Karras VE, DDIM and DDPM flax schedulers (#508)

* beta never changes removed from state

* fix typos in docs

* removed unused var

* initial ddim flax scheduler

* import

* added dummy objects

* fix style

* fix typo

* docs

* fix typo in comment

* set return type

* added flax ddom

* fix style

* remake

* pass PRNG key as argument and split before use

* fix doc string

* use config

* added flax Karras VE scheduler

* make style

* fix dummy

* fix ndarray type annotation

* replace returns a new state

* added lms_discrete scheduler

* use self.config

* add_noise needs state

* use config

* use config

* docstring

* added flax score sde ve

* fix imports

* fix typos
This commit is contained in:
Kashif Rasul 2022-09-15 15:55:48 +02:00 committed by GitHub
parent 83a7bb2aba
commit b34be039f9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
21 changed files with 1351 additions and 66 deletions

View File

@ -504,7 +504,9 @@ def main():
noise = torch.randn(latents.shape).to(latents.device)
bsz = latents.shape[0]
# Sample a random timestep for each image
timesteps = torch.randint(0, noise_scheduler.num_train_timesteps, (bsz,), device=latents.device).long()
timesteps = torch.randint(
0, noise_scheduler.config.num_train_timesteps, (bsz,), device=latents.device
).long()
# Add noise to the latents according to the noise magnitude at each timestep
# (this is the forward diffusion process)

View File

@ -130,7 +130,7 @@ def main(args):
bsz = clean_images.shape[0]
# Sample a random timestep for each image
timesteps = torch.randint(
0, noise_scheduler.num_train_timesteps, (bsz,), device=clean_images.device
0, noise_scheduler.config.num_train_timesteps, (bsz,), device=clean_images.device
).long()
# Add noise to the clean images according to the noise magnitude at each timestep

View File

@ -64,6 +64,13 @@ else:
if is_flax_available():
from .modeling_flax_utils import FlaxModelMixin
from .schedulers import FlaxPNDMScheduler
from .schedulers import (
FlaxDDIMScheduler,
FlaxDDPMScheduler,
FlaxKarrasVeScheduler,
FlaxLMSDiscreteScheduler,
FlaxPNDMScheduler,
FlaxScoreSdeVeScheduler,
)
else:
from .utils.dummy_flax_objects import * # noqa F403

View File

@ -386,7 +386,7 @@ class FlaxModelMixin:
raise ValueError from e
except (UnicodeDecodeError, ValueError):
raise EnvironmentError(f"Unable to convert {model_file} to Flax deserializable object. ")
# make sure all arrays are stored as jnp.arrays
# make sure all arrays are stored as jnp.ndarray
# NOTE: This is to prevent a bug this will be fixed in Flax >= v0.3.4:
# https://github.com/google/flax/issues/1261
state = jax.tree_util.tree_map(lambda x: jax.device_put(x, jax.devices("cpu")[0]), state)

View File

@ -80,7 +80,7 @@ class ScoreSdeVePipeline(DiffusionPipeline):
sigma_t = self.scheduler.sigmas[i] * torch.ones(shape[0], device=self.device)
# correction step
for _ in range(self.scheduler.correct_steps):
for _ in range(self.scheduler.config.correct_steps):
model_output = self.unet(sample, sigma_t).sample
sample = self.scheduler.step_correct(model_output, sample, generator=generator).prev_sample

View File

@ -28,7 +28,12 @@ else:
from ..utils.dummy_pt_objects import * # noqa F403
if is_flax_available():
from .scheduling_ddim_flax import FlaxDDIMScheduler
from .scheduling_ddpm_flax import FlaxDDPMScheduler
from .scheduling_karras_ve_flax import FlaxKarrasVeScheduler
from .scheduling_lms_discrete_flax import FlaxLMSDiscreteScheduler
from .scheduling_pndm_flax import FlaxPNDMScheduler
from .scheduling_sde_ve_flax import FlaxScoreSdeVeScheduler
else:
from ..utils.dummy_flax_objects import * # noqa F403

View File

@ -113,7 +113,7 @@ class DDIMScheduler(SchedulerMixin, ConfigMixin):
# At every step in ddim, we are looking into the previous alphas_cumprod
# For the final step, there is no previous alphas_cumprod because we are already at 0
# `set_alpha_to_one` decides whether we set this paratemer simply to one or
# `set_alpha_to_one` decides whether we set this parameter simply to one or
# whether we use the final alpha of the "non-previous" one.
self.final_alpha_cumprod = np.array(1.0) if set_alpha_to_one else self.alphas_cumprod[0]
@ -195,7 +195,7 @@ class DDIMScheduler(SchedulerMixin, ConfigMixin):
# - pred_original_sample -> f_theta(x_t, t) or x_0
# - std_dev_t -> sigma_t
# - eta -> η
# - pred_sample_direction -> "direction pointingc to x_t"
# - pred_sample_direction -> "direction pointing to x_t"
# - pred_prev_sample -> "x_t-1"
# 1. get previous step value (=t-1)

