mirror of https://github.com/vladmandic/automatic
265 lines
11 KiB
Python
265 lines
11 KiB
Python
# Copyright 2025 The RES4LYF Team (Clybius) and The HuggingFace Team. All rights reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
from typing import ClassVar, Literal
|
|
|
|
import numpy as np
|
|
import torch
|
|
from diffusers.configuration_utils import ConfigMixin, register_to_config
|
|
from diffusers.schedulers.scheduling_utils import KarrasDiffusionSchedulers, SchedulerMixin, SchedulerOutput
|
|
|
|
from diffusers.utils import logging
|
|
|
|
logger = logging.get_logger(__name__)
|
|
|
|
|
|
class RiemannianFlowScheduler(SchedulerMixin, ConfigMixin):
|
|
"""
|
|
Riemannian Flow scheduler using Exponential Integrator step.
|
|
"""
|
|
|
|
_compatibles: ClassVar[list[str]] = [e.name for e in KarrasDiffusionSchedulers]
|
|
order: ClassVar[int] = 1
|
|
|
|
@register_to_config
|
|
def __init__(
|
|
self,
|
|
num_train_timesteps: int = 1000,
|
|
beta_start: float = 0.00085,
|
|
beta_end: float = 0.012,
|
|
beta_schedule: str = "linear",
|
|
metric_type: Literal["euclidean", "hyperbolic", "spherical", "lorentzian"] = "hyperbolic",
|
|
curvature: float = 1.0,
|
|
prediction_type: str = "epsilon",
|
|
timestep_spacing: str = "linspace",
|
|
rescale_betas_zero_snr: bool = False,
|
|
use_karras_sigmas: bool = False,
|
|
use_exponential_sigmas: bool = False,
|
|
use_beta_sigmas: bool = False,
|
|
use_flow_sigmas: bool = False,
|
|
shift: float = 1.0,
|
|
use_dynamic_shifting: bool = False,
|
|
base_shift: float = 0.5,
|
|
max_shift: float = 1.15,
|
|
base_image_seq_len: int = 256,
|
|
max_image_seq_len: int = 4096,
|
|
):
|
|
from .scheduler_utils import betas_for_alpha_bar, rescale_zero_terminal_snr
|
|
|
|
if beta_schedule == "linear":
|
|
self.betas = torch.linspace(beta_start, beta_end, num_train_timesteps, dtype=torch.float32)
|
|
elif beta_schedule == "scaled_linear":
|
|
self.betas = torch.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=torch.float32) ** 2
|
|
elif beta_schedule == "squaredcos_cap_v2":
|
|
self.betas = betas_for_alpha_bar(num_train_timesteps)
|
|
else:
|
|
raise NotImplementedError(f"{beta_schedule} does not exist.")
|
|
|
|
if rescale_betas_zero_snr:
|
|
self.betas = rescale_zero_terminal_snr(self.betas)
|
|
|
|
self.alphas = 1.0 - self.betas
|
|
self.alphas_cumprod = torch.cumprod(self.alphas, dim=0)
|
|
|
|
# Standard deviation of the initial noise distribution
|
|
self.init_noise_sigma = 1.0
|
|
|
|
# Setable values
|
|
self.num_inference_steps = None
|
|
self.timesteps = torch.from_numpy(np.arange(0, num_train_timesteps)[::-1].copy())
|
|
self.sigmas = torch.zeros((num_train_timesteps,), dtype=torch.float32)
|
|
|
|
self._step_index = None
|
|
self._begin_index = None
|
|
|
|
@property
|
|
def step_index(self) -> int | None:
|
|
return self._step_index
|
|
|
|
@property
|
|
def begin_index(self) -> int | None:
|
|
return self._begin_index
|
|
|
|
def set_begin_index(self, begin_index: int = 0) -> None:
|
|
self._begin_index = begin_index
|
|
|
|
def set_timesteps(self, num_inference_steps: int, device: str | torch.device = None, mu: float | None = None, dtype: torch.dtype = torch.float32):
|
|
from .scheduler_utils import (
|
|
apply_shift,
|
|
get_dynamic_shift,
|
|
get_sigmas_beta,
|
|
get_sigmas_exponential,
|
|
get_sigmas_flow,
|
|
get_sigmas_karras,
|
|
)
|
|
|
|
self.num_inference_steps = num_inference_steps
|
|
timestep_spacing = getattr(self.config, "timestep_spacing", "linspace")
|
|
steps_offset = getattr(self.config, "steps_offset", 0)
|
|
|
|
if timestep_spacing == "linspace":
|
|
timesteps = np.linspace(0, self.config.num_train_timesteps - 1, num_inference_steps, dtype=float)[::-1].copy()
|
|
elif timestep_spacing == "leading":
|
|
step_ratio = self.config.num_train_timesteps // self.num_inference_steps
|
|
timesteps = (np.arange(0, num_inference_steps) * step_ratio).round()[::-1].copy()
|
|
timesteps += steps_offset
|
|
elif timestep_spacing == "trailing":
|
|
step_ratio = self.config.num_train_timesteps / self.num_inference_steps
|
|
timesteps = (np.arange(self.config.num_train_timesteps, 0, -step_ratio)).round().copy()
|
|
timesteps -= 1
|
|
else:
|
|
raise ValueError(f"timestep_spacing {timestep_spacing} is not supported.")
