feat: 初始提交

Ruyi-Models/ruyi/utils/diffusion_utils.py

+# Modified from OpenAI's diffusion repos
+#     GLIDE: https://github.com/openai/glide-text2im/blob/main/glide_text2im/gaussian_diffusion.py
+#     ADM:   https://github.com/openai/guided-diffusion/blob/main/guided_diffusion
+#     IDDPM: https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py
+
+import numpy as np
+import torch as th
+
+
+def normal_kl(mean1, logvar1, mean2, logvar2):
+    """
+    Compute the KL divergence between two gaussians.
+    Shapes are automatically broadcasted, so batches can be compared to
+    scalars, among other use cases.
+    """
+    tensor = next(
+        (
+            obj
+            for obj in (mean1, logvar1, mean2, logvar2)
+            if isinstance(obj, th.Tensor)
+        ),
+        None,
+    )
+    assert tensor is not None, "at least one argument must be a Tensor"
+
+    # Force variances to be Tensors. Broadcasting helps convert scalars to
+    # Tensors, but it does not work for th.exp().
+    logvar1, logvar2 = [
+        x if isinstance(x, th.Tensor) else th.tensor(x, device=tensor.device)
+        for x in (logvar1, logvar2)
+    ]
+
+    return 0.5 * (
+        -1.0
+        + logvar2
+        - logvar1
+        + th.exp(logvar1 - logvar2)
+        + ((mean1 - mean2) ** 2) * th.exp(-logvar2)
+    )
+
+
+def approx_standard_normal_cdf(x):
+    """
+    A fast approximation of the cumulative distribution function of the
+    standard normal.
+    """
+    return 0.5 * (1.0 + th.tanh(np.sqrt(2.0 / np.pi) * (x + 0.044715 * th.pow(x, 3))))
+
+
+def continuous_gaussian_log_likelihood(x, *, means, log_scales):
+    """
+    Compute the log-likelihood of a continuous Gaussian distribution.
+    :param x: the targets
+    :param means: the Gaussian mean Tensor.
+    :param log_scales: the Gaussian log stddev Tensor.
+    :return: a tensor like x of log probabilities (in nats).
+    """
+    centered_x = x - means
+    inv_stdv = th.exp(-log_scales)
+    normalized_x = centered_x * inv_stdv
+    return th.distributions.Normal(th.zeros_like(x), th.ones_like(x)).log_prob(
+        normalized_x
+    )
+
+
+def discretized_gaussian_log_likelihood(x, *, means, log_scales):
+    """
+    Compute the log-likelihood of a Gaussian distribution discretizing to a
+    given image.
+    :param x: the target images. It is assumed that this was uint8 values,
+              rescaled to the range [-1, 1].
+    :param means: the Gaussian mean Tensor.
+    :param log_scales: the Gaussian log stddev Tensor.
+    :return: a tensor like x of log probabilities (in nats).
+    """
+    assert x.shape == means.shape == log_scales.shape
+    centered_x = x - means
+    inv_stdv = th.exp(-log_scales)
+    plus_in = inv_stdv * (centered_x + 1.0 / 255.0)
+    cdf_plus = approx_standard_normal_cdf(plus_in)
+    min_in = inv_stdv * (centered_x - 1.0 / 255.0)
+    cdf_min = approx_standard_normal_cdf(min_in)
+    log_cdf_plus = th.log(cdf_plus.clamp(min=1e-12))
+    log_one_minus_cdf_min = th.log((1.0 - cdf_min).clamp(min=1e-12))
+    cdf_delta = cdf_plus - cdf_min
+    log_probs = th.where(
+        x < -0.999,
+        log_cdf_plus,
+        th.where(x > 0.999, log_one_minus_cdf_min, th.log(cdf_delta.clamp(min=1e-12))),
+    )
+    assert log_probs.shape == x.shape
+    return log_probs
\ No newline at end of file

