffusion.p_sample_loop(
+            sample_fn, z.shape, z, clip_denoised=False, model_kwargs=model_kwargs, progress=True, device=device
+        )
+
+    print(samples.shape)
+    if args.use_fp16:
+        samples = samples.to(dtype=torch.float16)
+    b, f, c, h, w = samples.shape
+    samples = rearrange(samples, 'b f c h w -> (b f) c h w')
+    samples = vae.decode(samples / 0.18215).sample
+    samples = rearrange(samples, '(b f) c h w -> b f c h w', b=b)
+    # Save and display images:
+
+    if not os.path.exists(args.save_video_path):
+        os.makedirs(args.save_video_path)
+
+
+    video_ = ((samples[0] * 0.5 + 0.5) * 255).add_(0.5).clamp_(0, 255).to(dtype=torch.uint8).cpu().permute(0, 2, 3, 1).contiguous()
+    video_save_path = os.path.join(args.save_video_path, 'sample' + '.mp4')
+    print(video_save_path)
+    imageio.mimwrite(video_save_path, video_, fps=8, quality=9)
+    print('save path {}'.format(args.save_video_path))
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--config", type=str, default="./configs/ucf101/ucf101_sample.yaml")
+    parser.add_argument("--ckpt", type=str, default="")
+    parser.add_argument("--save_video_path", type=str, default="./sample_videos/")
+    args = parser.parse_args()
+    omega_conf = OmegaConf.load(args.config)
+    omega_conf.ckpt = args.ckpt
+    omega_conf.save_video_path = args.save_video_path
+    main(omega_conf)