View File

@ -0,0 +1,274 @@
# Copyright 2022 Stanford University Team 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.
# DISCLAIMER: This code is strongly influenced by https://github.com/pesser/pytorch_diffusion
# and https://github.com/hojonathanho/diffusion
import math
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import flax
import jax.numpy as jnp
from jax import random
from ..configuration_utils import ConfigMixin, register_to_config
from .scheduling_utils import SchedulerMixin, SchedulerOutput
def betas_for_alpha_bar(num_diffusion_timesteps, max_beta=0.999) -> jnp.ndarray:
"""
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].
Contains a function alpha_bar that takes an argument t and transforms it to the cumulative product of (1-beta) up
to that part of the diffusion process.
Args:
num_diffusion_timesteps (`int`): the number of betas to produce.
max_beta (`float`): the maximum beta to use; use values lower than 1 to
prevent singularities.
Returns:
betas (`jnp.ndarray`): the betas used by the scheduler to step the model outputs
"""
def alpha_bar(time_step):
return math.cos((time_step + 0.008) / 1.008 * math.pi / 2) ** 2
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 jnp.array(betas, dtype=jnp.float32)
@flax.struct.dataclass
class DDIMSchedulerState:
# setable values
timesteps: jnp.ndarray
num_inference_steps: Optional[int] = None
@classmethod
def create(cls, num_train_timesteps: int):
return cls(timesteps=jnp.arange(0, num_train_timesteps)[::-1])
@dataclass
class FlaxSchedulerOutput(SchedulerOutput):
state: DDIMSchedulerState
class FlaxDDIMScheduler(SchedulerMixin, ConfigMixin):
"""
Denoising diffusion implicit models is a scheduler that extends the denoising procedure introduced in denoising
diffusion probabilistic models (DDPMs) with non-Markovian guidance.
[`~ConfigMixin`] takes care of storing all config attributes that are passed in the scheduler's `__init__`
function, such as `num_train_timesteps`. They can be accessed via `scheduler.config.num_train_timesteps`.
[`~ConfigMixin`] also provides general loading and saving functionality via the [`~ConfigMixin.save_config`] and
[`~ConfigMixin.from_config`] functions.
For more details, see the original paper: https://arxiv.org/abs/2010.02502
Args:
num_train_timesteps (`int`): number of diffusion steps used to train the model.
beta_start (`float`): the starting `beta` value of inference.
beta_end (`float`): the final `beta` value.
beta_schedule (`str`):
the beta schedule, a mapping from a beta range to a sequence of betas for stepping the model. Choose from
`linear`, `scaled_linear`, or `squaredcos_cap_v2`.
trained_betas (`jnp.ndarray`, optional):
option to pass an array of betas directly to the constructor to bypass `beta_start`, `beta_end` etc.
clip_sample (`bool`, default `True`):
option to clip predicted sample between -1 and 1 for numerical stability.
set_alpha_to_one (`bool`, default `True`):
if alpha for final step is 1 or the final alpha of the "non-previous" one.
"""
@register_to_config
def __init__(
self,
num_train_timesteps: int = 1000,
beta_start: float = 0.0001,
beta_end: float = 0.02,
beta_schedule: str = "linear",
trained_betas: Optional[jnp.ndarray] = None,
clip_sample: bool = True,
set_alpha_to_one: bool = True,
):
if trained_betas is not None:
self.betas = jnp.asarray(trained_betas)
if beta_schedule == "linear":
self.betas = jnp.linspace(beta_start, beta_end, num_train_timesteps, dtype=jnp.float32)
elif beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
self.betas = jnp.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=jnp.float32) ** 2
elif beta_schedule == "squaredcos_cap_v2":
# Glide cosine schedule
self.betas = betas_for_alpha_bar(num_train_timesteps)
else:
raise NotImplementedError(f"{beta_schedule} does is not implemented for {self.__class__}")
self.alphas = 1.0 - self.betas
self.alphas_cumprod = jnp.cumprod(self.alphas, axis=0)
# At every step in ddim, we are looking into the previous alphas_cumprod
# For the final step, there is no previous alphas_cumprod because we are already at 0
# `set_alpha_to_one` decides whether we set this parameter simply to one or
# whether we use the final alpha of the "non-previous" one.
self.final_alpha_cumprod = jnp.array(1.0) if set_alpha_to_one else self.alphas_cumprod[0]
self.state = DDIMSchedulerState.create(num_train_timesteps=num_train_timesteps)
def _get_variance(self, timestep, prev_timestep):
alpha_prod_t = self.alphas_cumprod[timestep]
alpha_prod_t_prev = self.alphas_cumprod[prev_timestep] if prev_timestep >= 0 else self.final_alpha_cumprod