Ruyi-Models/ruyi/utils/gaussian_diffusion.py

+# Modified from OpenAI's diffusion repos
+#     GLIDE: https://github.com/openai/glide-text2im/blob/main/glide_text2im/gaussian_diffusion.py
+#     ADM:   https://github.com/openai/guided-diffusion/blob/main/guided_diffusion
+#     IDDPM: https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py
+
+
+import enum
+import math
+
+import numpy as np
+import torch as th
+import torch.nn.functional as F
+from einops import rearrange
+
+from .diffusion_utils import discretized_gaussian_log_likelihood, normal_kl
+
+
+def mean_flat(tensor):
+    """
+    Take the mean over all non-batch dimensions.
+    """
+    return tensor.mean(dim=list(range(1, len(tensor.shape))))
+
+
+class ModelMeanType(enum.Enum):
+    """
+    Which type of output the model predicts.
+    """
+
+    PREVIOUS_X = enum.auto()  # the model predicts x_{t-1}
+    START_X = enum.auto()  # the model predicts x_0
+    EPSILON = enum.auto()  # the model predicts epsilon
+
+
+class ModelVarType(enum.Enum):
+    """
+    What is used as the model's output variance.
+    The LEARNED_RANGE option has been added to allow the model to predict
+    values between FIXED_SMALL and FIXED_LARGE, making its job easier.
+    """
+
+    LEARNED = enum.auto()
+    FIXED_SMALL = enum.auto()
+    FIXED_LARGE = enum.auto()
+    LEARNED_RANGE = enum.auto()
+
+
+class LossType(enum.Enum):
+    MSE = enum.auto()  # use raw MSE loss (and KL when learning variances)
+    RESCALED_MSE = (
+        enum.auto()
+    )  # use raw MSE loss (with RESCALED_KL when learning variances)
+    KL = enum.auto()  # use the variational lower-bound
+    RESCALED_KL = enum.auto()  # like KL, but rescale to estimate the full VLB
+
+    def is_vb(self):
+        return self in [LossType.KL, LossType.RESCALED_KL]
+
+
+def _warmup_beta(beta_start, beta_end, num_diffusion_timesteps, warmup_frac):
+    betas = beta_end * np.ones(num_diffusion_timesteps, dtype=np.float64)
+    warmup_time = int(num_diffusion_timesteps * warmup_frac)
+    betas[:warmup_time] = np.linspace(beta_start, beta_end, warmup_time, dtype=np.float64)
+    return betas
+
+
+def get_beta_schedule(beta_schedule, *, beta_start, beta_end, num_diffusion_timesteps):
+    """
+    This is the deprecated API for creating beta schedules.
+    See get_named_beta_schedule() for the new library of schedules.
+    """
+    if beta_schedule == "quad":
+        betas = (
+            np.linspace(
+                beta_start ** 0.5,
+                beta_end ** 0.5,
+                num_diffusion_timesteps,
+                dtype=np.float64,
+            )
+            ** 2
+        )
+    elif beta_schedule == "linear":
+        betas = np.linspace(beta_start, beta_end, num_diffusion_timesteps, dtype=np.float64)
+    elif beta_schedule == "warmup10":
+        betas = _warmup_beta(beta_start, beta_end, num_diffusion_timesteps, 0.1)
+    elif beta_schedule == "warmup50":
+        betas = _warmup_beta(beta_start, beta_end, num_diffusion_timesteps, 0.5)
+    elif beta_schedule == "const":
+        betas = beta_end * np.ones(num_diffusion_timesteps, dtype=np.float64)
+    elif beta_schedule == "jsd":  # 1/T, 1/(T-1), 1/(T-2), ..., 1
+        betas = 1.0 / np.linspace(
+            num_diffusion_timesteps, 1, num_diffusion_timesteps, dtype=np.float64
+        )
+    else:
+        raise NotImplementedError(beta_schedule)
+    assert betas.shape == (num_diffusion_timesteps,)
+    return betas
+
+
+def get_named_beta_schedule(schedule_name, num_diffusion_timesteps):
+    """
+    Get a pre-defined beta schedule for the given name.
+    The beta schedule library consists of beta schedules which remain similar
+    in the limit of num_diffusion_timesteps.
+    Beta schedules may be added, but should not be removed or changed once
+    they are committed to maintain backwards compatibility.
+    """
+    if schedule_name == "linear":
+        # Linear schedule from Ho et al, extended to work for any number of
+        # diffusion steps.
+        scale = 1000 / num_diffusion_timesteps
+        return get_beta_schedule(
+            "linear",
+            beta_start=scale * 0.0001,
+            beta_end=scale * 0.02,
+            num_diffusion_timesteps=num_diffusion_timesteps,
+        )
+    elif schedule_name == "squaredcos_cap_v2":
+        return betas_for_alpha_bar(
+            num_diffusion_timesteps,
+            lambda t: math.cos((t + 0.008) / 1.008 * math.pi / 2) ** 2,
+        )
+    else:
+        raise NotImplementedError(f"unknown beta schedule: {schedule_name}")
+
+
+def betas_for_alpha_bar(num_diffusion_timesteps, alpha_bar, max_beta=0.999):
+    """
+    Create a beta schedule that discretizes the given alpha_t_bar function,
+    which defines the cumulative product of (1-beta) over time from t = [0,1].
+    :param num_diffusion_timesteps: the number of betas to produce.
+    :param alpha_bar: a lambda that takes an argument t from 0 to 1 and
+                      produces the cumulative product of (1-beta) up to that
+                      part of the diffusion process.
+    :param max_beta: the maximum beta to use; use values lower than 1 to
+                     prevent singularities.
+    """
+    betas = []
+    for i in range(num_diffusion_timesteps):
+        t1 = i / num_diffusion_timesteps
+        t2 = (i + 1) / num_diffusion_timesteps
+        betas.append(min(1 - alpha_bar(t2) / alpha_bar(t1), max_beta))
+    return np.array(betas)
+
+
+class GaussianDiffusion:
+    """
+    Utilities for training and sampling diffusion models.
+    Original ported from this codebase:
+    https://github.com/hojonathanho/diffusion/blob/1e0dceb3b3495bbe19116a5e1b3596cd0706c543/diffusion_tf/diffusion_utils_2.py#L42
+    :param betas: a 1-D numpy array of betas for each diffusion timestep,
+                  starting at T and going to 1.
+    """
+
+    def __init__(
+        self,
+        *,
+        betas,
+        model_mean_type,
+        model_var_type,
+        loss_type,
+        snr=False,
+        return_startx=False,
+    ):
+
+        self.model_mean_type = model_mean_type
+        self.model_var_type = model_var_type
+        self.loss_type = loss_type
+        self.snr = snr
+        self.return_startx = return_startx
+
+        # Use float64 for accuracy.
+        betas = np.array(betas, dtype=np.float64)
+        self.betas = betas
+        assert len(betas.shape) == 1, "betas must be 1-D"
+        assert (betas > 0).all() and (betas <= 1).all()
+
+        self.num_timesteps = int(betas.shape[0])
+
+        alphas = 1.0 - betas
+        self.alphas_cumprod = np.cumprod(alphas, axis=0)
+        self.alphas_cumprod_prev = np.append(1.0, self.alphas_cumprod[:-1])
+        self.alphas_cumprod_next = np.append(self.alphas_cumprod[1:], 0.0)
+        assert self.alphas_cumprod_prev.shape == (self.num_timesteps,)
+
+        # calculations for diffusion q(x_t | x_{t-1}) and others
+        self.sqrt_alphas_cumprod = np.sqrt(self.alphas_cumprod)
+        self.sqrt_one_minus_alphas_cumprod = np.sqrt(1.0 - self.alphas_cumprod)
+        self.log_one_minus_alphas_cumprod = np.log(1.0 - self.alphas_cumprod)
+        self.sqrt_recip_alphas_cumprod = np.sqrt(1.0 / self.alphas_cumprod)
+        self.sqrt_recipm1_alphas_cumprod = np.sqrt(1.0 / self.alphas_cumprod - 1)
+
+        # calculations for posterior q(x_{t-1} | x_t, x_0)
+        self.posterior_variance = (
+            betas * (1.0 - self.alphas_cumprod_prev) / (1.0 - self.alphas_cumprod)
+        )
+        # below: log calculation clipped because the posterior variance is 0 at the beginning of the diffusion chain
+        self.posterior_log_variance_clipped = np.log(
+            np.append(self.posterior_variance[1], self.posterior_variance[1:])
+        ) if len(self.posterior_variance) > 1 else np.array([])
+
+        self.posterior_mean_coef1 = (
+            betas * np.sqrt(self.alphas_cumprod_prev) / (1.0 - self.alphas_cumprod)
+        )
+        self.posterior_mean_coef2 = (
+            (1.0 - self.alphas_cumprod_prev) * np.sqrt(alphas) / (1.0 - self.alphas_cumprod)
+        )
+
+    def q_mean_variance(self, x_start, t):
+        """
+        Get the distribution q(x_t | x_0).
+        :param x_start: the [N x C x ...] tensor of noiseless inputs.
+        :param t: the number of diffusion steps (minus 1). Here, 0 means one step.
+        :return: A tuple (mean, variance, log_variance), all of x_start's shape.
+        """
+        mean = _extract_into_tensor(self.sqrt_alphas_cumprod, t, x_start.shape) * x_start
+        variance = _extract_into_tensor(1.0 - self.alphas_cumprod, t, x_start.shape)
+        log_variance = _extract_into_tensor(self.log_one_minus_alphas_cumprod, t, x_start.shape)
+        return mean, variance, log_variance
+
+    def q_sample(self, x_start, t, noise=None):
+        """
+        Diffuse the data for a given number of diffusion steps.
+        In other words, sample from q(x_t | x_0).
+        :param x_start: the initial data batch.
+        :param t: the number of diffusion steps (minus 1). Here, 0 means one step.
+        :param noise: if specified, the split-out normal noise.
+        :return: A noisy version of x_start.
+        """
+        if noise is None:
+            noise = th.randn_like(x_start)
+        assert noise.shape == x_start.shape
+        return (
+            _extract_into_tensor(self.sqrt_alphas_cumprod, t, x_start.shape) * x_start
+            + _extract_into_tensor(self.sqrt_one_minus_alphas_cumprod, t, x_start.shape) * noise
+        )
+
+    def q_posterior_mean_variance(self, x_start, x_t, t):
+        """
+        Compute the mean and variance of the diffusion posterior:
+            q(x_{t-1} | x_t, x_0)
+        """
+        assert x_start.shape == x_t.shape
+        posterior_mean = (
+            _extract_into_tensor(self.posterior_mean_coef1, t, x_t.shape) * x_start
+            + _extract_into_tensor(self.posterior_mean_coef2, t, x_t.shape) * x_t
+        )
+        posterior_variance = _extract_into_tensor(self.posterior_variance, t, x_t.shape)
+        posterior_log_variance_clipped = _extract_into_tensor(
+            self.posterior_log_variance_clipped, t, x_t.shape
+        )
+        assert (
+            posterior_mean.shape[0]
+            == posterior_variance.shape[0]
+            == posterior_log_variance_clipped.shape[0]
+            == x_start.shape[0]
+        )
+        return posterior_mean, posterior_variance, posterior_log_variance_clipped
+
+    def p_mean_variance(self, model, x, t, clip_denoised=True, denoised_fn=None, model_kwargs=None):
+        """
+        Apply the model to get p(x_{t-1} | x_t), as well as a prediction of
+        the initial x, x_0.
+        :param model: the model, which takes a signal and a batch of timesteps
+                      as input.
+        :param x: the [N x C x ...] tensor at time t.
+        :param t: a 1-D Tensor of timesteps.
+        :param clip_denoised: if True, clip the denoised signal into [-1, 1].
+        :param denoised_fn: if not None, a function which applies to the
+            x_start prediction before it is used to sample. Applies before
+            clip_denoised.
+        :param model_kwargs: if not None, a dict of extra keyword arguments to
+            pass to the model. This can be used for conditioning.
+        :return: a dict with the following keys:
+                 - 'mean': the model mean output.
+                 - 'variance': the model variance output.
+                 - 'log_variance': the log of 'variance'.
+                 - 'pred_xstart': the prediction for x_0.
+        """
+        if model_kwargs is None:
+            model_kwargs = {}
+
+        B, C = x.shape[:2]
+        assert t.shape == (B,)
+        model_output = model(x, timestep=t, **model_kwargs)
+        if isinstance(model_output, tuple):
+            model_output, extra = model_output
+        else:
+            extra = None
+
+        if self.model_var_type in [ModelVarType.LEARNED, ModelVarType.LEARNED_RANGE]:
+            assert model_output.shape == (B, C * 2, *x.shape[2:])
+            model_output, model_var_values = th.split(model_output, C, dim=1)
+            min_log = _extract_into_tensor(self.posterior_log_variance_clipped, t, x.shape)
+            max_log = _extract_into_tensor(np.log(self.betas), t, x.shape)
+            # The model_var_values is [-1, 1] for [min_var, max_var].
+            frac = (model_var_values + 1) / 2
+            model_log_variance = frac * max_log + (1 - frac) * min_log
+            model_variance = th.exp(model_log_variance)
+        elif self.model_var_type in [ModelVarType.FIXED_LARGE, ModelVarType.FIXED_SMALL]:
+            model_variance, model_log_variance = {
+                # for fixedlarge, we set the initial (log-)variance like so
+                # to get a better decoder log likelihood.
+                ModelVarType.FIXED_LARGE: (
+                    np.append(self.posterior_variance[1], self.betas[1:]),
+                    np.log(np.append(self.posterior_variance[1], self.betas[1:])),
+                ),
+                ModelVarType.FIXED_SMALL: (
+                    self.posterior_variance,
+                    self.posterior_log_variance_clipped,
+                ),
+            }[self.model_var_type]
+            model_variance = _extract_into_tensor(model_variance, t, x.shape)
+            model_log_variance = _extract_into_tensor(model_log_variance, t, x.shape)
+        else:
+            model_variance = th.zeros_like(model_output)
+            model_log_variance = th.zeros_like(model_output)
+
+        def process_xstart(x):
+            if denoised_fn is not None:
+                x = denoised_fn(x)
+            return x.clamp(-1, 1) if clip_denoised else x
+
+        if self.model_mean_type == ModelMeanType.START_X:
+            pred_xstart = process_xstart(model_output)
+        else:
+            pred_xstart = process_xstart(
+                self._predict_xstart_from_eps(x_t=x, t=t, eps=model_output)
+            )
+        model_mean, _, _ = self.q_posterior_mean_variance(x_start=pred_xstart, x_t=x, t=t)
+
+        assert model_mean.shape == model_log_variance.shape == pred_xstart.shape == x.shape
+        return {
+            "mean": model_mean,
+            "variance": model_variance,
+            "log_variance": model_log_variance,
+            "pred_xstart": pred_xstart,
+            "extra": extra,
+        }
+
+    def _predict_xstart_from_eps(self, x_t, t, eps):
+        assert x_t.shape == eps.shape
+        return (
+            _extract_into_tensor(self.sqrt_recip_alphas_cumprod, t, x_t.shape) * x_t
+            - _extract_into_tensor(self.sqrt_recipm1_alphas_cumprod, t, x_t.shape) * eps
+        )
+
+    def _predict_eps_from_xstart(self, x_t, t, pred_xstart):
+        return (
+            _extract_into_tensor(self.sqrt_recip_alphas_cumprod, t, x_t.shape) * x_t - pred_xstart
+        ) / _extract_into_tensor(self.sqrt_recipm1_alphas_cumprod, t, x_t.shape)
+
+    def condition_mean(self, cond_fn, p_mean_var, x, t, model_kwargs=None):
+        """
+        Compute the mean for the previous step, given a function cond_fn that
+        computes the gradient of a conditional log probability with respect to
+        x. In particular, cond_fn computes grad(log(p(y|x))), and we want to
+        condition on y.
+        This uses the conditioning strategy from Sohl-Dickstein et al. (2015).
+        """
+        gradient = cond_fn(x, t, **model_kwargs)
+        return p_mean_var["mean"].float() + p_mean_var["variance"] * gradient.float()
+
+    def condition_score(self, cond_fn, p_mean_var, x, t, model_kwargs=None):
+        """
+        Compute what the p_mean_variance output would have been, should the
+        model's score function be conditioned by cond_fn.
+        See condition_mean() for details on cond_fn.
+        Unlike condition_mean(), this instead uses the conditioning strategy
+        from Song et al (2020).
+        """
+        alpha_bar = _extract_into_tensor(self.alphas_cumprod, t, x.shape)
+
+        eps = self._predict_eps_from_xstart(x, t, p_mean_var["pred_xstart"])
+        eps = eps - (1 - alpha_bar).sqrt() * cond_fn(x, t, **model_kwargs)
+
+        out = p_mean_var.copy()
+        out["pred_xstart"] = self._predict_xstart_from_eps(x, t, eps)
+        out["mean"], _, _ = self.q_posterior_mean_variance(x_start=out["pred_xstart"], x_t=x, t=t)
+        return out
+
+    def p_sample(
+        self,
+        model,
+        x,
+        t,
+        clip_denoised=True,
+        denoised_fn=None,
+        cond_fn=None,
+        model_kwargs=None,
+    ):
+        """
+        Sample x_{t-1} from the model at the given timestep.
+        :param model: the model to sample from.
+        :param x: the current tensor at x_{t-1}.
+        :param t: the value of t, starting at 0 for the first diffusion step.
+        :param clip_denoised: if True, clip the x_start prediction to [-1, 1].
+        :param denoised_fn: if not None, a function which applies to the
+            x_start prediction before it is used to sample.
+        :param cond_fn: if not None, this is a gradient function that acts
+                        similarly to the model.
+        :param model_kwargs: if not None, a dict of extra keyword arguments to
+            pass to the model. This can be used for conditioning.
+        :return: a dict containing the following keys:
+                 - 'sample': a random sample from the model.
+                 - 'pred_xstart': a prediction of x_0.
+        """
+        out = self.p_mean_variance(
+            model,
+            x,
+            t,
+            clip_denoised=clip_denoised,
+            denoised_fn=denoised_fn,
+            model_kwargs=model_kwargs,
+        )
+        noise = th.randn_like(x)
+        nonzero_mask = (
+            (t != 0).float().view(-1, *([1] * (len(x.shape) - 1)))
+        )  # no noise when t == 0
+        if cond_fn is not None:
+            out["mean"] = self.condition_mean(cond_fn, out, x, t, model_kwargs=model_kwargs)
+        sample = out["mean"] + nonzero_mask * th.exp(0.5 * out["log_variance"]) * noise
+        return {"sample": sample, "pred_xstart": out["pred_xstart"]}
+
+    def p_sample_loop(
+        self,
+        model,
+        shape,
+        noise=None,
+        clip_denoised=True,
+        denoised_fn=None,
+        cond_fn=None,
+        model_kwargs=None,
+        device=None,
+        progress=False,
+    ):
+        """
+        Generate samples from the model.
+        :param model: the model module.
+        :param shape: the shape of the samples, (N, C, H, W).
+        :param noise: if specified, the noise from the encoder to sample.
+                      Should be of the same shape as `shape`.
+        :param clip_denoised: if True, clip x_start predictions to [-1, 1].
+        :param denoised_fn: if not None, a function which applies to the
+            x_start prediction before it is used to sample.
+        :param cond_fn: if not None, this is a gradient function that acts
+                        similarly to the model.
+        :param model_kwargs: if not None, a dict of extra keyword arguments to
+            pass to the model. This can be used for conditioning.
+        :param device: if specified, the device to create the samples on.
+                       If not specified, use a model parameter's device.
+        :param progress: if True, show a tqdm progress bar.
+        :return: a non-differentiable batch of samples.
+        """
+        final = None
+        for sample in self.p_sample_loop_progressive(
+            model,
+            shape,
+            noise=noise,
+            clip_denoised=clip_denoised,
+            denoised_fn=denoised_fn,
+            cond_fn=cond_fn,
+            model_kwargs=model_kwargs,
+            device=device,
+            progress=progress,
+        ):
+            final = sample
+        return final["sample"]
+
+    def p_sample_loop_progressive(
+        self,
+        model,
+        shape,
+        noise=None,
+        clip_denoised=True,
+        denoised_fn=None,
+        cond_fn=None,
+        model_kwargs=None,
+        device=None,
+        progress=False,
+    ):
+        """
+        Generate samples from the model and yield intermediate samples from
+        each timestep of diffusion.
+        Arguments are the same as p_sample_loop().
+        Returns a generator over dicts, where each dict is the return value of
+        p_sample().
+        """
+        if device is None:
+            device = next(model.parameters()).device
+        assert isinstance(shape, (tuple, list))
+        img = noise if noise is not None else th.randn(*shape, device=device)
+        indices = list(range(self.num_timesteps))[::-1]
+
+        if progress:
+            # Lazy import so that we don't depend on tqdm.
+            from tqdm.auto import tqdm
+
+            indices = tqdm(indices)
+
+        for i in indices:
+            t = th.tensor([i] * shape[0], device=device)
+            with th.no_grad():
+                out = self.p_sample(