|
|
|
|
# Derived sigma range from alphas_cumprod
|
|
# In FM, we usually go from sigma_max to sigma_min
|
|
base_sigmas = np.array(((1 - self.alphas_cumprod) / self.alphas_cumprod) ** 0.5)
|
|
# Note: alphas_cumprod[0] is ~0.999 (small sigma), alphas_cumprod[-1] is ~0.0001 (large sigma)
|
|
start_sigma = base_sigmas[-1]
|
|
end_sigma = base_sigmas[0]
|
|
|
|
t = torch.linspace(0, 1, num_inference_steps, device=device)
|
|
metric_type = self.config.metric_type
|
|
curvature = self.config.curvature
|
|
|
|
if metric_type == "euclidean":
|
|
result = start_sigma * (1 - t) + end_sigma * t
|
|
elif metric_type == "hyperbolic":
|
|
x_start = torch.tanh(torch.tensor(start_sigma / 2, device=device))
|
|
x_end = torch.tanh(torch.tensor(end_sigma / 2, device=device))
|
|
d = torch.acosh(torch.clamp(1 + 2 * ((x_start - x_end)**2) / ((1 - x_start**2) * (1 - x_end**2) + 1e-9), min=1.0))
|
|
lambda_t = torch.sinh(t * d) / (torch.sinh(d) + 1e-9)
|
|
result = 2 * torch.atanh(torch.clamp((1 - lambda_t) * x_start + lambda_t * x_end, -0.999, 0.999))
|
|
elif metric_type == "spherical":
|
|
k = torch.tensor(curvature, device=device)
|
|
theta_start = start_sigma * torch.sqrt(k)
|
|
theta_end = end_sigma * torch.sqrt(k)
|
|
result = torch.sin((1 - t) * theta_start + t * theta_end) / torch.sqrt(k)
|
|
elif metric_type == "lorentzian":
|
|
gamma = 1 / torch.sqrt(torch.clamp(1 - curvature * t**2, min=1e-9))
|
|
result = (start_sigma * (1 - t) + end_sigma * t) * gamma
|
|
else:
|
|
result = start_sigma * (1 - t) + end_sigma * t
|
|
|
|
result = torch.clamp(result, min=min(start_sigma, end_sigma), max=max(start_sigma, end_sigma))
|
|
|
|
if start_sigma > end_sigma:
|
|
result, _ = torch.sort(result, descending=True)
|
|
|
|
sigmas = result.cpu().numpy()
|
|
|
|
if getattr(self.config, "use_karras_sigmas", False):
|
|
sigmas = get_sigmas_karras(num_inference_steps, sigmas[-1], sigmas[0], device=device, dtype=dtype).cpu().numpy()
|
|
elif getattr(self.config, "use_exponential_sigmas", False):
|
|
sigmas = get_sigmas_exponential(num_inference_steps, sigmas[-1], sigmas[0], device=device, dtype=dtype).cpu().numpy()
|
|
elif getattr(self.config, "use_beta_sigmas", False):
|
|
sigmas = get_sigmas_beta(num_inference_steps, sigmas[-1], sigmas[0], device=device, dtype=dtype).cpu().numpy()
|
|
elif getattr(self.config, "use_flow_sigmas", False):
|
|
sigmas = get_sigmas_flow(num_inference_steps, sigmas[-1], sigmas[0], device=device, dtype=dtype).cpu().numpy()
|
|
|
|
shift = getattr(self.config, "shift", 1.0)
|
|
use_dynamic_shifting = getattr(self.config, "use_dynamic_shifting", False)
|
|
if shift != 1.0 or use_dynamic_shifting:
|
|
if use_dynamic_shifting and mu is not None:
|
|
shift = get_dynamic_shift(
|
|
mu,
|
|
getattr(self.config, "base_shift", 0.5),
|
|
getattr(self.config, "max_shift", 1.5),
|
|
getattr(self.config, "base_image_seq_len", 256),
|
|
getattr(self.config, "max_image_seq_len", 4096),
|
|
)
|
|