sample/sample_ddp.py

+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""
+Samples a large number of images from a pre-trained Latte model using DDP.
+Subsequently saves a .npz file that can be used to compute FVD and other
+evaluation metrics via the ADM repo: https://github.com/openai/guided-diffusion/tree/main/evaluations
+
+For a simple single-GPU/CPU sampling script, see sample.py.
+"""
+import io
+import os
+import sys
+import torch
+sys.path.append(os.path.split(sys.path[0])[0])
+import torch.distributed as dist
+from download import find_model
+from diffusion import create_diffusion
+from diffusers.models import AutoencoderKL
+from tqdm import tqdm
+import os
+from PIL import Image
+import numpy as np
+import math
+import argparse
+import imageio
+from omegaconf import OmegaConf
+from models import get_models
+from einops import rearrange
+
+
+def create_npz_from_sample_folder(sample_dir, num=50_000):
+    """
+    Builds a single .npz file from a folder of .png samples.
+    """
+    samples = []
+    for i in tqdm(range(num), desc="Building .npz file from samples"):
+        sample_pil = Image.open(f"{sample_dir}/{i:06d}.png")
+        sample_np = np.asarray(sample_pil).astype(np.uint8)
+        samples.append(sample_np)
+    samples = np.stack(samples)
+    assert samples.shape == (num, samples.shape[1], samples.shape[2], 3)
+    npz_path = f"{sample_dir}.npz"
+    np.savez(npz_path, arr_0=samples)
+    print(f"Saved .npz file to {npz_path} [shape={samples.shape}].")
+    return npz_path
+
+
+def main(args):
+    """
+    Run sampling.
+    """
+    torch.backends.cuda.matmul.allow_tf32 = True  # True: fast but may lead to some small numerical differences
+    assert torch.cuda.is_available(), "Sampling with DDP requires at least one GPU. sample.py supports CPU-only usage"
+    torch.set_grad_enabled(False)
+
+    # Setup DDP:
+    dist.init_process_group("nccl")
+    rank = dist.get_rank()
+    device = rank % torch.cuda.device_count()
+    if args.seed:
+        seed = args.seed * dist.get_world_size() + rank
+        torch.manual_seed(seed)
+    torch.cuda.set_device(device)
+    # print(f"Starting rank={rank}, seed={seed}, world_size={dist.get_world_size()}.")
+
+    if args.ckpt is None:
+        assert args.model == "Latte-XL/2", "Only Latte-XL/2 models are available for auto-download."
+        assert args.image_size in [256, 512]
+        assert args.num_classes == 1000
+
+    # Load model:
+    latent_size = args.image_size // 8
+    args.latent_size = latent_size
+    model = get_models(args).to(device)
+
+    if args.use_compile:
+        model = torch.compile(model)
+
+    # a pre-trained model or load a custom Latte checkpoint from train.py:
+    ckpt_path = args.ckpt
+    state_dict = find_model(ckpt_path)
+    model.load_state_dict(state_dict)
+    model.eval()  # important!
+    diffusion = create_diffusion(str(args.num_sampling_steps))
+    # vae = AutoencoderKL.from_pretrained(f"stabilityai/sd-vae-ft-{args.vae}").to(device)
+    # vae = AutoencoderKL.from_pretrained(args.pretrained_model_path, subfolder="vae").to(device)
+    vae = AutoencoderKL.from_pretrained(args.pretrained_model_path, subfolder="sd-vae-ft-ema").to(device)
+    
+    if args.use_fp16:
+        print('WARNING: using half percision for inferencing!')
+        vae.to(dtype=torch.float16)
+        model.to(dtype=torch.float16)
+        # text_encoder.to(dtype=torch.float16)
+    
+    assert args.cfg_scale >= 1.0, "In almost all cases, cfg_scale be >= 1.0"
+    using_cfg = args.cfg_scale > 1.0
+
+    # Create folder to save samples:
+    # model_string_name = args.model.replace("/", "-")
+    # ckpt_string_name = os.path.basename(args.ckpt).replace(".pt", "") if args.ckpt else "pretrained"
+    # folder_name = f"{model_string_name}-{ckpt_string_name}-size-{args.image_size}-vae-{args.vae}-" \
+    #               f"cfg-{args.cfg_scale}-seed-{args.seed}"
+    # sample_folder_dir = f"{args.sample_dir}/{folder_name}"
+    sample_folder_dir = args.save_video_path
+    if args.seed:
+        sample_folder_dir = args.save_video_path + '-seed-' + str(args.seed)
+    if rank == 0:
+        os.makedirs(sample_folder_dir, exist_ok=True)
+        print(f"Saving .mp4 samples at {sample_folder_dir}")
+    dist.barrier()
+
+    # Figure out how many samples we need to generate on each GPU and how many iterations we need to run:
+    n = args.per_proc_batch_size
+    global_batch_size = n * dist.get_world_size()
+    # To make things evenly-divisible, we'll sample a bit more than we need and then discard the extra samples:
+    total_samples = int(math.ceil(args.num_fvd_samples / global_batch_size) * global_batch_size)
+    if rank == 0:
+        print(f"Total number of images that will be sampled: {total_samples}")
+    assert total_samples % dist.get_world_size() == 0, "total_samples must be divisible by world_size"
+    samples_needed_this_gpu = int(total_samples // dist.get_world_size())
+    assert samples_needed_this_gpu % n == 0, "samples_needed_this_gpu must be divisible by the per-GPU batch size"
+    iterations = int(samples_needed_this_gpu // n)
+    pbar = range(iterations)
+    pbar = tqdm(pbar) if rank == 0 else pbar
+    total = 0
+    for _ in pbar:
+        # Sample inputs:
+        if args.use_fp16:
+            z = torch.randn(n, args.num_frames, 4, latent_size, latent_size, dtype=torch.float16, device=device)
+        else:
+            z = torch.randn(n, args.num_frames, 4, latent_size, latent_size, device=device)
+        
+        # Setup classifier-free guidance:
+        if using_cfg:
+            z = torch.cat([z, z], 0)
+            y = torch.randint(0, args.num_classes, (n,), device=device)
+            y_null = torch.tensor([101] * n, device=device)
+            y = torch.cat([y, y_null], dim=0)
+            model_kwargs = dict(y=y, cfg_scale=args.cfg_scale, use_fp16=args.use_fp16)
+            sample_fn = model.forward_with_cfg
+        else:
+            model_kwargs = dict(y=None, use_fp16=args.use_fp16)
+            sample_fn = model.forward
+
+        # Sample images:
+        if args.sample_method == 'ddim':
+            samples = diffusion.ddim_sample_loop(
+                sample_fn, z.shape, z, clip_denoised=False, model_kwargs=model_kwargs, progress=False, device=device
+            )
+        elif args.sample_method == 'ddpm':
+            samples = diffusion.p_sample_loop(
+                sample_fn, z.shape, z, clip_denoised=False, model_kwargs=model_kwargs, progress=False, device=device
+            )
+
+
+        if using_cfg:
+            samples, _ = samples.chunk(2, dim=0)  # Remove null class samples
+
+        if args.use_fp16:
+            samples = samples.to(dtype=torch.float16)
+
+        b, f, c, h, w = samples.shape
+        samples = rearrange(samples, 'b f c h w -> (b f) c h w')
+        samples = vae.decode(samples / 0.18215).sample
+        samples = rearrange(samples, '(b f) c h w -> b f c h w', b=b)
+
+        # Save samples to disk as individual .png files
+        for i, sample in enumerate(samples):
+            sample = ((sample * 0.5 + 0.5) * 255).add_(0.5).clamp_(0, 255).to(dtype=torch.uint8).cpu().permute(0, 2, 3, 1).contiguous()
+            index = i * dist.get_world_size() + rank + total
+            # Image.fromarray(sample).save(f"{sample_folder_dir}/{index:04d}.png")
+            sample_save_path = f"{sample_folder_dir}/{index:04d}.mp4"
+            imageio.mimwrite(sample_save_path, sample, fps=8, quality=9)
+        total += global_batch_size
+
+    # Make sure all processes have finished saving their samples before attempting to convert to .npz
+    dist.barrier()
+    # if rank == 0:
+    #     create_npz_from_sample_folder(sample_folder_dir, args.num_fvd_samples)
+    #     print("Done.")
+    # dist.barrier()
+    dist.destroy_process_group()
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--config", type=str, default="./configs/tuneavideo.yaml")
+    parser.add_argument("--ckpt", type=str, default="")
+    parser.add_argument("--save_video_path", type=str, default="./sample_videos/")
+    parser.add_argument("--save_ceph", default=False, action='store_true')
+    args = parser.parse_args()
+    omega_conf = OmegaConf.load(args.config)
+    omega_conf.ckpt = args.ckpt
+    omega_conf.save_video_path = args.save_video_path
+    omega_conf.save_ceph = args.save_ceph
+    main(omega_conf)
\ No newline at end of file