+                    model,
+                    img,
+                    t,
+                    clip_denoised=clip_denoised,
+                    denoised_fn=denoised_fn,
+                    cond_fn=cond_fn,
+                    model_kwargs=model_kwargs,
+                )
+                yield out
+                img = out["sample"]
+
+    def ddim_sample(
+        self,
+        model,
+        x,
+        t,
+        clip_denoised=True,
+        denoised_fn=None,
+        cond_fn=None,
+        model_kwargs=None,
+        eta=0.0,
+    ):
+        """
+        Sample x_{t-1} from the model using DDIM.
+        Same usage as p_sample().
+        """
+        out = self.p_mean_variance(
+            model,
+            x,
+            t,
+            clip_denoised=clip_denoised,
+            denoised_fn=denoised_fn,
+            model_kwargs=model_kwargs,
+        )
+        if cond_fn is not None:
+            out = self.condition_score(cond_fn, out, x, t, model_kwargs=model_kwargs)
+
+        # Usually our model outputs epsilon, but we re-derive it
+        # in case we used x_start or x_prev prediction.
+        eps = self._predict_eps_from_xstart(x, t, out["pred_xstart"])
+
+        alpha_bar = _extract_into_tensor(self.alphas_cumprod, t, x.shape)
+        alpha_bar_prev = _extract_into_tensor(self.alphas_cumprod_prev, t, x.shape)
+        sigma = (
+            eta
+            * th.sqrt((1 - alpha_bar_prev) / (1 - alpha_bar))
+            * th.sqrt(1 - alpha_bar / alpha_bar_prev)
+        )
+        # Equation 12.
+        noise = th.randn_like(x)
+        mean_pred = (
+            out["pred_xstart"] * th.sqrt(alpha_bar_prev)
+            + th.sqrt(1 - alpha_bar_prev - sigma ** 2) * eps
+        )
+        nonzero_mask = (
+            (t != 0).float().view(-1, *([1] * (len(x.shape) - 1)))
+        )  # no noise when t == 0
+        sample = mean_pred + nonzero_mask * sigma * noise
+        return {"sample": sample, "pred_xstart": out["pred_xstart"]}
+
+    def ddim_reverse_sample(
+        self,
+        model,
+        x,
+        t,
+        clip_denoised=True,
+        denoised_fn=None,
+        cond_fn=None,
+        model_kwargs=None,
+        eta=0.0,
+    ):
+        """
+        Sample x_{t+1} from the model using DDIM reverse ODE.
+        """
+        assert eta == 0.0, "Reverse ODE only for deterministic path"
+        out = self.p_mean_variance(
+            model,
+            x,
+            t,
+            clip_denoised=clip_denoised,
+            denoised_fn=denoised_fn,
+            model_kwargs=model_kwargs,
+        )
+        if cond_fn is not None:
+            out = self.condition_score(cond_fn, out, x, t, model_kwargs=model_kwargs)
+        # Usually our model outputs epsilon, but we re-derive it
+        # in case we used x_start or x_prev prediction.
+        eps = (
+            _extract_into_tensor(self.sqrt_recip_alphas_cumprod, t, x.shape) * x
+            - out["pred_xstart"]
+        ) / _extract_into_tensor(self.sqrt_recipm1_alphas_cumprod, t, x.shape)
+        alpha_bar_next = _extract_into_tensor(self.alphas_cumprod_next, t, x.shape)
+
+        # Equation 12. reversed
+        mean_pred = out["pred_xstart"] * th.sqrt(alpha_bar_next) + th.sqrt(1 - alpha_bar_next) * eps
+
+        return {"sample": mean_pred, "pred_xstart": out["pred_xstart"]}
+
+    def ddim_sample_loop(
+        self,
+        model,
+        shape,
+        noise=None,
+        clip_denoised=True,
+        denoised_fn=None,
+        cond_fn=None,
+        model_kwargs=None,
+        device=None,
+        progress=False,
+        eta=0.0,
+    ):
+        """
+        Generate samples from the model using DDIM.
+        Same usage as p_sample_loop().
+        """
+        final = None
+        for sample in self.ddim_sample_loop_progressive(
+            model,
+            shape,
+            noise=noise,
+            clip_denoised=clip_denoised,
+            denoised_fn=denoised_fn,
+            cond_fn=cond_fn,
+            model_kwargs=model_kwargs,
+            device=device,
+            progress=progress,
+            eta=eta,
+        ):
+            final = sample
+        return final["sample"]
+
+    def ddim_sample_loop_progressive(
+        self,
+        model,
+        shape,
+        noise=None,
+        clip_denoised=True,
+        denoised_fn=None,
+        cond_fn=None,
+        model_kwargs=None,
+        device=None,
+        progress=False,
+        eta=0.0,
+    ):
+        """
+        Use DDIM to sample from the model and yield intermediate samples from
+        each timestep of DDIM.
+        Same usage as p_sample_loop_progressive().
+        """
+        if device is None:
+            device = next(model.parameters()).device
+        assert isinstance(shape, (tuple, list))
+        img = noise if noise is not None else th.randn(*shape, device=device)
+        indices = list(range(self.num_timesteps))[::-1]
+
+        if progress:
+            # Lazy import so that we don't depend on tqdm.
+            from tqdm.auto import tqdm
+
+            indices = tqdm(indices)
+
+        for i in indices:
+            t = th.tensor([i] * shape[0], device=device)
+            with th.no_grad():
+                out = self.ddim_sample(
+                    model,
+                    img,
+                    t,
+                    clip_denoised=clip_denoised,
+                    denoised_fn=denoised_fn,
+                    cond_fn=cond_fn,
+                    model_kwargs=model_kwargs,
+                    eta=eta,
+                )
+                yield out
+                img = out["sample"]
+
+    def _vb_terms_bpd(
+            self, model, x_start, x_t, t, clip_denoised=True, model_kwargs=None
+    ):
+        """
+        Get a term for the variational lower-bound.
+        The resulting units are bits (rather than nats, as one might expect).
+        This allows for comparison to other papers.
+        :return: a dict with the following keys:
+                 - 'output': a shape [N] tensor of NLLs or KLs.
+                 - 'pred_xstart': the x_0 predictions.
+        """
+        true_mean, _, true_log_variance_clipped = self.q_posterior_mean_variance(
+            x_start=x_start, x_t=x_t, t=t
+        )
+        out = self.p_mean_variance(
+            model, x_t, t, clip_denoised=clip_denoised, model_kwargs=model_kwargs
+        )
+        kl = normal_kl(
+            true_mean, true_log_variance_clipped, out["mean"], out["log_variance"]
+        )
+        kl = mean_flat(kl) / np.log(2.0)
+
+        decoder_nll = -discretized_gaussian_log_likelihood(
+            x_start, means=out["mean"], log_scales=0.5 * out["log_variance"]
+        )
+        assert decoder_nll.shape == x_start.shape
+        decoder_nll = mean_flat(decoder_nll) / np.log(2.0)
+
+        # At the first timestep return the decoder NLL,
+        # otherwise return KL(q(x_{t-1}|x_t,x_0) || p(x_{t-1}|x_t))
+        output = th.where((t == 0), decoder_nll, kl)
+        return {"output": output, "pred_xstart": out["pred_xstart"]}
+
+    def training_losses(self, model, x_start, timestep, model_kwargs=None, noise=None, skip_noise=False):
+        """
+        Compute training losses for a single timestep.
+        :param model: the model to evaluate loss on.
+        :param x_start: the [N x C x ...] tensor of inputs.
+        :param t: a batch of timestep indices.
+        :param model_kwargs: if not None, a dict of extra keyword arguments to
+            pass to the model. This can be used for conditioning.
+        :param noise: if specified, the specific Gaussian noise to try to remove.
+        :return: a dict with the key "loss" containing a tensor of shape [N].
+                 Some mean or variance settings may also have other keys.
+        """
+        t = timestep
+        if model_kwargs is None:
+            model_kwargs = {}
+        if skip_noise:
+            x_t = x_start
+        else:
+            if noise is None:
+                noise = th.randn_like(x_start)
+            x_t = self.q_sample(x_start, t, noise=noise)
+
+        terms = {}
+
+        if self.loss_type == LossType.KL or self.loss_type == LossType.RESCALED_KL:
+            terms["loss"] = self._vb_terms_bpd(
+                model=model,
+                x_start=x_start,
+                x_t=x_t,
+                t=t,
+                clip_denoised=False,
+                model_kwargs=model_kwargs,
+            )["output"]
+            if self.loss_type == LossType.RESCALED_KL:
+                terms["loss"] *= self.num_timesteps
+        elif self.loss_type in [LossType.MSE, LossType.RESCALED_MSE]:
+            model_output = model(x_t, timestep=t, **model_kwargs)[0]
+
+            if isinstance(model_output, dict) and model_output.get('x', None) is not None:
+                output = model_output['x']
+            else:
+                output = model_output
+
+            if self.return_startx and self.model_mean_type == ModelMeanType.EPSILON:
+                return self._extracted_from_training_losses_diffusers(x_t, output, t)
+            # self.model_var_type = ModelVarType.LEARNED_RANGE:4
+            if self.model_var_type in [
+                ModelVarType.LEARNED,
+                ModelVarType.LEARNED_RANGE,
+            ]:
+                B, C = x_t.shape[:2]
+                assert output.shape == (B, C * 2, *x_t.shape[2:])
+                output, model_var_values = th.split(output, C, dim=1)
+                # Learn the variance using the variational bound, but don't let it affect our mean prediction.
+                frozen_out = th.cat([output.detach(), model_var_values], dim=1)
+                # vb variational bound
+                terms["vb"] = self._vb_terms_bpd(
+                    model=lambda *args, r=frozen_out, **kwargs: r,
+                    x_start=x_start,
+                    x_t=x_t,
+                    t=t,
+                    clip_denoised=False,
+                )["output"]
+                if self.loss_type == LossType.RESCALED_MSE:
+                    # Divide by 1000 for equivalence with initial implementation.
+                    # Without a factor of 1/1000, the VB term hurts the MSE term.
+                    terms["vb"] *= self.num_timesteps / 1000.0
+
+            target = {
+                ModelMeanType.PREVIOUS_X: self.q_posterior_mean_variance(
+                    x_start=x_start, x_t=x_t, t=t
+                )[0],
+                ModelMeanType.START_X: x_start,
+                ModelMeanType.EPSILON: noise,
+            }[self.model_mean_type]
+            assert output.shape == target.shape == x_start.shape
+            if self.snr:
+                if self.model_mean_type == ModelMeanType.START_X:
+                    pred_noise = self._predict_eps_from_xstart(x_t=x_t, t=t, pred_xstart=output)
+                    pred_startx = output
+                elif self.model_mean_type == ModelMeanType.EPSILON:
+                    pred_noise = output
+                    pred_startx = self._predict_xstart_from_eps(x_t=x_t, t=t, eps=output)
+                # terms["mse_eps"] = mean_flat((noise - pred_noise) ** 2)
+                # terms["mse_x0"] = mean_flat((x_start - pred_startx) ** 2)
+
+                t = t[:, None, None, None, None].expand(pred_startx.shape)  # [128, 4, 32, 32]
+                # best
+                target = th.where(t > 249, noise, x_start)
+                output = th.where(t > 249, pred_noise, pred_startx)
+            loss = (target - output) ** 2
+            if model_kwargs.get('mask_ratio', False) and model_kwargs['mask_ratio'] > 0:
+                assert 'mask' in model_output
+                loss = F.avg_pool2d(loss.mean(dim=1), model.model.module.patch_size).flatten(1)
+                mask = model_output['mask']
+                unmask = 1 - mask
+                terms['mse'] = mean_flat(loss * unmask) * unmask.shape[1]/unmask.sum(1)
+                if model_kwargs['mask_loss_coef'] > 0:
+                    terms['mae'] = model_kwargs['mask_loss_coef'] * mean_flat(loss * mask) * mask.shape[1]/mask.sum(1)
+            else:
+                terms["mse"] = mean_flat(loss)
+            terms["loss"] = terms["mse"] + terms["vb"] if "vb" in terms else terms["mse"]
+            if "mae" in terms:
+                terms["loss"] = terms["loss"] + terms["mae"]
+        else:
+            raise NotImplementedError(self.loss_type)
+
+        return terms
+
+    def training_losses_diffusers(self, model, x_start, timestep, model_kwargs=None, noise=None, skip_noise=False):
+        """
+        Compute training losses for a single timestep.
+        :param model: the model to evaluate loss on.
+        :param x_start: the [N x C x ...] tensor of inputs.
+        :param t: a batch of timestep indices.
+        :param model_kwargs: if not None, a dict of extra keyword arguments to
+            pass to the model. This can be used for conditioning.
+        :param noise: if specified, the specific Gaussian noise to try to remove.
+        :return: a dict with the key "loss" containing a tensor of shape [N].
+                 Some mean or variance settings may also have other keys.
+        """
+        t = timestep
+        if model_kwargs is None:
+            model_kwargs = {}
+        if skip_noise:
+            x_t = x_start
+        else:
+            if noise is None:
+                noise = th.randn_like(x_start)
+            x_t = self.q_sample(x_start, t, noise=noise)
+
+        terms = {}
+
+        if self.loss_type in [LossType.KL, LossType.RESCALED_KL]:
+            terms["loss"] = self._vb_terms_bpd(
+                model=model,
+                x_start=x_start,
+                x_t=x_t,
+                t=t,
+                clip_denoised=False,
+                model_kwargs=model_kwargs,
+            )["output"]
+            if self.loss_type == LossType.RESCALED_KL:
+                terms["loss"] *= self.num_timesteps
+        elif self.loss_type in [LossType.MSE, LossType.RESCALED_MSE]:
+            output = model(x_t, timestep=t, **model_kwargs, return_dict=False)[0]
+
+            if self.return_startx and self.model_mean_type == ModelMeanType.EPSILON:
+                return self._extracted_from_training_losses_diffusers(x_t, output, t)
+
+            if self.model_var_type in [
+                ModelVarType.LEARNED,
+                ModelVarType.LEARNED_RANGE,
+            ]:
+                B, C = x_t.shape[:2]
+                assert output.shape == (B, C * 2, *x_t.shape[2:])
+                output, model_var_values = th.split(output, C, dim=1)
+                # Learn the variance using the variational bound, but don't let it affect our mean prediction.
+                frozen_out = th.cat([output.detach(), model_var_values], dim=1)
+                terms["vb"] = self._vb_terms_bpd(
+                    model=lambda *args, r=frozen_out, **kwargs: r,
+                    x_start=x_start,
+                    x_t=x_t,
+                    t=t,
+                    clip_denoised=False,
+                )["output"]
+                if self.loss_type == LossType.RESCALED_MSE:
+                    # Divide by 1000 for equivalence with initial implementation.
+                    # Without a factor of 1/1000, the VB term hurts the MSE term.
+                    terms["vb"] *= self.num_timesteps / 1000.0
+
+            target = {
+                ModelMeanType.PREVIOUS_X: self.q_posterior_mean_variance(
+                    x_start=x_start, x_t=x_t, t=t
+                )[0],
+                ModelMeanType.START_X: x_start,
+                ModelMeanType.EPSILON: noise,
+            }[self.model_mean_type]
+            assert output.shape == target.shape == x_start.shape
+            if self.snr:
+                if self.model_mean_type == ModelMeanType.START_X:
+                    pred_noise = self._predict_eps_from_xstart(x_t=x_t, t=t, pred_xstart=output)
+                    pred_startx = output
+                elif self.model_mean_type == ModelMeanType.EPSILON:
+                    pred_noise = output
+                    pred_startx = self._predict_xstart_from_eps(x_t=x_t, t=t, eps=output)
+                # terms["mse_eps"] = mean_flat((noise - pred_noise) ** 2)
+                # terms["mse_x0"] = mean_flat((x_start - pred_startx) ** 2)
+
+                t = t[:, None, None, None].expand(pred_startx.shape)  # [128, 4, 32, 32]
+                # best
+                target = th.where(t > 249, noise, x_start)
+                output = th.where(t > 249, pred_noise, pred_startx)
+            loss = (target - output) ** 2
+            terms["mse"] = mean_flat(loss)
+            terms["loss"] = terms["mse"] + terms["vb"] if "vb" in terms else terms["mse"]
+            if "mae" in terms:
+                terms["loss"] = terms["loss"] + terms["mae"]
+        else:
+            raise NotImplementedError(self.loss_type)
+
+        return terms
+
+    def _extracted_from_training_losses_diffusers(self, x_t, output, t):
+        B, C = x_t.shape[:2]
+        assert output.shape == (B, C * 2, *x_t.shape[2:])
+        output = th.split(output, C, dim=1)[0]
+        return output, self._predict_xstart_from_eps(x_t=x_t, t=t, eps=output), x_t
+
+    def _prior_bpd(self, x_start):
+        """
+        Get the prior KL term for the variational lower-bound, measured in
+        bits-per-dim.
+        This term can't be optimized, as it only depends on the encoder.
+        :param x_start: the [N x C x ...] tensor of inputs.
+        :return: a batch of [N] KL values (in bits), one per batch element.
+        """
+        batch_size = x_start.shape[0]
+        t = th.tensor([self.num_timesteps - 1] * batch_size, device=x_start.device)
+        qt_mean, _, qt_log_variance = self.q_mean_variance(x_start, t)
+        kl_prior = normal_kl(
+            mean1=qt_mean, logvar1=qt_log_variance, mean2=0.0, logvar2=0.0
+        )
+        return mean_flat(kl_prior) / np.log(2.0)
+
+    def calc_bpd_loop(self, model, x_start, clip_denoised=True, model_kwargs=None):
+        """
+        Compute the entire variational lower-bound, measured in bits-per-dim,
+        as well as other related quantities.
+        :param model: the model to evaluate loss on.
+        :param x_start: the [N x C x ...] tensor of inputs.
+        :param clip_denoised: if True, clip denoised samples.
+        :param model_kwargs: if not None, a dict of extra keyword arguments to
+            pass to the model. This can be used for conditioning.
+        :return: a dict containing the following keys:
+                 - total_bpd: the total variational lower-bound, per batch element.
+                 - prior_bpd: the prior term in the lower-bound.
+                 - vb: an [N x T] tensor of terms in the lower-bound.
+                 - xstart_mse: an [N x T] tensor of x_0 MSEs for each timestep.
+                 - mse: an [N x T] tensor of epsilon MSEs for each timestep.
+        """
+        device = x_start.device
+        batch_size = x_start.shape[0]
+
+        vb = []
+        xstart_mse = []
+        mse = []
+        for t in list(range(self.num_timesteps))[::-1]:
+            t_batch = th.tensor([t] * batch_size, device=device)
+            noise = th.randn_like(x_start)
+            x_t = self.q_sample(x_start=x_start, t=t_batch, noise=noise)
+            # Calculate VLB term at the current timestep
+            with th.no_grad():
+                out = self._vb_terms_bpd(
+                    model,
+                    x_start=x_start,
+                    x_t=x_t,
+                    t=t_batch,
+                    clip_denoised=clip_denoised,
+                    model_kwargs=model_kwargs,
+                )
+            vb.append(out["output"])
+            xstart_mse.append(mean_flat((out["pred_xstart"] - x_start) ** 2))
+            eps = self._predict_eps_from_xstart(x_t, t_batch, out["pred_xstart"])
+            mse.append(mean_flat((eps - noise) ** 2))
+
+        vb = th.stack(vb, dim=1)
+        xstart_mse = th.stack(xstart_mse, dim=1)
+        mse = th.stack(mse, dim=1)
+
+        prior_bpd = self._prior_bpd(x_start)
+        total_bpd = vb.sum(dim=1) + prior_bpd
+        return {
+            "total_bpd": total_bpd,
+            "prior_bpd": prior_bpd,
+            "vb": vb,
+            "xstart_mse": xstart_mse,
+            "mse": mse,
+        }
+
+
+def _extract_into_tensor(arr, timesteps, broadcast_shape):
+    """
+    Extract values from a 1-D numpy array for a batch of indices.
+    :param arr: the 1-D numpy array.
+    :param timesteps: a tensor of indices into the array to extract.
+    :param broadcast_shape: a larger shape of K dimensions with the batch
+                            dimension equal to the length of timesteps.
+    :return: a tensor of shape [batch_size, 1, ...] where the shape has K dims.
+    """
+    res = th.from_numpy(arr).to(device=timesteps.device)[timesteps].float()
+    while len(res.shape) < len(broadcast_shape):
+        res = res[..., None]
+    return res + th.zeros(broadcast_shape, device=timesteps.device)
\ No newline at end of file

