""" This code started out as a PyTorch port of Ho et al's diffusion models: https://github.com/hojonathanho/diffusion/blob/1e0dceb3b3495bbe19116a5e1b3596cd0706c543/diffusion_tf/diffusion_utils_2.py Docstrings have been added, as well as DDIM sampling and a new collection of beta schedules. """ from model.unet_autoenc import AutoencReturn from config_base import BaseConfig import enum import math import numpy as np import torch as th from model import * from model.nn import mean_flat from typing import NamedTuple, Tuple from choices import * from torch.cuda.amp import autocast import torch.nn.functional as F from dataclasses import dataclass @dataclass class GaussianDiffusionBeatGansConfig(BaseConfig): gen_type: GenerativeType betas: Tuple[float] model_type: ModelType model_mean_type: ModelMeanType model_var_type: ModelVarType loss_type: LossType rescale_timesteps: bool fp16: bool train_pred_xstart_detach: bool = True def make_sampler(self): return GaussianDiffusionBeatGans(self) class GaussianDiffusionBeatGans: """ Utilities for training and sampling diffusion models. Ported directly from here, and then adapted over time to further experimentation. 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. :param model_mean_type: a ModelMeanType determining what the model outputs. :param model_var_type: a ModelVarType determining how variance is output. :param loss_type: a LossType determining the loss function to use. :param rescale_timesteps: if True, pass floating point timesteps into the model so that they are always scaled like in the original paper (0 to 1000). """ def __init__(self, conf: GaussianDiffusionBeatGansConfig): self.conf = conf self.model_mean_type = conf.model_mean_type self.model_var_type = conf.model_var_type self.loss_type = conf.loss_type self.rescale_timesteps = conf.rescale_timesteps # Use float64 for accuracy. betas = np.array(conf.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)) # 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:])) 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 training_losses(self, model, motion_direction_start: th.Tensor, motion_target: th.Tensor, motion_start: th.Tensor, audio_feats: th.Tensor, face_location: th.Tensor, face_scale: th.Tensor, yaw_pitch_roll: th.Tensor, t: th.Tensor, model_kwargs=None, noise: th.Tensor = None): """ 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. """ if model_kwargs is None: model_kwargs = {} if noise is None: noise = th.randn_like(motion_target) x_t = self.q_sample(motion_target, t, noise=noise) terms = {'x_t': x_t} if self.loss_type in [ LossType.mse, LossType.l1, ]: with autocast(self.conf.fp16): # x_t is static wrt. to the diffusion process predicted_direction, predicted_location, predicted_scale, predicted_pose = model.forward(motion_start, motion_direction_start, audio_feats, face_location, face_scale, yaw_pitch_roll, x_t.detach(), self._scale_timesteps(t), control_flag=False) target_types = { ModelMeanType.eps: noise, } target = target_types[self.model_mean_type] assert predicted_direction.shape == target.shape == motion_target.shape if self.loss_type == LossType.mse: if self.model_mean_type == ModelMeanType.eps: direction_loss = mean_flat((target - predicted_direction)**2) # import pdb;pdb.set_trace() location_loss = mean_flat((face_location.unsqueeze(-1) - predicted_location)**2) scale_loss = mean_flat((face_scale - predicted_scale)**2) pose_loss = mean_flat((yaw_pitch_roll - predicted_pose)**2) terms["mse"] = direction_loss + location_loss + scale_loss + pose_loss else: raise NotImplementedError() elif self.loss_type == LossType.l1: # (n, c, h, w) => (n, ) terms["mse"] = mean_flat((target - predicted_direction).abs()) else: raise NotImplementedError() if "vb" in terms: # if learning the variance also use the vlb loss terms["loss"] = terms["mse"] + terms["vb"] else: terms["loss"] = terms["mse"] else: raise NotImplementedError(self.loss_type) return terms def sample(self, model: Model, shape=None, noise=None, cond=None, x_start=None, clip_denoised=True, model_kwargs=None, progress=False): """ Args: x_start: given for the autoencoder """ if model_kwargs is None: model_kwargs = {} if self.conf.model_type.has_autoenc(): model_kwargs['x_start'] = x_start model_kwargs['cond'] = cond if self.conf.gen_type == GenerativeType.ddpm: return self.p_sample_loop(model, shape=shape, noise=noise, clip_denoised=clip_denoised, model_kwargs=model_kwargs, progress=progress) elif self.conf.gen_type == GenerativeType.ddim: return self.ddim_sample_loop(model, shape=shape, noise=noise, clip_denoised=clip_denoised, model_kwargs=model_kwargs, progress=progress) else: raise NotImplementedError() 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 = {} motion_start = model_kwargs['start'] audio_feats = model_kwargs['audio_driven'] face_location = model_kwargs['face_location'] face_scale = model_kwargs['face_scale'] yaw_pitch_roll = model_kwargs['yaw_pitch_roll'] motion_direction_start = model_kwargs['motion_direction_start'] control_flag = model_kwargs['control_flag'] B, C = x.shape[:2] assert t.shape == (B, ) with autocast(self.conf.fp16): model_forward, _, _, _ = model.forward(motion_start, motion_direction_start, audio_feats, face_location, face_scale, yaw_pitch_roll, x, self._scale_timesteps(t), control_flag) model_output = model_forward if 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) def process_xstart(x): if denoised_fn is not None: x = denoised_fn(x) if clip_denoised: return x.clamp(-1, 1) return x if self.model_mean_type in [ ModelMeanType.eps, ]: if self.model_mean_type == ModelMeanType.eps: pred_xstart = process_xstart( self._predict_xstart_from_eps(x_t=x, t=t, eps=model_output)) else: raise NotImplementedError() model_mean, _, _ = self.q_posterior_mean_variance( x_start=pred_xstart, x_t=x, t=t) else: raise NotImplementedError(self.model_mean_type) 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, 'model_forward': model_forward, } 