sigmas = apply_shift(torch.from_numpy(sigmas), shift).numpy()
|
|
|
|
self.sigmas = torch.from_numpy(np.concatenate([sigmas, [0.0]])).to(device=device, dtype=dtype)
|
|
self.timesteps = torch.from_numpy(timesteps).to(device=device, dtype=dtype)
|
|
self.init_noise_sigma = self.sigmas.max().item() if self.sigmas.numel() > 0 else 1.0
|
|
|
|
self._step_index = None
|
|
self._begin_index = None
|
|
|
|
def index_for_timestep(self, timestep, schedule_timesteps=None):
|
|
from .scheduler_utils import index_for_timestep
|
|
if schedule_timesteps is None:
|
|
schedule_timesteps = self.timesteps
|
|
return index_for_timestep(timestep, schedule_timesteps)
|
|
|
|
def add_noise(
|
|
self,
|
|
original_samples: torch.Tensor,
|
|
noise: torch.Tensor,
|
|
timesteps: torch.Tensor,
|
|
) -> torch.Tensor:
|
|
from .scheduler_utils import add_noise_to_sample
|
|
return add_noise_to_sample(original_samples, noise, self.sigmas, timesteps, self.timesteps)
|
|
|
|
def scale_model_input(self, sample: torch.Tensor, timestep: float | torch.Tensor) -> torch.Tensor:
|
|
if self._step_index is None:
|
|
self._init_step_index(timestep)
|
|
if self.config.prediction_type == "flow_prediction":
|
|
return sample
|
|
sigma = self.sigmas[self._step_index]
|
|
sample = sample / ((sigma**2 + 1) ** 0.5)
|
|
return sample
|
|
|
|
def step(
|
|
self,
|
|
model_output: torch.Tensor,
|
|
timestep: float | torch.Tensor,
|
|
sample: torch.Tensor,
|
|
return_dict: bool = True,
|
|
) -> SchedulerOutput | tuple:
|
|
if self._step_index is None:
|
|
self._init_step_index(timestep)
|
|
|
|
step_index = self._step_index
|
|
sigma = self.sigmas[step_index]
|
|
sigma_next = self.sigmas[step_index + 1]
|
|
|
|
# Determine denoised (x_0 prediction)
|
|
if self.config.prediction_type == "epsilon":
|
|
x0 = sample - sigma * model_output
|
|
elif self.config.prediction_type == "v_prediction":
|
|
alpha_t = 1.0 / (sigma**2 + 1)**0.5
|
|
sigma_t = sigma * alpha_t
|
|
x0 = alpha_t * sample - sigma_t * model_output
|
|
elif self.config.prediction_type == "sample":
|
|
x0 = model_output
|
|
elif self.config.prediction_type == "flow_prediction":
|
|
x0 = sample - sigma * model_output
|
|
else:
|
|
x0 = model_output
|
|
|
|
# Exponential Integrator Update (1st order)
|
|
if sigma_next == 0:
|
|
x_next = x0
|
|
else:
|
|
h = -torch.log(sigma_next / sigma) if sigma > 0 and sigma_next > 0 else torch.zeros_like(sigma)
|
|
x_next = torch.exp(-h) * sample + (1 - torch.exp(-h)) * x0
|
|
|
|
self._step_index += 1
|
|
|
|
if not return_dict:
|
|
return (x_next,)
|
|
return SchedulerOutput(prev_sample=x_next)
|
|
|
|
def _init_step_index(self, timestep):
|
|
if self.begin_index is None:
|
|
if isinstance(timestep, torch.Tensor):
|
|
timestep = timestep.to(self.timesteps.device)
|
|
self._step_index = self.index_for_timestep(timestep)
|
|
else:
|
|
self._step_index = self._begin_index
|
|
|
|
def __len__(self):
|
|
return self.config.num_train_timesteps
|