Ruyi-Models/ruyi/utils/lora_utils.py

+# LoRA network module
+# reference:
+# https://github.com/microsoft/LoRA/blob/main/loralib/layers.py
+# https://github.com/cloneofsimo/lora/blob/master/lora_diffusion/lora.py
+# https://github.com/bmaltais/kohya_ss
+
+import hashlib
+import math
+import os
+from collections import defaultdict
+from io import BytesIO
+from typing import List, Optional, Type, Union
+
+import safetensors.torch
+import torch
+import torch.utils.checkpoint
+from diffusers.models.lora import LoRACompatibleConv, LoRACompatibleLinear
+from safetensors.torch import load_file
+from transformers import T5EncoderModel
+
+
+class LoRAModule(torch.nn.Module):
+    """
+    replaces forward method of the original Linear, instead of replacing the original Linear module.
+    """
+
+    def __init__(
+        self,
+        lora_name,
+        org_module: torch.nn.Module,
+        multiplier=1.0,
+        lora_dim=4,
+        alpha=1,
+        dropout=None,
+        rank_dropout=None,
+        module_dropout=None,
+    ):
+        """if alpha == 0 or None, alpha is rank (no scaling)."""
+        super().__init__()
+        self.lora_name = lora_name
+
+        if org_module.__class__.__name__ == "Conv2d":
+            in_dim = org_module.in_channels
+            out_dim = org_module.out_channels
+        else:
+            in_dim = org_module.in_features
+            out_dim = org_module.out_features
+
+        self.lora_dim = lora_dim
+        if org_module.__class__.__name__ == "Conv2d":
+            kernel_size = org_module.kernel_size
+            stride = org_module.stride
+            padding = org_module.padding
+            self.lora_down = torch.nn.Conv2d(in_dim, self.lora_dim, kernel_size, stride, padding, bias=False)
+            self.lora_up = torch.nn.Conv2d(self.lora_dim, out_dim, (1, 1), (1, 1), bias=False)
+        else:
+            self.lora_down = torch.nn.Linear(in_dim, self.lora_dim, bias=False)
+            self.lora_up = torch.nn.Linear(self.lora_dim, out_dim, bias=False)
+
+        if type(alpha) == torch.Tensor:
+            alpha = alpha.detach().float().numpy()  # without casting, bf16 causes error
+        alpha = self.lora_dim if alpha is None or alpha == 0 else alpha
+        self.scale = alpha / self.lora_dim
+        self.register_buffer("alpha", torch.tensor(alpha))
+
+        # same as microsoft's
+        torch.nn.init.kaiming_uniform_(self.lora_down.weight, a=math.sqrt(5))
+        torch.nn.init.zeros_(self.lora_up.weight)
+
+        self.multiplier = multiplier
+        self.org_module = org_module  # remove in applying
+        self.dropout = dropout
+        self.rank_dropout = rank_dropout
+        self.module_dropout = module_dropout
+
+    def apply_to(self):
+        self.org_forward = self.org_module.forward
+        self.org_module.forward = self.forward
+        del self.org_module
+
+    def forward(self, x, *args, **kwargs):
+        weight_dtype = x.dtype
+        org_forwarded = self.org_forward(x)
+
+        # module dropout
+        if self.module_dropout is not None and self.training:
+            if torch.rand(1) < self.module_dropout:
+                return org_forwarded
+
+        lx = self.lora_down(x.to(self.lora_down.weight.dtype))
+
+        # normal dropout
+        if self.dropout is not None and self.training:
+            lx = torch.nn.functional.dropout(lx, p=self.dropout)
+
+        # rank dropout
+        if self.rank_dropout is not None and self.training:
+            mask = torch.rand((lx.size(0), self.lora_dim), device=lx.device) > self.rank_dropout
+            if len(lx.size()) == 3:
+                mask = mask.unsqueeze(1)  # for Text Encoder
+            elif len(lx.size()) == 4:
+                mask = mask.unsqueeze(-1).unsqueeze(-1)  # for Conv2d
+            lx = lx * mask
+
+            # scaling for rank dropout: treat as if the rank is changed
+            scale = self.scale * (1.0 / (1.0 - self.rank_dropout))  # redundant for readability
+        else:
+            scale = self.scale
+
+        lx = self.lora_up(lx)
+
+        return org_forwarded.to(weight_dtype) + lx.to(weight_dtype) * self.multiplier * scale
+
+
+def addnet_hash_legacy(b):
+    """Old model hash used by sd-webui-additional-networks for .safetensors format files"""
+    m = hashlib.sha256()
+
+    b.seek(0x100000)
+    m.update(b.read(0x10000))
+    return m.hexdigest()[0:8]
+
+
+def addnet_hash_safetensors(b):
+    """New model hash used by sd-webui-additional-networks for .safetensors format files"""
+    hash_sha256 = hashlib.sha256()
+    blksize = 1024 * 1024
+
+    b.seek(0)
+    header = b.read(8)
+    n = int.from_bytes(header, "little")
+
+    offset = n + 8
+    b.seek(offset)
+    for chunk in iter(lambda: b.read(blksize), b""):
+        hash_sha256.update(chunk)
+
+    return hash_sha256.hexdigest()
+
+
+def precalculate_safetensors_hashes(tensors, metadata):
+    """Precalculate the model hashes needed by sd-webui-additional-networks to
+    save time on indexing the model later."""
+
+    # Because writing user metadata to the file can change the result of
+    # sd_models.model_hash(), only retain the training metadata for purposes of
+    # calculating the hash, as they are meant to be immutable
+    metadata = {k: v for k, v in metadata.items() if k.startswith("ss_")}
+
+    bytes = safetensors.torch.save(tensors, metadata)
+    b = BytesIO(bytes)
+
+    model_hash = addnet_hash_safetensors(b)
+    legacy_hash = addnet_hash_legacy(b)
+    return model_hash, legacy_hash
+
+
+class LoRANetwork(torch.nn.Module):
+    TRANSFORMER_TARGET_REPLACE_MODULE = ["Transformer2DModel", "Transformer3DModel", "HunyuanTransformer3DModel"]
+    TEXT_ENCODER_TARGET_REPLACE_MODULE = ["T5LayerSelfAttention", "T5LayerFF", "BertEncoder"]
+    LORA_PREFIX_TRANSFORMER = "lora_unet"
+    LORA_PREFIX_TEXT_ENCODER = "lora_te"
+    def __init__(
+        self,
+        text_encoder: Union[List[T5EncoderModel], T5EncoderModel],
+        unet,
+        multiplier: float = 1.0,
+        lora_dim: int = 4,
+        alpha: float = 1,
+        dropout: Optional[float] = None,
+        module_class: Type[object] = LoRAModule,
+        add_lora_in_attn_temporal: bool = False,
+        varbose: Optional[bool] = False,
+    ) -> None:
+        super().__init__()
+        self.multiplier = multiplier
+
+        self.lora_dim = lora_dim
+        self.alpha = alpha
+        self.dropout = dropout
+
+        print(f"create LoRA network. base dim (rank): {lora_dim}, alpha: {alpha}")
+        print(f"neuron dropout: p={self.dropout}")
+
+        # create module instances
+        def create_modules(
+            is_unet: bool,
+            root_module: torch.nn.Module,
+            target_replace_modules: List[torch.nn.Module],
+        ) -> List[LoRAModule]:
+            prefix = (
+                self.LORA_PREFIX_TRANSFORMER
+                if is_unet
+                else self.LORA_PREFIX_TEXT_ENCODER
+            )
+            loras = []
+            skipped = []
+            for name, module in root_module.named_modules():
+                if module.__class__.__name__ in target_replace_modules:
+                    for child_name, child_module in module.named_modules():
+                        is_linear = child_module.__class__.__name__ == "Linear" or child_module.__class__.__name__ == "LoRACompatibleLinear"
+                        is_conv2d = child_module.__class__.__name__ == "Conv2d" or child_module.__class__.__name__ == "LoRACompatibleConv"
+                        is_conv2d_1x1 = is_conv2d and child_module.kernel_size == (1, 1)
+                        
+                        if not add_lora_in_attn_temporal:
+                            if "attn_temporal" in child_name:
+                                continue
+
+                        if is_linear or is_conv2d:
+                            lora_name = prefix + "." + name + "." + child_name
+                            lora_name = lora_name.replace(".", "_")
+
+                            dim = None
+                            alpha = None
+
+                            if is_linear or is_conv2d_1x1:
+                                dim = self.lora_dim
+                                alpha = self.alpha
+
+                            if dim is None or dim == 0:
+                                if is_linear or is_conv2d_1x1:
+                                    skipped.append(lora_name)
+                                continue
+
+                            lora = module_class(
+                                lora_name,
+                                child_module,
+                                self.multiplier,
+                                dim,
+                                alpha,
+                                dropout=dropout,
+                            )
+                            loras.append(lora)
+            return loras, skipped
+
+        text_encoders = text_encoder if type(text_encoder) == list else [text_encoder]
+
+        self.text_encoder_loras = []
+        skipped_te = []
+        for i, text_encoder in enumerate(text_encoders):
+            if text_encoder is not None:
+                text_encoder_loras, skipped = create_modules(False, text_encoder, LoRANetwork.TEXT_ENCODER_TARGET_REPLACE_MODULE)
+                self.text_encoder_loras.extend(text_encoder_loras)
+                skipped_te += skipped
+        print(f"create LoRA for Text Encoder: {len(self.text_encoder_loras)} modules.")
+
+        self.unet_loras, skipped_un = create_modules(True, unet, LoRANetwork.TRANSFORMER_TARGET_REPLACE_MODULE)
+        print(f"create LoRA for U-Net: {len(self.unet_loras)} modules.")
+
+        # assertion
+        names = set()
+        for lora in self.text_encoder_loras + self.unet_loras:
+            assert lora.lora_name not in names, f"duplicated lora name: {lora.lora_name}"
+            names.add(lora.lora_name)
+
+    def apply_to(self, text_encoder, unet, apply_text_encoder=True, apply_unet=True):
+        if apply_text_encoder:
+            print("enable LoRA for text encoder")
+        else:
+            self.text_encoder_loras = []
+
+        if apply_unet:
+            print("enable LoRA for U-Net")
+        else:
+            self.unet_loras = []
+
+        for lora in self.text_encoder_loras + self.unet_loras:
+            lora.apply_to()
+            self.add_module(lora.lora_name, lora)
+
+    def set_multiplier(self, multiplier):
+        self.multiplier = multiplier
+        for lora in self.text_encoder_loras + self.unet_loras:
+            lora.multiplier = self.multiplier
+
+    def load_weights(self, file):
+        if os.path.splitext(file)[1] == ".safetensors":
+            from safetensors.torch import load_file
+
+            weights_sd = load_file(file)
+        else:
+            weights_sd = torch.load(file, map_location="cpu")
+        info = self.load_state_dict(weights_sd, False)
+        return info
+
+    def prepare_optimizer_params(self, text_encoder_lr, unet_lr, default_lr):
+        self.requires_grad_(True)
+        all_params = []
+
+        def enumerate_params(loras):
+            params = []
+            for lora in loras:
+                params.extend(lora.parameters())
+            return params
+
+        if self.text_encoder_loras:
+            param_data = {"params": enumerate_params(self.text_encoder_loras)}
+            if text_encoder_lr is not None:
+                param_data["lr"] = text_encoder_lr
+            all_params.append(param_data)
+
+        if self.unet_loras:
+            param_data = {"params": enumerate_params(self.unet_loras)}
+            if unet_lr is not None:
+                param_data["lr"] = unet_lr
+            all_params.append(param_data)
+
+        return all_params
+
+    def enable_gradient_checkpointing(self):
+        pass
+
+    def get_trainable_params(self):
+        return self.parameters()
+
+    def save_weights(self, file, dtype, metadata):
+        if metadata is not None and len(metadata) == 0:
+            metadata = None
+
+        state_dict = self.state_dict()
+
+        if dtype is not None:
+            for key in list(state_dict.keys()):
+                v = state_dict[key]
+                v = v.detach().clone().to("cpu").to(dtype)
+                state_dict[key] = v
+
+        if os.path.splitext(file)[1] == ".safetensors":
+            from safetensors.torch import save_file
+
+            # Precalculate model hashes to save time on indexing
+            if metadata is None:
+                metadata = {}
+            model_hash, legacy_hash = precalculate_safetensors_hashes(state_dict, metadata)
+            metadata["sshs_model_hash"] = model_hash
+            metadata["sshs_legacy_hash"] = legacy_hash
+
+            save_file(state_dict, file, metadata)
+        else:
+            torch.save(state_dict, file)
+
+def create_network(
+    multiplier: float,
+    network_dim: Optional[int],
+    network_alpha: Optional[float],
+    text_encoder: Union[T5EncoderModel, List[T5EncoderModel]],
+    transformer,
+    neuron_dropout: Optional[float] = None,
+    add_lora_in_attn_temporal: bool = False,
+    **kwargs,
+):
+    if network_dim is None:
+        network_dim = 4  # default
+    if network_alpha is None:
+        network_alpha = 1.0
+
+    network = LoRANetwork(
+        text_encoder,
+        transformer,
+        multiplier=multiplier,
+        lora_dim=network_dim,
+        alpha=network_alpha,
+        dropout=neuron_dropout,
+        add_lora_in_attn_temporal=add_lora_in_attn_temporal,
+        varbose=True,
+    )
+    return network
+
+def merge_lora(pipeline, lora_path, multiplier, device='cpu', dtype=torch.float32, state_dict=None, transformer_only=False):
+    LORA_PREFIX_TRANSFORMER = "lora_unet"
+    LORA_PREFIX_TEXT_ENCODER = "lora_te"
+    if state_dict is None:
+        state_dict = load_file(lora_path, device=device)
+    else:
+        state_dict = state_dict
+    updates = defaultdict(dict)
+    for key, value in state_dict.items():
+        layer, elem = key.split('.', 1)
+        updates[layer][elem] = value
+
+    for layer, elems in updates.items():
+
+        if "lora_te" in layer:
+            if transformer_only:
+                continue
+            else:
+                layer_infos = layer.split(LORA_PREFIX_TEXT_ENCODER + "_")[-1].split("_")
+                curr_layer = pipeline.text_encoder
+        else:
+            layer_infos = layer.split(LORA_PREFIX_TRANSFORMER + "_")[-1].split("_")
+            curr_layer = pipeline.transformer
+
+        temp_name = layer_infos.pop(0)
+        while len(layer_infos) > -1:
+            try:
+                curr_layer = curr_layer.__getattr__(temp_name)
+                if len(layer_infos) > 0:
+                    temp_name = layer_infos.pop(0)
+                elif len(layer_infos) == 0:
+                    break
+            except Exception:
+                if len(layer_infos) == 0:
+                    print('Error loading layer')
+                if len(temp_name) > 0:
+                    temp_name += "_" + layer_infos.pop(0)
+                else:
+                    temp_name = layer_infos.pop(0)
+
+        weight_up = elems['lora_up.weight'].to(dtype)
+        weight_down = elems['lora_down.weight'].to(dtype)
+        if 'alpha' in elems.keys():
+            alpha = elems['alpha'].item() / weight_up.shape[1]
+        else:
+            alpha = 1.0
+
+        curr_layer.weight.data = curr_layer.weight.data.to(device)
+        if len(weight_up.shape) == 4:
+            curr_layer.weight.data += multiplier * alpha * torch.mm(weight_up.squeeze(3).squeeze(2),
+                                                                    weight_down.squeeze(3).squeeze(2)).unsqueeze(
+                2).unsqueeze(3)
+        else:
+            curr_layer.weight.data += multiplier * alpha * torch.mm(weight_up, weight_down)
+
+    return pipeline
+
+# TODO: Refactor with merge_lora.
+def unmerge_lora(pipeline, lora_path, multiplier=1, device="cpu", dtype=torch.float32):
+    """Unmerge state_dict in LoRANetwork from the pipeline in diffusers."""
+    LORA_PREFIX_UNET = "lora_unet"
+    LORA_PREFIX_TEXT_ENCODER = "lora_te"
+    state_dict = load_file(lora_path, device=device)
+
+    updates = defaultdict(dict)
+    for key, value in state_dict.items():
+        layer, elem = key.split('.', 1)
+        updates[layer][elem] = value
+
+    for layer, elems in updates.items():
+
+        if "lora_te" in layer:
+            layer_infos = layer.split(LORA_PREFIX_TEXT_ENCODER + "_")[-1].split("_")
+            curr_layer = pipeline.text_encoder
+        else:
+            layer_infos = layer.split(LORA_PREFIX_UNET + "_")[-1].split("_")
+            curr_layer = pipeline.transformer
+
+        temp_name = layer_infos.pop(0)
+        while len(layer_infos) > -1:
+            try:
+                curr_layer = curr_layer.__getattr__(temp_name)
+                if len(layer_infos) > 0:
+                    temp_name = layer_infos.pop(0)
+                elif len(layer_infos) == 0:
+                    break
+            except Exception:
+                if len(layer_infos) == 0:
+                    print('Error loading layer')
+                if len(temp_name) > 0:
+                    temp_name += "_" + layer_infos.pop(0)
+                else:
+                    temp_name = layer_infos.pop(0)
+
+        weight_up = elems['lora_up.weight'].to(dtype)
+        weight_down = elems['lora_down.weight'].to(dtype)
+        if 'alpha' in elems.keys():
+            alpha = elems['alpha'].item() / weight_up.shape[1]
+        else:
+            alpha = 1.0
+
+        curr_layer.weight.data = curr_layer.weight.data.to(device)
+        if len(weight_up.shape) == 4:
+            curr_layer.weight.data -= multiplier * alpha * torch.mm(weight_up.squeeze(3).squeeze(2),
+                                                                    weight_down.squeeze(3).squeeze(2)).unsqueeze(2).unsqueeze(3)
+        else:
+            curr_layer.weight.data -= multiplier * alpha * torch.mm(weight_up, weight_down)
+
+    return pipeline
\ No newline at end of file