sample/sample_t2v.py

+import os
+import torch
+import argparse
+import torchvision
+
+
+from diffusers.schedulers import (DDIMScheduler, DDPMScheduler, PNDMScheduler, 
+                                  EulerDiscreteScheduler, DPMSolverMultistepScheduler, 
+                                  HeunDiscreteScheduler, EulerAncestralDiscreteScheduler,
+                                  DEISMultistepScheduler, KDPM2AncestralDiscreteScheduler)
+from diffusers.schedulers.scheduling_dpmsolver_singlestep import DPMSolverSinglestepScheduler
+from diffusers.models import AutoencoderKL
+from omegaconf import OmegaConf
+from transformers import T5EncoderModel, T5Tokenizer
+
+import os, sys
+sys.path.append(os.path.split(sys.path[0])[0])
+from download import find_model
+from pipeline_videogen import VideoGenPipeline
+from models import get_models
+from utils import save_video_grid
+import imageio
+from copy import deepcopy
+
+def main(args):
+    # torch.manual_seed(args.seed)
+    torch.set_grad_enabled(False)
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    transformer_model = get_models(args).to(device, dtype=torch.float16)
+    state_dict = find_model(args.ckpt)
+    transformer_model.load_state_dict(state_dict)
+
+    vae = AutoencoderKL.from_pretrained(args.pretrained_model_path, subfolder="vae", torch_dtype=torch.float16).to(device)
+    tokenizer = T5Tokenizer.from_pretrained(args.pretrained_model_path, subfolder="tokenizer")
+    text_encoder = T5EncoderModel.from_pretrained(args.pretrained_model_path, subfolder="text_encoder", torch_dtype=torch.float16).to(device)
+
+    # set eval mode
+    transformer_model.eval()
+    vae.eval()
+    text_encoder.eval()
+
+    if args.sample_method == 'DDIM':
+        scheduler = DDIMScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'EulerDiscrete':
+        scheduler = EulerDiscreteScheduler.from_pretrained(args.pretrained_model_path, 
+                                                        subfolder="scheduler",
+                                                        beta_start=args.beta_start, 
+                                                        beta_end=args.beta_end, 
+                                                        beta_schedule=args.beta_schedule,
+                                                        variance_type=args.variance_type)
+    elif args.sample_method == 'DDPM':
+        scheduler = DDPMScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'DPMSolverMultistep':
+        scheduler = DPMSolverMultistepScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'DPMSolverSinglestep':
+        scheduler = DPMSolverSinglestepScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'PNDM':
+        scheduler = PNDMScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'HeunDiscrete':
+        scheduler = HeunDiscreteScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'EulerAncestralDiscrete':
+        scheduler = EulerAncestralDiscreteScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'DEISMultistep':
+        scheduler = DEISMultistepScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+    elif args.sample_method == 'KDPM2AncestralDiscrete':
+        scheduler = KDPM2AncestralDiscreteScheduler.from_pretrained(args.pretrained_model_path, 
+                                                  subfolder="scheduler",
+                                                  beta_start=args.beta_start, 
+                                                  beta_end=args.beta_end, 
+                                                  beta_schedule=args.beta_schedule,
+                                                  variance_type=args.variance_type)
+
+
+    videogen_pipeline = VideoGenPipeline(vae=vae, 
+                                 text_encoder=text_encoder, 
+                                 tokenizer=tokenizer, 
+                                 scheduler=scheduler, 
+                                 transformer=transformer_model).to(device)
+    # videogen_pipeline.enable_xformers_memory_efficient_attention()
+
+    if not os.path.exists(args.save_img_path):
+        os.makedirs(args.save_img_path)
+
+    video_grids = []
+    for prompt in args.text_prompt:
+        print('Processing the ({}) prompt'.format(prompt))
+        videos = videogen_pipeline(prompt, 
+                                video_length=args.video_length, 
+                                height=args.image_size[0], 
+                                width=args.image_size[1], 
+                                num_inference_steps=args.num_sampling_steps,
+                                guidance_scale=args.guidance_scale,
+                                enable_temporal_attentions=args.enable_temporal_attentions,
+                                num_images_per_prompt=1,
+                                mask_feature=True,
+                                ).video
+        print(videos.shape)
+        try:
+            imageio.mimwrite(args.save_img_path + prompt.replace(' ', '_') + '_%04d' % args.run_time + 'webv-imageio.mp4', videos[0], fps=8, quality=9) # highest quality is 10, lowest is 0
+        except:
+            print('Error when saving {}'.format(prompt))
+        video_grids.append(videos)
+    video_grids = torch.cat(video_grids, dim=0)
+    print(video_grids.shape)
+
+    video_grids = save_video_grid(video_grids)
+
+    # torchvision.io.write_video(args.save_img_path + '_%04d' % args.run_time + '-.mp4', video_grids, fps=6)
+    imageio.mimwrite(args.save_img_path + '_%04d' % args.run_time + '-.mp4', video_grids, fps=8, quality=5)
+    print('save path {}'.format(args.save_img_path))
+
+    # save_videos_grid(video, f"./{prompt}.gif")
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--config", type=str, default="./configs/wbv10m_train.yaml")
+    args = parser.parse_args()
+
+    main(OmegaConf.load(args.config))
+