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_xstart_from_xprev(self, x_t, t, xprev): assert x_t.shape == xprev.shape return ( # (xprev - coef2*x_t) / coef1 _extract_into_tensor(1.0 / self.posterior_mean_coef1, t, x_t.shape) * xprev - _extract_into_tensor( self.posterior_mean_coef2 / self.posterior_mean_coef1, t, x_t.shape) * x_t) def _predict_xstart_from_scaled_xstart(self, t, scaled_xstart): return scaled_xstart * _extract_into_tensor( self.sqrt_recip_alphas_cumprod, t, scaled_xstart.shape) 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 _predict_eps_from_scaled_xstart(self, x_t, t, scaled_xstart): """ Args: scaled_xstart: is supposed to be sqrt(alphacum) * x_0 """ # 1 / sqrt(1-alphabar) * (x_t - scaled xstart) return (x_t - scaled_xstart) / _extract_into_tensor( self.sqrt_one_minus_alphas_cumprod, t, x_t.shape) def _scale_timesteps(self, t): if self.rescale_timesteps: # scale t to be maxed out at 1000 steps return t.float() * (1000.0 / self.num_timesteps) return t 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, self._scale_timesteps(t), **model_kwargs) new_mean = (p_mean_var["mean"].float() + p_mean_var["variance"] * gradient.float()) return new_mean 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, self._scale_timesteps(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: 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: Model, shape=None, 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: Model, shape=None, 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 if noise is not None: img = noise else: assert isinstance(shape, (tuple, list)) img = 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) t = th.tensor([i] * len(img), 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: 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: Model, x, t, clip_denoised=True, denoised_fn=None, model_kwargs=None, eta=0.0, ): """ Sample x_{t+1} from the model using DDIM reverse ODE. NOTE: never used ? """ 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, ) # 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 (DDIM paper) (th.sqrt == torch.sqrt) 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_reverse_sample_loop( self, model: Model, x, clip_denoised=True, denoised_fn=None, model_kwargs=None, eta=0.0, device=None, ): if device is None: device = next(model.parameters()).device sample_t = [] xstart_t = [] T = [] indices = list(range(self.num_timesteps)) sample = x for i in indices: t = th.tensor([i] * len(sample), device=device) with th.no_grad(): out = self.ddim_reverse_sample(model, sample, t=t, clip_denoised=clip_denoised, denoised_fn=denoised_fn, model_kwargs=model_kwargs, eta=eta) sample = out['sample'] # [1, ..., T] sample_t.append(sample) # [0, ...., T-1] xstart_t.append(out['pred_xstart']) # [0, ..., T-1] ready to use T.append(t) return { # xT " 'sample': sample, # (1, ..., T) 'sample_t': sample_t, # xstart here is a bit different from sampling from T = T-1 to T = 0 # may not be exact 'xstart_t': xstart_t, 'T': T, } def ddim_sample_loop( self, model: Model, shape=None, 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: Model, shape=None, 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 if noise is not None: img = noise else: assert isinstance(shape, (tuple, list)) img = 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: if isinstance(model_kwargs, list): # index dependent model kwargs # (T-1, ..., 0) _kwargs = model_kwargs[i] else: _kwargs = model_kwargs t = th.tensor([i] * len(img), 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=_kwargs, eta=eta, ) out['t'] = t yield out img = out["sample"] def _vb_terms_bpd(self, model: 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"], 'model_forward': out['model_forward'], } 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: 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.expand(broadcast_shape) 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 beta_start = scale * 0.0001 beta_end = scale * 0.02 return np.linspace(beta_start, beta_end, num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "cosine": return betas_for_alpha_bar( num_diffusion_timesteps, lambda t: math.cos((t + 0.008) / 1.008 * math.pi / 2)**2, ) elif schedule_name == "const0.01": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.01] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.015": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.015] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.008": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.008] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.0065": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.0065] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.0055": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.0055] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.0045": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.0045] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.0035": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.0035] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.0025": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.0025] * num_diffusion_timesteps, dtype=np.float64) elif schedule_name == "const0.0015": scale = 1000 / num_diffusion_timesteps return np.array([scale * 0.0015] * num_diffusion_timesteps, dtype=np.float64) 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) 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 = None for obj in (mean1, logvar1, mean2, logvar2): if isinstance(obj, th.Tensor): tensor = obj break 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).to(tensor) 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 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 class DummyModel(th.nn.Module): def __init__(self, pred): super().__init__() self.pred = pred def forward(self, *args, **kwargs): return DummyReturn(pred=self.pred) class DummyReturn(NamedTuple): pred: th.Tensor