Ruyi-Models/ruyi/utils/respace.py

+# Modified from OpenAI's diffusion repos
+#     GLIDE: https://github.com/openai/glide-text2im/blob/main/glide_text2im/gaussian_diffusion.py
+#     ADM:   https://github.com/openai/guided-diffusion/blob/main/guided_diffusion
+#     IDDPM: https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py
+
+import numpy as np
+import torch as th
+
+from .gaussian_diffusion import GaussianDiffusion
+
+
+def space_timesteps(num_timesteps, section_counts):
+    """
+    Create a list of timesteps to use from an original diffusion process,
+    given the number of timesteps we want to take from equally-sized portions
+    of the original process.
+    For example, if there's 300 timesteps and the section counts are [10,15,20]
+    then the first 100 timesteps are strided to be 10 timesteps, the second 100
+    are strided to be 15 timesteps, and the final 100 are strided to be 20.
+    If the stride is a string starting with "ddim", then the fixed striding
+    from the DDIM paper is used, and only one section is allowed.
+    :param num_timesteps: the number of diffusion steps in the original
+                          process to divide up.
+    :param section_counts: either a list of numbers, or a string containing
+                           comma-separated numbers, indicating the step count
+                           per section. As a special case, use "ddimN" where N
+                           is a number of steps to use the striding from the
+                           DDIM paper.
+    :return: a set of diffusion steps from the original process to use.
+    """
+    if isinstance(section_counts, str):
+        if section_counts.startswith("ddim"):
+            desired_count = int(section_counts[len("ddim") :])
+            for i in range(1, num_timesteps):
+                if len(range(0, num_timesteps, i)) == desired_count:
+                    return set(range(0, num_timesteps, i))
+            raise ValueError(
+                f"cannot create exactly {num_timesteps} steps with an integer stride"
+            )
+        section_counts = [int(x) for x in section_counts.split(",")]
+    size_per = num_timesteps // len(section_counts)
+    extra = num_timesteps % len(section_counts)
+    start_idx = 0
+    all_steps = []
+    for i, section_count in enumerate(section_counts):
+        size = size_per + (1 if i < extra else 0)
+        if size < section_count:
+            raise ValueError(
+                f"cannot divide section of {size} steps into {section_count}"
+            )
+        frac_stride = 1 if section_count <= 1 else (size - 1) / (section_count - 1)
+        cur_idx = 0.0
+        taken_steps = []
+        for _ in range(section_count):
+            taken_steps.append(start_idx + round(cur_idx))
+            cur_idx += frac_stride
+        all_steps += taken_steps
+        start_idx += size
+    return set(all_steps)
+
+
+class SpacedDiffusion(GaussianDiffusion):
+    """
+    A diffusion process which can skip steps in a base diffusion process.
+    :param use_timesteps: a collection (sequence or set) of timesteps from the
+                          original diffusion process to retain.
+    :param kwargs: the kwargs to create the base diffusion process.
+    """
+
+    def __init__(self, use_timesteps, **kwargs):
+        self.use_timesteps = set(use_timesteps)
+        self.timestep_map = []
+        self.original_num_steps = len(kwargs["betas"])
+
+        base_diffusion = GaussianDiffusion(**kwargs)  # pylint: disable=missing-kwoa
+        last_alpha_cumprod = 1.0
+        new_betas = []
+        for i, alpha_cumprod in enumerate(base_diffusion.alphas_cumprod):
+            if i in self.use_timesteps:
+                new_betas.append(1 - alpha_cumprod / last_alpha_cumprod)
+                last_alpha_cumprod = alpha_cumprod
+                self.timestep_map.append(i)
+        kwargs["betas"] = np.array(new_betas)
+        super().__init__(**kwargs)
+
+    def p_mean_variance(
+        self, model, *args, **kwargs
+    ):  # pylint: disable=signature-differs
+        return super().p_mean_variance(self._wrap_model(model), *args, **kwargs)
+
+    def training_losses(
+        self, model, *args, **kwargs
+    ):  # pylint: disable=signature-differs
+        return super().training_losses(self._wrap_model(model), *args, **kwargs)
+
+    def training_losses_diffusers(
+        self, model, *args, **kwargs
+    ):  # pylint: disable=signature-differs
+        return super().training_losses_diffusers(self._wrap_model(model), *args, **kwargs)
+
+    def condition_mean(self, cond_fn, *args, **kwargs):
+        return super().condition_mean(self._wrap_model(cond_fn), *args, **kwargs)
+
+    def condition_score(self, cond_fn, *args, **kwargs):
+        return super().condition_score(self._wrap_model(cond_fn), *args, **kwargs)
+
+    def _wrap_model(self, model):
+        if isinstance(model, _WrappedModel):
+            return model
+        return _WrappedModel(
+            model, self.timestep_map, self.original_num_steps
+        )
+
+    def _scale_timesteps(self, t):
+        # Scaling is done by the wrapped model.
+        return t
+
+
+class _WrappedModel:
+    def __init__(self, model, timestep_map, original_num_steps):
+        self.model = model
+        self.timestep_map = timestep_map
+        # self.rescale_timesteps = rescale_timesteps
+        self.original_num_steps = original_num_steps
+
+    def __call__(self, x, timestep, **kwargs):
+        map_tensor = th.tensor(self.timestep_map, device=timestep.device, dtype=timestep.dtype)
+        new_ts = map_tensor[timestep]
+        # if self.rescale_timesteps:
+        #     new_ts = new_ts.float() * (1000.0 / self.original_num_steps)
+        return self.model(x, timestep=new_ts, **kwargs)
\ No newline at end of file

Ruyi-Models/ruyi/utils/utils.py

+import os
+import gc
+import imageio
+import numpy as np
+import torch
+import torchvision
+import cv2
+from einops import rearrange
+from PIL import Image
+
+def get_width_and_height_from_image_and_base_resolution(image, base_resolution):
+    target_pixels = int(base_resolution) * int(base_resolution)
+    original_width, original_height = Image.open(image).size
+    ratio = (target_pixels / (original_width * original_height)) ** 0.5
+    width_slider = round(original_width * ratio)
+    height_slider = round(original_height * ratio)
+    return height_slider, width_slider
+
+def color_transfer(sc, dc):
+    """
+    Transfer color distribution from of sc, referred to dc.
+
+    Args:
+        sc (numpy.ndarray): input image to be transfered.
+        dc (numpy.ndarray): reference image
+
+    Returns:
+        numpy.ndarray: Transferred color distribution on the sc.
+    """
+
+    def get_mean_and_std(img):
+        x_mean, x_std = cv2.meanStdDev(img)
+        x_mean = np.hstack(np.around(x_mean, 2))
+        x_std = np.hstack(np.around(x_std, 2))
+        return x_mean, x_std
+
+    sc = cv2.cvtColor(sc, cv2.COLOR_RGB2LAB)
+    s_mean, s_std = get_mean_and_std(sc)
+    dc = cv2.cvtColor(dc, cv2.COLOR_RGB2LAB)
+    t_mean, t_std = get_mean_and_std(dc)
+    img_n = ((sc - s_mean) * (t_std / s_std)) + t_mean
+    np.putmask(img_n, img_n > 255, 255)
+    np.putmask(img_n, img_n < 0, 0)
+    dst = cv2.cvtColor(cv2.convertScaleAbs(img_n), cv2.COLOR_LAB2RGB)
+    return dst
+
+def save_videos_grid(videos: torch.Tensor, path: str, rescale=False, n_rows=6, fps=12, imageio_backend=True, color_transfer_post_process=False):
+    videos = rearrange(videos, "b c t h w -> t b c h w")
+    outputs = []
+    for x in videos:
+        x = torchvision.utils.make_grid(x, nrow=n_rows)
+        x = x.transpose(0, 1).transpose(1, 2).squeeze(-1)
+        if rescale:
+            x = (x + 1.0) / 2.0  # -1,1 -> 0,1
+        x = (x * 255).numpy().astype(np.uint8)
+        outputs.append(Image.fromarray(x))
+
+    if color_transfer_post_process:
+        for i in range(1, len(outputs)):
+            outputs[i] = Image.fromarray(color_transfer(np.uint8(outputs[i]), np.uint8(outputs[0])))
+
+    os.makedirs(os.path.dirname(path), exist_ok=True)
+    if imageio_backend:
+        if path.endswith("mp4"):
+            imageio.mimsave(path, outputs, fps=fps)
+        else:
+            imageio.mimsave(path, outputs, duration=(1000 * 1/fps))
+    else:
+        if path.endswith("mp4"):
+            path = path.replace('.mp4', '.gif')
+        outputs[0].save(path, format='GIF', append_images=outputs, save_all=True, duration=100, loop=0)
+
+def get_image_to_video_latent(validation_image_start, validation_image_end, video_length, sample_size):
+    if validation_image_start is not None and validation_image_end is not None:
+        if type(validation_image_start) is str and os.path.isfile(validation_image_start):
+            image_start = clip_image = Image.open(validation_image_start).convert("RGB")
+            image_start = image_start.resize([sample_size[1], sample_size[0]])
+            clip_image = clip_image.resize([sample_size[1], sample_size[0]])
+        else:
+            image_start = clip_image = validation_image_start
+            image_start = [_image_start.resize([sample_size[1], sample_size[0]]) for _image_start in image_start]
+            clip_image = [_clip_image.resize([sample_size[1], sample_size[0]]) for _clip_image in clip_image]
+
+        if type(validation_image_end) is str and os.path.isfile(validation_image_end):
+            image_end = Image.open(validation_image_end).convert("RGB")
+            image_end = image_end.resize([sample_size[1], sample_size[0]])
+        else:
+            image_end = validation_image_end
+            image_end = [_image_end.resize([sample_size[1], sample_size[0]]) for _image_end in image_end]
+
+        if type(image_start) is list:
+            clip_image = clip_image[0]
+            start_video = torch.cat(
+                [torch.from_numpy(np.array(_image_start)).permute(2, 0, 1).unsqueeze(1).unsqueeze(0) for _image_start in image_start], 
+                dim=2
+            )
+            input_video = torch.tile(start_video[:, :, :1], [1, 1, video_length, 1, 1])
+            input_video[:, :, :len(image_start)] = start_video
+            
+            input_video_mask = torch.zeros_like(input_video[:, :1])
+            input_video_mask[:, :, len(image_start):] = 255
+        else:
+            input_video = torch.tile(
+                torch.from_numpy(np.array(image_start)).permute(2, 0, 1).unsqueeze(1).unsqueeze(0), 
+                [1, 1, video_length, 1, 1]
+            )
+            input_video_mask = torch.zeros_like(input_video[:, :1])
+            input_video_mask[:, :, 3:] = 255
+
+        if type(image_end) is list:
+            image_end = [_image_end.resize(image_start[0].size if type(image_start) is list else image_start.size) for _image_end in image_end]
+            end_video = torch.cat(
+                [torch.from_numpy(np.array(_image_end)).permute(2, 0, 1).unsqueeze(1).unsqueeze(0) for _image_end in image_end], 
+                dim=2
+            )
+            input_video[:, :, -len(end_video):] = end_video
+            
+            input_video_mask[:, :, -len(image_end):] = 0
+        else:
+            image_end = image_end.resize(image_start[0].size if type(image_start) is list else image_start.size)
+            input_video[:, :, -3:] = torch.from_numpy(np.array(image_end)).permute(2, 0, 1).unsqueeze(1).unsqueeze(0)
+            input_video_mask[:, :, -3:] = 0
+
+        input_video = input_video / 255
+
+    elif validation_image_start is not None:
+        if type(validation_image_start) is str and os.path.isfile(validation_image_start):
+            image_start = clip_image = Image.open(validation_image_start).convert("RGB")
+            image_start = image_start.resize([sample_size[1], sample_size[0]])
+            clip_image = clip_image.resize([sample_size[1], sample_size[0]])
+        else:
+            image_start = clip_image = validation_image_start
+            # image_start = clip_image = [Image.open(_image_start).convert("RGB") for _image_start in validation_image_start]
+            image_start = [_image_start.resize([sample_size[1], sample_size[0]]) for _image_start in image_start]
+            clip_image = [_clip_image.resize([sample_size[1], sample_size[0]]) for _clip_image in clip_image]
+        image_end = None
+        
+        if type(image_start) is list:
+            clip_image = clip_image[0]
+            start_video = torch.cat(
+                [torch.from_numpy(np.array(_image_start)).permute(2, 0, 1).unsqueeze(1).unsqueeze(0) for _image_start in image_start], 
+                dim=2
+            )
+            input_video = torch.tile(start_video[:, :, :1], [1, 1, video_length, 1, 1])
+            input_video[:, :, :len(image_start)] = start_video
+            input_video = input_video / 255
+            
+            input_video_mask = torch.zeros_like(input_video[:, :1])
+            input_video_mask[:, :, len(image_start):] = 255
+        else:
+            input_video = torch.tile(
+                torch.from_numpy(np.array(image_start)).permute(2, 0, 1).unsqueeze(1).unsqueeze(0), 
+                [1, 1, video_length, 1, 1]
+            ) / 255
+            input_video_mask = torch.zeros_like(input_video[:, :1])
+            # input_video_mask[:, :, 3:, ] = 255
+            input_video_mask[:, :, 1:, ] = 255
+    else:
+        image_start = None
+        image_end = None
+        input_video = torch.zeros([1, 3, video_length, sample_size[0], sample_size[1]])
+        input_video_mask = torch.ones([1, 1, video_length, sample_size[0], sample_size[1]]) * 255
+        clip_image = None
+
+    del image_start
+    del image_end
+    gc.collect()
+
+    return  input_video, input_video_mask, clip_image
+
+def get_video_to_video_latent(input_video_path, video_length, sample_size):
+    if type(input_video_path) is str:
+        cap = cv2.VideoCapture(input_video_path)
+        input_video = []
+        while True:
+            ret, frame = cap.read()
+            if not ret:
+                break
+            frame = cv2.resize(frame, (sample_size[1], sample_size[0]))
+            input_video.append(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
+        cap.release()
+    else:
+        input_video = input_video_path
+
+    input_video = torch.from_numpy(np.array(input_video))[:video_length]
+    input_video = input_video.permute([3, 0, 1, 2]).unsqueeze(0) / 255
+
+    input_video_mask = torch.zeros_like(input_video[:, :1])
+    input_video_mask[:, :, :] = 255
+
+    return  input_video, input_video_mask, None
\ No newline at end of file