sample/sky.sh

+#!/bin/bash
+export CUDA_VISIBLE_DEVICES=7
+
+python sample/sample.py \
+--config ./configs/sky/sky_sample.yaml \
+--ckpt ./share_ckpts/skytimelapse.pt \
+--save_video_path ./test
\ No newline at end of file

sample/sky_ddp.sh

+#!/bin/bash
+export CUDA_VISIBLE_DEVICES=6,7
+
+torchrun --nnodes=1 --nproc_per_node=2 sample/sample_ddp.py \
+--config ./configs/sky/sky_sample.yaml \
+--ckpt ./share_ckpts/skytimelapse.pt \
+--save_video_path ./test

sample/t2v.sh

+python sample/sample_t2v.py --config configs/t2v/t2v_sample.yaml
\ No newline at end of file

sample/taichi.sh

+#!/bin/bash
+export CUDA_VISIBLE_DEVICES=7
+
+python sample/sample.py \
+--config ./configs/ucf101/taichi_sample.yaml \
+--ckpt  ./share_ckpts/taichi-hd.pt \
+--save_video_path ./test
\ No newline at end of file

sample/taichi_ddp.sh

+#!/bin/bash
+export CUDA_VISIBLE_DEVICES=6,7
+
+torchrun --nnodes=1 --nproc_per_node=2 sample/sample_ddp.py \
+--config ./configs/taichi/taichi_sample.yaml \
+--ckpt ./share_ckpts/taichi-hd.pt \
+--save_video_path ./test \

sample/ucf101.sh

+#!/bin/bash
+export CUDA_VISIBLE_DEVICES=7
+python sample/sample.py \
+--config ./configs/ucf101/ucf101_sample.yaml \
+--ckpt  ./share_ckpts/ucf101.pt \
+--save_video_path ./test
\ No newline at end of file

sample/ucf101_ddp.sh

+#!/bin/bash
+export CUDA_VISIBLE_DEVICES=6,7
+torchrun --nnodes=1 --nproc_per_node=2 --master_port=29520 sample/sample_ddp.py \
+--config ./configs/ucf101/ucf101_sample.yaml \
+--ckpt  ./share_ckpts/ucf101.pt \
+--save_video_path ./test