Ruyi-Models/ruyi/vae/LICENSE

+Copyright (c) 2022 Robin Rombach and Patrick Esser and contributors
+
+CreativeML Open RAIL-M
+dated August 22, 2022
+
+Section I: PREAMBLE
+
+Multimodal generative models are being widely adopted and used, and have the potential to transform the way artists, among other individuals, conceive and benefit from AI or ML technologies as a tool for content creation.
+
+Notwithstanding the current and potential benefits that these artifacts can bring to society at large, there are also concerns about potential misuses of them, either due to their technical limitations or ethical considerations.
+
+In short, this license strives for both the open and responsible downstream use of the accompanying model. When it comes to the open character, we took inspiration from open source permissive licenses regarding the grant of IP rights. Referring to the downstream responsible use, we added use-based restrictions not permitting the use of the Model in very specific scenarios, in order for the licensor to be able to enforce the license in case potential misuses of the Model may occur. At the same time, we strive to promote open and responsible research on generative models for art and content generation.
+
+Even though downstream derivative versions of the model could be released under different licensing terms, the latter will always have to include - at minimum - the same use-based restrictions as the ones in the original license (this license). We believe in the intersection between open and responsible AI development; thus, this License aims to strike a balance between both in order to enable responsible open-science in the field of AI.
+
+This License governs the use of the model (and its derivatives) and is informed by the model card associated with the model.
+
+NOW THEREFORE, You and Licensor agree as follows:
+
+1. Definitions
+
+- "License" means the terms and conditions for use, reproduction, and Distribution as defined in this document.
+- "Data" means a collection of information and/or content extracted from the dataset used with the Model, including to train, pretrain, or otherwise evaluate the Model. The Data is not licensed under this License.
+- "Output" means the results of operating a Model as embodied in informational content resulting therefrom.
+- "Model" means any accompanying machine-learning based assemblies (including checkpoints), consisting of learnt weights, parameters (including optimizer states), corresponding to the model architecture as embodied in the Complementary Material, that have been trained or tuned, in whole or in part on the Data, using the Complementary Material.
+- "Derivatives of the Model" means all modifications to the Model, works based on the Model, or any other model which is created or initialized by transfer of patterns of the weights, parameters, activations or output of the Model, to the other model, in order to cause the other model to perform similarly to the Model, including - but not limited to - distillation methods entailing the use of intermediate data representations or methods based on the generation of synthetic data by the Model for training the other model.
+- "Complementary Material" means the accompanying source code and scripts used to define, run, load, benchmark or evaluate the Model, and used to prepare data for training or evaluation, if any. This includes any accompanying documentation, tutorials, examples, etc, if any.
+- "Distribution" means any transmission, reproduction, publication or other sharing of the Model or Derivatives of the Model to a third party, including providing the Model as a hosted service made available by electronic or other remote means - e.g. API-based or web access.
+- "Licensor" means the copyright owner or entity authorized by the copyright owner that is granting the License, including the persons or entities that may have rights in the Model and/or distributing the Model.
+- "You" (or "Your") means an individual or Legal Entity exercising permissions granted by this License and/or making use of the Model for whichever purpose and in any field of use, including usage of the Model in an end-use application - e.g. chatbot, translator, image generator.
+- "Third Parties" means individuals or legal entities that are not under common control with Licensor or You.
+- "Contribution" means any work of authorship, including the original version of the Model and any modifications or additions to that Model or Derivatives of the Model thereof, that is intentionally submitted to Licensor for inclusion in the Model by the copyright owner or by an individual or Legal Entity authorized to submit on behalf of the copyright owner. For the purposes of this definition, "submitted" means any form of electronic, verbal, or written communication sent to the Licensor or its representatives, including but not limited to communication on electronic mailing lists, source code control systems, and issue tracking systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and improving the Model, but excluding communication that is conspicuously marked or otherwise designated in writing by the copyright owner as "Not a Contribution."
+- "Contributor" means Licensor and any individual or Legal Entity on behalf of whom a Contribution has been received by Licensor and subsequently incorporated within the Model.
+
+Section II: INTELLECTUAL PROPERTY RIGHTS
+
+Both copyright and patent grants apply to the Model, Derivatives of the Model and Complementary Material. The Model and Derivatives of the Model are subject to additional terms as described in Section III.
+
+2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare, publicly display, publicly perform, sublicense, and distribute the Complementary Material, the Model, and Derivatives of the Model.
+3. Grant of Patent License. Subject to the terms and conditions of this License and where and as applicable, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this paragraph) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Model and the Complementary Material, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Model to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Model and/or Complementary Material or a Contribution incorporated within the Model and/or Complementary Material constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for the Model and/or Work shall terminate as of the date such litigation is asserted or filed.
+
+Section III: CONDITIONS OF USAGE, DISTRIBUTION AND REDISTRIBUTION
+
+4. Distribution and Redistribution. You may host for Third Party remote access purposes (e.g. software-as-a-service), reproduce and distribute copies of the Model or Derivatives of the Model thereof in any medium, with or without modifications, provided that You meet the following conditions:
+Use-based restrictions as referenced in paragraph 5 MUST be included as an enforceable provision by You in any type of legal agreement (e.g. a license) governing the use and/or distribution of the Model or Derivatives of the Model, and You shall give notice to subsequent users You Distribute to, that the Model or Derivatives of the Model are subject to paragraph 5. This provision does not apply to the use of Complementary Material.
+You must give any Third Party recipients of the Model or Derivatives of the Model a copy of this License;
+You must cause any modified files to carry prominent notices stating that You changed the files;
+You must retain all copyright, patent, trademark, and attribution notices excluding those notices that do not pertain to any part of the Model, Derivatives of the Model.
+You may add Your own copyright statement to Your modifications and may provide additional or different license terms and conditions - respecting paragraph 4.a. - for use, reproduction, or Distribution of Your modifications, or for any such Derivatives of the Model as a whole, provided Your use, reproduction, and Distribution of the Model otherwise complies with the conditions stated in this License.
+5. Use-based restrictions. The restrictions set forth in Attachment A are considered Use-based restrictions. Therefore You cannot use the Model and the Derivatives of the Model for the specified restricted uses. You may use the Model subject to this License, including only for lawful purposes and in accordance with the License. Use may include creating any content with, finetuning, updating, running, training, evaluating and/or reparametrizing the Model. You shall require all of Your users who use the Model or a Derivative of the Model to comply with the terms of this paragraph (paragraph 5).
+6. The Output You Generate. Except as set forth herein, Licensor claims no rights in the Output You generate using the Model. You are accountable for the Output you generate and its subsequent uses. No use of the output can contravene any provision as stated in the License.
+
+Section IV: OTHER PROVISIONS
+
+7. Updates and Runtime Restrictions. To the maximum extent permitted by law, Licensor reserves the right to restrict (remotely or otherwise) usage of the Model in violation of this License, update the Model through electronic means, or modify the Output of the Model based on updates. You shall undertake reasonable efforts to use the latest version of the Model.
+8. Trademarks and related. Nothing in this License permits You to make use of Licensors’ trademarks, trade names, logos or to otherwise suggest endorsement or misrepresent the relationship between the parties; and any rights not expressly granted herein are reserved by the Licensors.
+9. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides the Model and the Complementary Material (and each Contributor provides its Contributions) on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied, including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining the appropriateness of using or redistributing the Model, Derivatives of the Model, and the Complementary Material and assume any risks associated with Your exercise of permissions under this License.
+10. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Model and the Complementary Material (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages.
+11. Accepting Warranty or Additional Liability. While redistributing the Model, Derivatives of the Model and the Complementary Material thereof, You may choose to offer, and charge a fee for, acceptance of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License. However, in accepting such obligations, You may act only on Your own behalf and on Your sole responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor by reason of your accepting any such warranty or additional liability.
+12. If any provision of this License is held to be invalid, illegal or unenforceable, the remaining provisions shall be unaffected thereby and remain valid as if such provision had not been set forth herein.
+
+END OF TERMS AND CONDITIONS
+
+
+
+
+Attachment A
+
+Use Restrictions
+
+You agree not to use the Model or Derivatives of the Model:
+- In any way that violates any applicable national, federal, state, local or international law or regulation;
+- For the purpose of exploiting, harming or attempting to exploit or harm minors in any way;
+- To generate or disseminate verifiably false information and/or content with the purpose of harming others;
+- To generate or disseminate personal identifiable information that can be used to harm an individual;
+- To defame, disparage or otherwise harass others;
+- For fully automated decision making that adversely impacts an individual’s legal rights or otherwise creates or modifies a binding, enforceable obligation;
+- For any use intended to or which has the effect of discriminating against or harming individuals or groups based on online or offline social behavior or known or predicted personal or personality characteristics;
+- To exploit any of the vulnerabilities of a specific group of persons based on their age, social, physical or mental characteristics, in order to materially distort the behavior of a person pertaining to that group in a manner that causes or is likely to cause that person or another person physical or psychological harm;
+- For any use intended to or which has the effect of discriminating against individuals or groups based on legally protected characteristics or categories;
+- To provide medical advice and medical results interpretation;
+- To generate or disseminate information for the purpose to be used for administration of justice, law enforcement, immigration or asylum processes, such as predicting an individual will commit fraud/crime commitment (e.g. by text profiling, drawing causal relationships between assertions made in documents, indiscriminate and arbitrarily-targeted use).

Ruyi-Models/ruyi/vae/README.md

+## VAE Training
+
+English | [简体中文](./README_zh-CN.md)
+
+After completing data preprocessing, we can obtain the following dataset:
+
+```
+📦 project/
+├── 📂 datasets/
+│   ├── 📂 internal_datasets/
+│       ├── 📂 videos/
+│       │   ├── 📄 00000001.mp4
+│       │   ├── 📄 00000001.jpg
+│       │   └── 📄 .....
+│       └── 📄 json_of_internal_datasets.json
+```
+
+The json_of_internal_datasets.json is a standard JSON file. The file_path in the json can to be set as relative path, as shown in below:
+```json
+[
+    {
+      "file_path": "videos/00000001.mp4",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "video"
+    },
+    {
+      "file_path": "train/00000001.jpg",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "image"
+    },
+    .....
+]
+```
+
+You can also set the path as absolute path as follow:
+```json
+[
+    {
+      "file_path": "/mnt/data/videos/00000001.mp4",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "video"
+    },
+    {
+      "file_path": "/mnt/data/train/00000001.jpg",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "image"
+    },
+    .....
+]
+```
+
+## Train Video VAE
+We need to set config in ```easyanimate/vae/configs/autoencoder``` at first. The default config is ```autoencoder_kl_32x32x4_slice.yaml```. We need to set the some params in yaml file. 
+
+- ```data_json_path``` corresponds to the JSON file of the dataset. 
+- ```data_root``` corresponds to the root path of the dataset. If you want to use absolute path in json file, please delete this line.
+- ```ckpt_path``` corresponds to the pretrained weights of the vae. 
+- ```gpus``` and num_nodes need to be set as the actual situation of your machine. 
+
+The we run shell file as follow: 
+```
+sh scripts/train_vae.sh
+```
\ No newline at end of file

Ruyi-Models/ruyi/vae/README_zh-CN.md

+## VAE 训练
+
+[English](./README.md) | 简体中文
+
+在完成数据预处理后，你可以获得这样的数据格式:
+
+```
+📦 project/
+├── 📂 datasets/
+│   ├── 📂 internal_datasets/
+│       ├── 📂 videos/
+│       │   ├── 📄 00000001.mp4
+│       │   ├── 📄 00000001.jpg
+│       │   └── 📄 .....
+│       └── 📄 json_of_internal_datasets.json
+```
+
+json_of_internal_datasets.json是一个标准的json文件。json中的file_path可以被设置为相对路径，如下所示：
+```json
+[
+    {
+      "file_path": "videos/00000001.mp4",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "video"
+    },
+    {
+      "file_path": "train/00000001.jpg",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "image"
+    },
+    .....
+]
+```
+
+你也可以将路径设置为绝对路径：
+```json
+[
+    {
+      "file_path": "/mnt/data/videos/00000001.mp4",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "video"
+    },
+    {
+      "file_path": "/mnt/data/train/00000001.jpg",
+      "text": "A group of young men in suits and sunglasses are walking down a city street.",
+      "type": "image"
+    },
+    .....
+]
+```
+
+## 训练 Video VAE
+我们首先需要修改 ```easyanimate/vae/configs/autoencoder``` 中的配置文件。默认的配置文件是 ```autoencoder_kl_32x32x4_slice.yaml```。你需要修改以下参数： 
+
+- ```data_json_path``` json file 所在的目录。 
+- ```data_root``` 数据的根目录。如果你在json file中使用了绝对路径，请设置为空。
+- ```ckpt_path``` 预训练的vae模型路径。 
+- ```gpus``` 以及 ```num_nodes``` 需要设置为你机器的实际gpu数目。 
+
+运行以下的脚本来训练vae: 
+```
+sh scripts/train_vae.sh
+```
\ No newline at end of file

Ruyi-Models/ruyi/vae/configs/autoencoder/autoencoder_kl_32x32x4_mag.yaml

+model:
+  base_learning_rate: 1.0e-04
+  target: easyanimate.vae.ldm.models.omnigen_casual3dcnn.AutoencoderKLMagvit_fromOmnigen
+  params:
+    monitor: train/rec_loss
+    ckpt_path: models/videoVAE_omnigen_8x8x4_from_vae-ft-mse-840000-ema-pruned.ckpt
+    down_block_types: ("SpatialDownBlock3D", "SpatialTemporalDownBlock3D", "SpatialTemporalDownBlock3D",
+      "SpatialTemporalDownBlock3D",)
+    up_block_types: ("SpatialUpBlock3D", "SpatialTemporalUpBlock3D", "SpatialTemporalUpBlock3D",
+      "SpatialTemporalUpBlock3D",)
+    lossconfig:
+      target: easyanimate.vae.ldm.modules.losses.LPIPSWithDiscriminator
+      params:
+        disc_start: 50001
+        kl_weight: 1.0e-06
+        disc_weight: 0.5
+        l2_loss_weight: 0.1
+        l1_loss_weight: 1.0
+        perceptual_weight: 1.0
+
+data:
+  target: train_vae.DataModuleFromConfig
+
+  params:
+    batch_size: 2
+    wrap: true
+    num_workers: 4
+    train:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRTrain
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 128
+        degradation: pil_nearest
+        video_size: 128
+        video_len: 9
+        slice_interval: 1
+    validation:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRValidation
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 128
+        degradation: pil_nearest
+        video_size: 128
+        video_len: 9
+        slice_interval: 1
+
+lightning:
+  callbacks:
+    image_logger:
+      target: train_vae.ImageLogger
+      params:
+        batch_frequency: 5000
+        max_images: 8
+        increase_log_steps: True
+
+  trainer:
+    benchmark: True
+    accumulate_grad_batches: 1
+    gpus: "0"
+    num_nodes: 1
\ No newline at end of file

Ruyi-Models/ruyi/vae/configs/autoencoder/autoencoder_kl_32x32x4_slice.yaml

+model:
+  base_learning_rate: 1.0e-04
+  target: easyanimate.vae.ldm.models.omnigen_casual3dcnn.AutoencoderKLMagvit_fromOmnigen
+  params:
+    slice_compression_vae: true
+    mini_batch_encoder: 8
+    mini_batch_decoder: 2
+    monitor: train/rec_loss
+    ckpt_path: models/Diffusion_Transformer/EasyAnimateV2-XL-2-512x512/vae/diffusion_pytorch_model.safetensors
+    down_block_types: ("SpatialDownBlock3D", "SpatialTemporalDownBlock3D", "SpatialTemporalDownBlock3D",
+      "SpatialTemporalDownBlock3D",)
+    up_block_types: ("SpatialUpBlock3D", "SpatialTemporalUpBlock3D", "SpatialTemporalUpBlock3D",
+      "SpatialTemporalUpBlock3D",)
+    lossconfig:
+      target: easyanimate.vae.ldm.modules.losses.LPIPSWithDiscriminator
+      params:
+        disc_start: 50001
+        kl_weight: 1.0e-06
+        disc_weight: 0.5
+        l2_loss_weight: 0.0
+        l1_loss_weight: 1.0
+        perceptual_weight: 1.0
+
+data:
+  target: train_vae.DataModuleFromConfig
+
+  params:
+    batch_size: 1
+    wrap: true
+    num_workers: 8
+    train:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRTrain
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 256
+        degradation: pil_nearest
+        video_size: 256
+        video_len: 25
+        slice_interval: 1
+    validation:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRValidation
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 256
+        degradation: pil_nearest
+        video_size: 256
+        video_len: 25
+        slice_interval: 1
+
+lightning:
+  callbacks:
+    image_logger:
+      target: train_vae.ImageLogger
+      params:
+        batch_frequency: 5000
+        max_images: 8
+        increase_log_steps: True
+
+  trainer:
+    benchmark: True
+    accumulate_grad_batches: 1
+    gpus: "0"
+    num_nodes: 1
\ No newline at end of file

Ruyi-Models/ruyi/vae/configs/autoencoder/autoencoder_kl_32x32x4_slice_decoder_only.yaml

+model:
+  base_learning_rate: 1.0e-04
+  target: easyanimate.vae.ldm.models.omnigen_casual3dcnn.AutoencoderKLMagvit_fromOmnigen
+  params:
+    slice_compression_vae: true
+    train_decoder_only: true
+    mini_batch_encoder: 8
+    mini_batch_decoder: 2
+    monitor: train/rec_loss
+    ckpt_path: models/Diffusion_Transformer/EasyAnimateV2-XL-2-512x512/vae/diffusion_pytorch_model.safetensors
+    down_block_types: ("SpatialDownBlock3D", "SpatialTemporalDownBlock3D", "SpatialTemporalDownBlock3D",
+      "SpatialTemporalDownBlock3D",)
+    up_block_types: ("SpatialUpBlock3D", "SpatialTemporalUpBlock3D", "SpatialTemporalUpBlock3D",
+      "SpatialTemporalUpBlock3D",)
+    lossconfig:
+      target: easyanimate.vae.ldm.modules.losses.LPIPSWithDiscriminator
+      params:
+        disc_start: 50001
+        kl_weight: 1.0e-06
+        disc_weight: 0.5
+        l2_loss_weight: 1.0
+        l1_loss_weight: 0.0
+        perceptual_weight: 1.0
+
+data:
+  target: train_vae.DataModuleFromConfig
+
+  params:
+    batch_size: 1
+    wrap: true
+    num_workers: 8
+    train:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRTrain
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 256
+        degradation: pil_nearest
+        video_size: 256
+        video_len: 25
+        slice_interval: 1
+    validation:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRValidation
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 256
+        degradation: pil_nearest
+        video_size: 256
+        video_len: 25
+        slice_interval: 1
+
+lightning:
+  callbacks:
+    image_logger:
+      target: train_vae.ImageLogger
+      params:
+        batch_frequency: 5000
+        max_images: 8
+        increase_log_steps: True
+
+  trainer:
+    benchmark: True
+    accumulate_grad_batches: 1
+    gpus: "0"
+    num_nodes: 1
\ No newline at end of file

Ruyi-Models/ruyi/vae/configs/autoencoder/autoencoder_kl_32x32x4_slice_t_downsample_8.yaml

+model:
+  base_learning_rate: 1.0e-04
+  target: easyanimate.vae.ldm.models.omnigen_casual3dcnn.AutoencoderKLMagvit_fromOmnigen
+  params:
+    slice_compression_vae: true
+    mini_batch_encoder: 8
+    mini_batch_decoder: 1
+    monitor: train/rec_loss
+    ckpt_path: models/Diffusion_Transformer/EasyAnimateV2-XL-2-512x512/vae/diffusion_pytorch_model.safetensors
+    down_block_types: ("SpatialTemporalDownBlock3D", "SpatialTemporalDownBlock3D", "SpatialTemporalDownBlock3D",
+      "SpatialTemporalDownBlock3D",)
+    up_block_types: ("SpatialTemporalUpBlock3D", "SpatialTemporalUpBlock3D", "SpatialTemporalUpBlock3D",
+      "SpatialTemporalUpBlock3D",)
+    lossconfig:
+      target: easyanimate.vae.ldm.modules.losses.LPIPSWithDiscriminator
+      params:
+        disc_start: 50001
+        kl_weight: 1.0e-06
+        disc_weight: 0.5
+        l2_loss_weight: 0.0
+        l1_loss_weight: 1.0
+        perceptual_weight: 1.0
+
+
+data:
+  target: train_vae.DataModuleFromConfig
+
+  params:
+    batch_size: 1
+    wrap: true
+    num_workers: 8
+    train:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRTrain
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 256
+        degradation: pil_nearest
+        video_size: 256
+        video_len: 33
+        slice_interval: 1
+    validation:
+      target: easyanimate.vae.ldm.data.dataset_image_video.CustomSRValidation
+      params:
+        data_json_path: pretrain.json
+        data_root: /your_data_root # This is used in relative path
+        size: 256
+        degradation: pil_nearest
+        video_size: 256
+        video_len: 33
+        slice_interval: 1
+
+lightning:
+  callbacks:
+    image_logger:
+      target: train_vae.ImageLogger
+      params:
+        batch_frequency: 5000
+        max_images: 8
+        increase_log_steps: True
+
+  trainer:
+    benchmark: True
+    accumulate_grad_batches: 1
+    gpus: "0"
+    num_nodes: 1
\ No newline at end of file

Ruyi-Models/ruyi/vae/environment.yaml

+name: ldm
+channels:
+  - pytorch
+  - defaults
+dependencies:
+  - python=3.8.5
+  - pip=20.3
+  - cudatoolkit=11.3
+  - pytorch=1.11.0
+  - torchvision=0.12.0
+  - numpy=1.19.2
+  - pip:
+    - albumentations==0.4.3
+    - diffusers
+    - opencv-python==4.1.2.30
+    - pudb==2019.2
+    - invisible-watermark
+    - imageio==2.9.0
+    - imageio-ffmpeg==0.4.2
+    - pytorch-lightning==1.4.2
+    - omegaconf==2.1.1
+    - test-tube>=0.7.5
+    - streamlit>=0.73.1
+    - einops==0.3.0
+    - torch-fidelity==0.3.0
+    - transformers==4.19.2
+    - torchmetrics==0.6.0
+    - kornia==0.6
+    - -e .

Ruyi-Models/ruyi/vae/ldm/data/base.py

+from abc import abstractmethod
+
+from torch.utils.data import (ChainDataset, ConcatDataset, Dataset,
+                              IterableDataset)
+
+
+class Txt2ImgIterableBaseDataset(IterableDataset):
+    '''
+    Define an interface to make the IterableDatasets for text2img data chainable
+    '''
+    def __init__(self, num_records=0, valid_ids=None, size=256):
+        super().__init__()
+        self.num_records = num_records
+        self.valid_ids = valid_ids
+        self.sample_ids = valid_ids
+        self.size = size
+
+        print(f'{self.__class__.__name__} dataset contains {self.__len__()} examples.')
+
+    def __len__(self):
+        return self.num_records
+
+    @abstractmethod
+    def __iter__(self):
+        pass
\ No newline at end of file

Ruyi-Models/ruyi/vae/ldm/data/dataset_callback.py

+#-*- encoding:utf-8 -*-
+from pytorch_lightning.callbacks import Callback
+
+class DatasetCallback(Callback):
+    def __init__(self):
+        self.sampler_pos_start = 0
+        self.preload_used_idx_flag = False
+
+    def on_train_start(self, trainer, pl_module):
+        if not self.preload_used_idx_flag:
+            self.preload_used_idx_flag = True
+            trainer.train_dataloader.batch_sampler.sampler_pos_reload = self.sampler_pos_start
+
+    def on_save_checkpoint(self, trainer, pl_module, checkpoint):
+        if trainer.train_dataloader is not None:
+            # Save sampler_pos_start parameters in the checkpoint
+            checkpoint['sampler_pos_start'] = trainer.train_dataloader.batch_sampler.sampler_pos_start
+
+    def on_load_checkpoint(self, trainer, pl_module, checkpoint):
+        # Restore sampler_pos_start parameters from the checkpoint
+        if 'sampler_pos_start' in checkpoint:
+            self.sampler_pos_start = checkpoint.get('sampler_pos_start', 0)
+            print('Load sampler_pos_start from checkpoint, sampler_pos_start = %d' % self.sampler_pos_start)
+        else:
+            print('The sampler_pos_start is not in checkpoint')
\ No newline at end of file

Ruyi-Models/ruyi/vae/ldm/data/dataset_image_video.py

+import glob
+import json
+import os
+import pickle
+import random
+import shutil
+import tarfile
+from functools import partial
+
+import albumentations
+import cv2
+import numpy as np
+import PIL
+import torchvision.transforms.functional as TF
+import yaml
+from decord import VideoReader
+from func_timeout import FunctionTimedOut, func_set_timeout
+from omegaconf import OmegaConf
+from PIL import Image
+from torch.utils.data import (BatchSampler, Dataset, Sampler)
+from tqdm import tqdm
+
+from ..modules.image_degradation import (degradation_fn_bsr,
+                                         degradation_fn_bsr_light)
+
+
+class ImageVideoSampler(BatchSampler):
+    """A sampler wrapper for grouping images with similar aspect ratio into a same batch.
+
+    Args:
+        sampler (Sampler): Base sampler.
+        dataset (Dataset): Dataset providing data information.
+        batch_size (int): Size of mini-batch.
+        drop_last (bool): If ``True``, the sampler will drop the last batch if
+            its size would be less than ``batch_size``.
+        aspect_ratios (dict): The predefined aspect ratios.
+    """
+
+    def __init__(self,
+                 sampler: Sampler,
+                 dataset: Dataset,
+                 batch_size: int,
+                 drop_last: bool = False
+                ) -> None:
+        if not isinstance(sampler, Sampler):
+            raise TypeError('sampler should be an instance of ``Sampler``, '
+                            f'but got {sampler}')
+        if not isinstance(batch_size, int) or batch_size <= 0:
+            raise ValueError('batch_size should be a positive integer value, '
+                             f'but got batch_size={batch_size}')
+        self.sampler = sampler
+        self.dataset = dataset
+        self.batch_size = batch_size
+        self.drop_last = drop_last
+
+        self.sampler_pos_start = 0
+        self.sampler_pos_reload = 0
+
+        self.num_samples_random = len(self.sampler)
+        # buckets for each aspect ratio
+        self.bucket = {'image':[], 'video':[]}
+            
+    def set_epoch(self, epoch):
+        if hasattr(self.sampler, "set_epoch"):
+            self.sampler.set_epoch(epoch)
+
+    def __iter__(self):
+        for index_sampler, idx in enumerate(self.sampler):
+            if self.sampler_pos_reload != 0 and self.sampler_pos_reload < self.num_samples_random:
+                if index_sampler < self.sampler_pos_reload:
+                    self.sampler_pos_start = (self.sampler_pos_start + 1) % self.num_samples_random
+                    continue
+                elif index_sampler == self.sampler_pos_reload:
+                    self.sampler_pos_reload = 0
+                        
+            content_type = self.dataset.data.get_type(idx)
+            bucket = self.bucket[content_type]
+            bucket.append(idx)
+            # yield a batch of indices in the same aspect ratio group
+            if len(self.bucket['video']) == self.batch_size:
+                yield self.bucket['video']
+                self.bucket['video'] = []
+            elif len(self.bucket['image']) == self.batch_size:
+                yield self.bucket['image']
+                self.bucket['image'] = []
+            self.sampler_pos_start = (self.sampler_pos_start + 1) % self.num_samples_random
+
+class ImageVideoDataset(Dataset): 
+    # update __getitem__() from ImageNetSR. If timeout for Pandas70M, throw exception.
+    # If caught exception(timeout or others), try another index until successful and return.
+    def __init__(self, size=None, video_size=128, video_len=25,
+                 degradation=None, downscale_f=4,  random_crop=True, min_crop_f=0.25, max_crop_f=1.,
+                 s_t=None, slice_interval=None, data_root=None
+                ):
+        """
+        Imagenet Superresolution Dataloader
+        Performs following ops in order:
+        1.  crops a crop of size s from image either as random or center crop
+        2.  resizes crop to size with cv2.area_interpolation
+        3.  degrades resized crop with degradation_fn
+
+        :param size: resizing to size after cropping
+        :param degradation: degradation_fn, e.g. cv_bicubic or bsrgan_light
+        :param downscale_f: Low Resolution Downsample factor
+        :param min_crop_f: determines crop size s,
+          where s = c * min_img_side_len with c sampled from interval (min_crop_f, max_crop_f)
+        :param max_crop_f: ""
+        :param data_root:
+        :param random_crop:
+        """
+        self.base = self.get_base()
+        assert size
+        assert (size / downscale_f).is_integer()
+        self.size = size
+        self.LR_size = int(size / downscale_f)
+        self.min_crop_f = min_crop_f
+        self.max_crop_f = max_crop_f
+        assert(max_crop_f <= 1.)
+        self.center_crop = not random_crop
+        self.s_t = s_t
+        self.slice_interval = slice_interval
+
+        self.image_rescaler = albumentations.SmallestMaxSize(max_size=size, interpolation=cv2.INTER_AREA)
+        self.video_rescaler = albumentations.SmallestMaxSize(max_size=video_size, interpolation=cv2.INTER_AREA)
+        self.video_len = video_len
+        self.video_size = video_size
+        self.data_root = data_root
+
+        self.pil_interpolation = False # gets reset later if incase interp_op is from pillow
+
+        if degradation == "bsrgan":
+            self.degradation_process = partial(degradation_fn_bsr, sf=downscale_f)
+
+        elif degradation == "bsrgan_light":
+            self.degradation_process = partial(degradation_fn_bsr_light, sf=downscale_f)
+        else:
+            interpolation_fn = {
+                "cv_nearest": cv2.INTER_NEAREST,
+                "cv_bilinear": cv2.INTER_LINEAR,
+                "cv_bicubic": cv2.INTER_CUBIC,
+                "cv_area": cv2.INTER_AREA,
+                "cv_lanczos": cv2.INTER_LANCZOS4,
+                "pil_nearest": PIL.Image.NEAREST,  
+                "pil_bilinear": PIL.Image.BILINEAR,
+                "pil_bicubic": PIL.Image.BICUBIC,
+                "pil_box": PIL.Image.BOX,
+                "pil_hamming": PIL.Image.HAMMING,
+                "pil_lanczos": PIL.Image.LANCZOS,
+            }[degradation]
+
+            self.pil_interpolation = degradation.startswith("pil_")
+
+            if self.pil_interpolation:                
+                self.degradation_process = partial(TF.resize, size=self.LR_size, interpolation=interpolation_fn)
+
+            else:
+                self.degradation_process = albumentations.SmallestMaxSize(max_size=self.LR_size,
+                                                                          interpolation=interpolation_fn)
+
+    def __len__(self):
+        return len(self.base)
+    
+    def get_type(self, index):
+        return self.base[index].get('type', 'image')
+
+    def __getitem__(self, i):
+        @func_set_timeout(3) # time wait 3 seconds
+        def get_video_item(example):
+            if self.data_root is not None:
+                video_reader = VideoReader(os.path.join(self.data_root, example['file_path']))
+            else:
+                video_reader = VideoReader(example['file_path'])
+            video_length = len(video_reader)
+            
+            clip_length = min(video_length, (self.video_len - 1) * self.slice_interval + 1)
+            start_idx   = random.randint(0, video_length - clip_length)
+            batch_index = np.linspace(start_idx, start_idx + clip_length - 1, self.video_len, dtype=int)
+
+            pixel_values = video_reader.get_batch(batch_index).asnumpy()
+            
+            del video_reader
+            out_images = []
+            LR_out_images = []
+            min_side_len = min(pixel_values[0].shape[:2])
+
+            crop_side_len = min_side_len * np.random.uniform(self.min_crop_f, self.max_crop_f, size=None)
+            crop_side_len = int(crop_side_len)
+            if self.center_crop:
+                self.cropper = albumentations.CenterCrop(height=crop_side_len, width=crop_side_len)
+            else:
+                self.cropper = albumentations.RandomCrop(height=crop_side_len, width=crop_side_len)
+
+            imgs = np.transpose(pixel_values, (1, 2, 3, 0))
+            imgs = self.cropper(image=imgs)["image"]
+            imgs = np.transpose(imgs, (3, 0, 1, 2))
+            for img in imgs:
+                image = self.video_rescaler(image=img)["image"]
+                out_images.append(image[None, :, :, :])
+                if self.pil_interpolation:
+                    image_pil = PIL.Image.fromarray(image)
+                    LR_image = self.degradation_process(image_pil)
+                    LR_image = np.array(LR_image).astype(np.uint8)
+                else:
+                    LR_image = self.degradation_process(image=image)["image"]
+                LR_out_images.append(LR_image[None, :, :, :])
+
+            example = {}
+            example['image'] = (np.concatenate(out_images) / 127.5 - 1.0).astype(np.float32)
+            example['LR_image'] = (np.concatenate(LR_out_images) / 127.5 - 1.0).astype(np.float32)
+            return example
+
+        example = self.base[i]
+        if example.get('type', 'image') == 'video':
+            while True:  
+                try:
+                    example = self.base[i]
+                    return get_video_item(example)
+                except FunctionTimedOut:
+                    print("stt catch: Function 'extract failed' timed out.")
+                    i = random.randint(0, self.__len__() - 1)
+                except Exception as e:
+                    print('stt catch', e)
+                    i = random.randint(0, self.__len__() - 1)
+        elif example.get('type', 'image') == 'image':
+            while True:
+                try:
+                    example = self.base[i]
+                    if self.data_root is not None:
+                        image = Image.open(os.path.join(self.data_root, example['file_path']))
+                    else:
+                        image = Image.open(example['file_path'])
+                    image = image.convert("RGB")
+                    image = np.array(image).astype(np.uint8)
+
+                    min_side_len    = min(image.shape[:2])
+                    crop_side_len   = min_side_len * np.random.uniform(self.min_crop_f, self.max_crop_f, size=None)
+                    crop_side_len   = int(crop_side_len)
+
+                    if self.center_crop:
+                        self.cropper = albumentations.CenterCrop(height=crop_side_len, width=crop_side_len)
+
+                    else:
+                        self.cropper = albumentations.RandomCrop(height=crop_side_len, width=crop_side_len)
+
+                    image = self.cropper(image=image)["image"]
+
+                    image = self.image_rescaler(image=image)["image"]
+
+                    if self.pil_interpolation:
+                        image_pil = PIL.Image.fromarray(image)
+                        LR_image = self.degradation_process(image_pil)
+                        LR_image = np.array(LR_image).astype(np.uint8)
+
+                    else:
+                        LR_image = self.degradation_process(image=image)["image"]
+                    
+                    example = {}
+                    example["image"] = (image/127.5 - 1.0).astype(np.float32)
+                    example["LR_image"] = (LR_image/127.5 - 1.0).astype(np.float32)
+                    return example
+                except Exception as e:
+                    print("catch", e)
+                    i = random.randint(0, self.__len__() - 1)
+
+class CustomSRTrain(ImageVideoDataset):
+    def __init__(self, data_json_path, **kwargs):
+        self.data_json_path = data_json_path
+        super().__init__(**kwargs)
+
+    def get_base(self):
+        return [ann for ann in json.load(open(self.data_json_path))]
+
+class CustomSRValidation(ImageVideoDataset):
+    def __init__(self, data_json_path, **kwargs):
+        self.data_json_path = data_json_path
+        super().__init__(**kwargs)
+        self.data_json_path = data_json_path
+
+    def get_base(self):
+        return [ann for ann in json.load(open(self.data_json_path))][:100] + \
+              [ann for ann in json.load(open(self.data_json_path))][-100:]

Ruyi-Models/ruyi/vae/ldm/lr_scheduler.py

+import numpy as np
+
+
+class LambdaWarmUpCosineScheduler:
+    """
+    note: use with a base_lr of 1.0
+    """
+    def __init__(self, warm_up_steps, lr_min, lr_max, lr_start, max_decay_steps, verbosity_interval=0):
+        self.lr_warm_up_steps = warm_up_steps
+        self.lr_start = lr_start
+        self.lr_min = lr_min
+        self.lr_max = lr_max
+        self.lr_max_decay_steps = max_decay_steps
+        self.last_lr = 0.
+        self.verbosity_interval = verbosity_interval
+
+    def schedule(self, n, **kwargs):
+        if self.verbosity_interval > 0:
+            if n % self.verbosity_interval == 0: print(f"current step: {n}, recent lr-multiplier: {self.last_lr}")
+        if n < self.lr_warm_up_steps:
+            lr = (self.lr_max - self.lr_start) / self.lr_warm_up_steps * n + self.lr_start
+            self.last_lr = lr
+            return lr
+        else:
+            t = (n - self.lr_warm_up_steps) / (self.lr_max_decay_steps - self.lr_warm_up_steps)
+            t = min(t, 1.0)
+            lr = self.lr_min + 0.5 * (self.lr_max - self.lr_min) * (
+                    1 + np.cos(t * np.pi))
+            self.last_lr = lr
+            return lr
+
+    def __call__(self, n, **kwargs):
+        return self.schedule(n,**kwargs)
+
+
+class LambdaWarmUpCosineScheduler2:
+    """
+    supports repeated iterations, configurable via lists
+    note: use with a base_lr of 1.0.
+    """
+    def __init__(self, warm_up_steps, f_min, f_max, f_start, cycle_lengths, verbosity_interval=0):
+        assert len(warm_up_steps) == len(f_min) == len(f_max) == len(f_start) == len(cycle_lengths)
+        self.lr_warm_up_steps = warm_up_steps
+        self.f_start = f_start
+        self.f_min = f_min
+        self.f_max = f_max
+        self.cycle_lengths = cycle_lengths
+        self.cum_cycles = np.cumsum([0] + list(self.cycle_lengths))
+        self.last_f = 0.
+        self.verbosity_interval = verbosity_interval
+
+    def find_in_interval(self, n):
+        interval = 0
+        for cl in self.cum_cycles[1:]:
+            if n <= cl:
+                return interval
+            interval += 1
+
+    def schedule(self, n, **kwargs):
+        cycle = self.find_in_interval(n)
+        n = n - self.cum_cycles[cycle]
+        if self.verbosity_interval > 0:
+            if n % self.verbosity_interval == 0: print(f"current step: {n}, recent lr-multiplier: {self.last_f}, "
+                                                       f"current cycle {cycle}")
+        if n < self.lr_warm_up_steps[cycle]:
+            f = (self.f_max[cycle] - self.f_start[cycle]) / self.lr_warm_up_steps[cycle] * n + self.f_start[cycle]
+            self.last_f = f
+            return f
+        else:
+            t = (n - self.lr_warm_up_steps[cycle]) / (self.cycle_lengths[cycle] - self.lr_warm_up_steps[cycle])
+            t = min(t, 1.0)
+            f = self.f_min[cycle] + 0.5 * (self.f_max[cycle] - self.f_min[cycle]) * (
+                    1 + np.cos(t * np.pi))
+            self.last_f = f
+            return f
+
+    def __call__(self, n, **kwargs):
+        return self.schedule(n, **kwargs)
+
+
+class LambdaLinearScheduler(LambdaWarmUpCosineScheduler2):
+
+    def schedule(self, n, **kwargs):
+        cycle = self.find_in_interval(n)
+        n = n - self.cum_cycles[cycle]
+        if self.verbosity_interval > 0:
+            if n % self.verbosity_interval == 0: print(f"current step: {n}, recent lr-multiplier: {self.last_f}, "
+                                                       f"current cycle {cycle}")
+
+        if n < self.lr_warm_up_steps[cycle]:
+            f = (self.f_max[cycle] - self.f_start[cycle]) / self.lr_warm_up_steps[cycle] * n + self.f_start[cycle]
+            self.last_f = f
+            return f
+        else:
+            f = self.f_min[cycle] + (self.f_max[cycle] - self.f_min[cycle]) * (self.cycle_lengths[cycle] - n) / (self.cycle_lengths[cycle])
+            self.last_f = f
+            return f
+

Ruyi-Models/ruyi/vae/ldm/models/autoencoder.py

+import time
+from contextlib import contextmanager
+
+import pytorch_lightning as pl
+import torch
+import torch.nn.functional as F
+
+from ..modules.diffusionmodules.model import Decoder, Encoder
+from ..modules.distributions.distributions import DiagonalGaussianDistribution
+from ..util import instantiate_from_config
+from .enc_dec_pytorch import Decoder as Mag_Decoder
+from .enc_dec_pytorch import Encoder as Mag_Encoder
+
+
+class AutoencoderKLMagvit(pl.LightningModule):
+    def __init__(self,
+                 ddconfig,
+                 lossconfig,
+                 embed_dim,
+                 ckpt_path=None,
+                 ignore_keys=[],
+                 image_key="image",
+                 colorize_nlabels=None,
+                 monitor=None,
+                 ):
+        super().__init__()
+        self.image_key = image_key
+        self.encoder = Mag_Encoder()
+        self.decoder = Mag_Decoder()
+        self.loss = instantiate_from_config(lossconfig)
+        self.quant_conv = torch.nn.Conv3d(16, 16, 1)
+        self.post_quant_conv = torch.nn.Conv3d(8, 8, 1)
+        self.embed_dim = embed_dim
+        if colorize_nlabels is not None:
+            assert type(colorize_nlabels)==int
+            self.register_buffer("colorize", torch.randn(3, colorize_nlabels, 1, 1))
+        if monitor is not None:
+            self.monitor = monitor
+        if ckpt_path is not None:
+            self.init_from_ckpt(ckpt_path, ignore_keys=ignore_keys)
+
+    def init_from_ckpt(self, path, ignore_keys=list()):
+        sd = torch.load(path, map_location="cpu")["state_dict"]
+        keys = list(sd.keys())
+        for k in keys:
+            for ik in ignore_keys:
+                if k.startswith(ik):
+                    print("Deleting key {} from state_dict.".format(k))
+                    del sd[k]
+        self.load_state_dict(sd, strict=False)
+        print(f"Restored from {path}")
+
+    def encode(self, x):
+        h = self.encoder(x)
+        moments = self.quant_conv(h)
+        posterior = DiagonalGaussianDistribution(moments)
+        return posterior
+
+    def decode(self, z):
+        z = self.post_quant_conv(z)
+        dec = self.decoder(z)
+        return dec
+
+    def forward(self, input, sample_posterior=True):
+        if input.ndim==4:
+            input = input.unsqueeze(2)
+        posterior = self.encode(input)
+        if sample_posterior:
+            z = posterior.sample()
+        else:
+            z = posterior.mode()
+        dec = self.decode(z)
+        return dec, posterior
+
+    def get_input(self, batch, k):
+        x = batch[k]
+        if x.ndim==5:
+            x = x.permute(0, 4, 1, 2, 3).to(memory_format=torch.contiguous_format).float()
+            return x
+        if len(x.shape) == 3:
+            x = x[..., None]
+        x = x.permute(0, 3, 1, 2).to(memory_format=torch.contiguous_format).float()
+        return x
+
+    def training_step(self, batch, batch_idx, optimizer_idx):
+        # tic = time.time()
+        inputs = self.get_input(batch, self.image_key)
+        # print(f"get_input time {time.time() - tic}")
+        # tic = time.time()
+        reconstructions, posterior = self(inputs)
+        # print(f"model forward time {time.time() - tic}")
+
+        if optimizer_idx == 0:
+            # train encoder+decoder+logvar
+            aeloss, log_dict_ae = self.loss(inputs, reconstructions, posterior, optimizer_idx, self.global_step,
+                                            last_layer=self.get_last_layer(), split="train")
+            self.log("aeloss", aeloss, prog_bar=True, logger=True, on_step=True, on_epoch=True)
+            self.log_dict(log_dict_ae, prog_bar=False, logger=True, on_step=True, on_epoch=False)
+            # print(f"cal loss time {time.time() - tic}")
+            return aeloss
+
+        if optimizer_idx == 1:
+            # train the discriminator
+            discloss, log_dict_disc = self.loss(inputs, reconstructions, posterior, optimizer_idx, self.global_step,
+                                                last_layer=self.get_last_layer(), split="train")
+
+            self.log("discloss", discloss, prog_bar=True, logger=True, on_step=True, on_epoch=True)
+            self.log_dict(log_dict_disc, prog_bar=False, logger=True, on_step=True, on_epoch=False)
+            # print(f"cal loss time {time.time() - tic}")
+            return discloss
+
+    def validation_step(self, batch, batch_idx):
+        with torch.no_grad():
+            inputs = self.get_input(batch, self.image_key)
+            reconstructions, posterior = self(inputs)
+            aeloss, log_dict_ae = self.loss(inputs, reconstructions, posterior, 0, self.global_step,
+                                            last_layer=self.get_last_layer(), split="val")
+
+            discloss, log_dict_disc = self.loss(inputs, reconstructions, posterior, 1, self.global_step,
+                                                last_layer=self.get_last_layer(), split="val")
+
+        self.log("val/rec_loss", log_dict_ae["val/rec_loss"])
+        self.log_dict(log_dict_ae)
+        self.log_dict(log_dict_disc)
+        return self.log_dict
+
+    def configure_optimizers(self):
+        lr = self.learning_rate
+        opt_ae = torch.optim.Adam(list(self.encoder.parameters())+
+                                  list(self.decoder.parameters())+
+                                  list(self.quant_conv.parameters())+
+                                  list(self.post_quant_conv.parameters()),
+                                  lr=lr, betas=(0.5, 0.9))
+        opt_disc = torch.optim.Adam(self.loss.discriminator.parameters(),
+                                    lr=lr, betas=(0.5, 0.9))
+        return [opt_ae, opt_disc], []
+
+    def get_last_layer(self):
+        return self.decoder.conv_out.weight
+
+    @torch.no_grad()
+    def log_images(self, batch, only_inputs=False, **kwargs):
+        log = dict()
+        x = self.get_input(batch, self.image_key)
+        x = x.to(self.device)
+        if not only_inputs:
+            xrec, posterior = self(x)
+            if x.shape[1] > 3:
+                # colorize with random projection
+                assert xrec.shape[1] > 3
+                x = self.to_rgb(x)
+                xrec = self.to_rgb(xrec)
+            log["samples"] = self.decode(torch.randn_like(posterior.sample()))
+            log["reconstructions"] = xrec
+        log["inputs"] = x
+        return log
+
+    def to_rgb(self, x):
+        assert self.image_key == "segmentation"
+        if not hasattr(self, "colorize"):
+            self.register_buffer("colorize", torch.randn(3, x.shape[1], 1, 1).to(x))
+        x = F.conv2d(x, weight=self.colorize)
+        x = 2.*(x-x.min())/(x.max()-x.min()) - 1.
+        return x
+
+class AutoencoderKL(pl.LightningModule):
+    def __init__(self,
+                 ddconfig,
+                 lossconfig,
+                 embed_dim,
+                 ckpt_path=None,
+                 ignore_keys=[],
+                 image_key="image",
+                 colorize_nlabels=None,
+                 monitor=None,
+                 ):
+        super().__init__()
+        self.image_key = image_key
+        self.encoder = Encoder(**ddconfig)
+        self.decoder = Decoder(**ddconfig)
+        self.loss = instantiate_from_config(lossconfig)
+        assert ddconfig["double_z"]
+        self.quant_conv = torch.nn.Conv2d(2*ddconfig["z_channels"], 2*embed_dim, 1)
+        self.post_quant_conv = torch.nn.Conv2d(embed_dim, ddconfig["z_channels"], 1)
+        self.embed_dim = embed_dim
+        if colorize_nlabels is not None:
+            assert type(colorize_nlabels)==int
+            self.register_buffer("colorize", torch.randn(3, colorize_nlabels, 1, 1))
+        if monitor is not None:
+            self.monitor = monitor
+        if ckpt_path is not None:
+            self.init_from_ckpt(ckpt_path, ignore_keys=ignore_keys)
+
+    def init_from_ckpt(self, path, ignore_keys=list()):
+        sd = torch.load(path, map_location="cpu")["state_dict"]
+        keys = list(sd.keys())
+        for k in keys:
+            for ik in ignore_keys:
+                if k.startswith(ik):
+                    print("Deleting key {} from state_dict.".format(k))
+                    del sd[k]
+        self.load_state_dict(sd, strict=False)
+        print(f"Restored from {path}")
+
+    def encode(self, x):
+        h = self.encoder(x)
+        moments = self.quant_conv(h)
+        posterior = DiagonalGaussianDistribution(moments)
+        return posterior
+
+    def decode(self, z):
+        z = self.post_quant_conv(z)
+        dec = self.decoder(z)
+        return dec
+
+    def forward(self, input, sample_posterior=True):
+        posterior = self.encode(input)
+        if sample_posterior:
+            z = posterior.sample()
+        else:
+            z = posterior.mode()
+        dec = self.decode(z)
+        return dec, posterior
+
+    def get_input(self, batch, k):
+        x = batch[k]
+        if len(x.shape) == 3:
+            x = x[..., None]
+        x = x.permute(0, 3, 1, 2).to(memory_format=torch.contiguous_format).float()
+        return x
+
+    def training_step(self, batch, batch_idx, optimizer_idx):
+        # tic = time.time()
+        inputs = self.get_input(batch, self.image_key)
+        # print(f"get_input time {time.time() - tic}")
+        # tic = time.time()
+        reconstructions, posterior = self(inputs)
+        # print(f"model forward time {time.time() - tic}")
+        tic = time.time()
+
+        if optimizer_idx == 0:
+            # train encoder+decoder+logvar
+            aeloss, log_dict_ae = self.loss(inputs, reconstructions, posterior, optimizer_idx, self.global_step,
+                                            last_layer=self.get_last_layer(), split="train")
+            self.log("aeloss", aeloss, prog_bar=True, logger=True, on_step=True, on_epoch=True)
+            self.log_dict(log_dict_ae, prog_bar=False, logger=True, on_step=True, on_epoch=False)
+            # print(f"cal loss time {time.time() - tic}")
+            return aeloss
+
+        if optimizer_idx == 1:
+            # train the discriminator
+            discloss, log_dict_disc = self.loss(inputs, reconstructions, posterior, optimizer_idx, self.global_step,
+                                                last_layer=self.get_last_layer(), split="train")
+
+            self.log("discloss", discloss, prog_bar=True, logger=True, on_step=True, on_epoch=True)
+            self.log_dict(log_dict_disc, prog_bar=False, logger=True, on_step=True, on_epoch=False)
+            # print(f"cal loss time {time.time() - tic}")
+            return discloss
+
+    def validation_step(self, batch, batch_idx):
+        tic = time.time()
+        inputs = self.get_input(batch, self.image_key)
+        print(f"get_input time {time.time() - tic}")
+        tic = time.time()
+        reconstructions, posterior = self(inputs)
+        print(f"val forward time {time.time() - tic}")
+        tic = time.time()
+        aeloss, log_dict_ae = self.loss(inputs, reconstructions, posterior, 0, self.global_step,
+                                        last_layer=self.get_last_layer(), split="val")
+
+        discloss, log_dict_disc = self.loss(inputs, reconstructions, posterior, 1, self.global_step,
+                                            last_layer=self.get_last_layer(), split="val")
+
+        self.log("val/rec_loss", log_dict_ae["val/rec_loss"])
+        self.log_dict(log_dict_ae)
+        self.log_dict(log_dict_disc)
+        print(f"val end time {time.time() - tic}")
+        return self.log_dict
+
+    def configure_optimizers(self):
+        lr = self.learning_rate
+        opt_ae = torch.optim.Adam(list(self.encoder.parameters())+
+                                  list(self.decoder.parameters())+
+                                  list(self.quant_conv.parameters())+
+                                  list(self.post_quant_conv.parameters()),
+                                  lr=lr, betas=(0.5, 0.9))
+        opt_disc = torch.optim.Adam(self.loss.discriminator.parameters(),
+                                    lr=lr, betas=(0.5, 0.9))
+        return [opt_ae, opt_disc], []
+
+    def get_last_layer(self):
+        return self.decoder.conv_out.weight
+
+    @torch.no_grad()
+    def log_images(self, batch, only_inputs=False, **kwargs):
+        log = dict()
+        x = self.get_input(batch, self.image_key)
+        x = x.to(self.device)
+        if not only_inputs:
+            xrec, posterior = self(x)
+            if x.shape[1] > 3:
+                # colorize with random projection
+                assert xrec.shape[1] > 3
+                x = self.to_rgb(x)
+                xrec = self.to_rgb(xrec)
+            log["samples"] = self.decode(torch.randn_like(posterior.sample()))
+            log["reconstructions"] = xrec
+        log["inputs"] = x
+        return log
+
+    def to_rgb(self, x):
+        assert self.image_key == "segmentation"
+        if not hasattr(self, "colorize"):
+            self.register_buffer("colorize", torch.randn(3, x.shape[1], 1, 1).to(x))
+        x = F.conv2d(x, weight=self.colorize)
+        x = 2.*(x-x.min())/(x.max()-x.min()) - 1.
+        return x
+
+
+class IdentityFirstStage(torch.nn.Module):
+    def __init__(self, *args, vq_interface=False, **kwargs):
+        self.vq_interface = vq_interface  # TODO: Should be true by default but check to not break older stuff
+        super().__init__()
+
+    def encode(self, x, *args, **kwargs):
+        return x
+
+    def decode(self, x, *args, **kwargs):
+        return x
+
+    def quantize(self, x, *args, **kwargs):
+        if self.vq_interface:
+            return x, None, [None, None, None]
+        return x
+
+    def forward(self, x, *args, **kwargs):
+        return x