beta_prod_t = 1 - alpha_prod_t
beta_prod_t_prev = 1 - alpha_prod_t_prev
variance = (beta_prod_t_prev / beta_prod_t) * (1 - alpha_prod_t / alpha_prod_t_prev)
return variance
def set_timesteps(
self, state: DDIMSchedulerState, num_inference_steps: int, offset: int = 0
) -> DDIMSchedulerState:
"""
Sets the discrete timesteps used for the diffusion chain. Supporting function to be run before inference.
Args:
state (`DDIMSchedulerState`):
the `FlaxDDIMScheduler` state data class instance.
num_inference_steps (`int`):
the number of diffusion steps used when generating samples with a pre-trained model.
offset (`int`):
optional value to shift timestep values up by. A value of 1 is used in stable diffusion for inference.
"""
step_ratio = self.config.num_train_timesteps // num_inference_steps
# creates integer timesteps by multiplying by ratio
# casting to int to avoid issues when num_inference_step is power of 3
timesteps = (jnp.arange(0, num_inference_steps) * step_ratio).round()[::-1]
timesteps = timesteps + offset
return state.replace(num_inference_steps=num_inference_steps, timesteps=timesteps)
def step(
self,
state: DDIMSchedulerState,
model_output: jnp.ndarray,
timestep: int,
sample: jnp.ndarray,
key: random.KeyArray,
eta: float = 0.0,
use_clipped_model_output: bool = False,
return_dict: bool = True,
) -> Union[FlaxSchedulerOutput, Tuple]:
"""
Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion
process from the learned model outputs (most often the predicted noise).
Args:
state (`DDIMSchedulerState`): the `FlaxDDIMScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`jnp.ndarray`):
current instance of sample being created by diffusion process.
key (`random.KeyArray`): a PRNG key.
eta (`float`): weight of noise for added noise in diffusion step.
use_clipped_model_output (`bool`): TODO
return_dict (`bool`): option for returning tuple rather than SchedulerOutput class
Returns:
[`FlaxSchedulerOutput`] or `tuple`: [`FlaxSchedulerOutput`] if `return_dict` is True, otherwise a `tuple`.
When returning a tuple, the first element is the sample tensor.
"""
if state.num_inference_steps is None:
raise ValueError(
"Number of inference steps is 'None', you need to run 'set_timesteps' after creating the scheduler"
)
# See formulas (12) and (16) of DDIM paper https://arxiv.org/pdf/2010.02502.pdf
# Ideally, read DDIM paper in-detail understanding
# Notation (<variable name> -> <name in paper>
# - pred_noise_t -> e_theta(x_t, t)
# - pred_original_sample -> f_theta(x_t, t) or x_0
# - std_dev_t -> sigma_t
# - eta -> η
# - pred_sample_direction -> "direction pointing to x_t"
# - pred_prev_sample -> "x_t-1"
# 1. get previous step value (=t-1)
prev_timestep = timestep - self.config.num_train_timesteps // state.num_inference_steps
# 2. compute alphas, betas
alpha_prod_t = self.alphas_cumprod[timestep]
alpha_prod_t_prev = self.alphas_cumprod[prev_timestep] if prev_timestep >= 0 else self.final_alpha_cumprod
beta_prod_t = 1 - alpha_prod_t
# 3. compute predicted original sample from predicted noise also called
# "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf
pred_original_sample = (sample - beta_prod_t ** (0.5) * model_output) / alpha_prod_t ** (0.5)
# 4. Clip "predicted x_0"
if self.config.clip_sample:
pred_original_sample = jnp.clip(pred_original_sample, -1, 1)
# 5. compute variance: "sigma_t(η)" -> see formula (16)
# σ_t = sqrt((1 α_t1)/(1 α_t)) * sqrt(1 α_t/α_t1)
variance = self._get_variance(timestep, prev_timestep)
std_dev_t = eta * variance ** (0.5)
if use_clipped_model_output:
# the model_output is always re-derived from the clipped x_0 in Glide
model_output = (sample - alpha_prod_t ** (0.5) * pred_original_sample) / beta_prod_t ** (0.5)
# 6. compute "direction pointing to x_t" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf
pred_sample_direction = (1 - alpha_prod_t_prev - std_dev_t**2) ** (0.5) * model_output
# 7. compute x_t without "random noise" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf
prev_sample = alpha_prod_t_prev ** (0.5) * pred_original_sample + pred_sample_direction
if eta > 0:
key = random.split(key, num=1)
noise = random.normal(key=key, shape=model_output.shape)
variance = self._get_variance(timestep, prev_timestep) ** (0.5) * eta * noise
prev_sample = prev_sample + variance
if not return_dict:
return (prev_sample, state)
return FlaxSchedulerOutput(prev_sample=prev_sample, state=state)
def add_noise(
self,
original_samples: jnp.ndarray,
noise: jnp.ndarray,
timesteps: jnp.ndarray,
) -> jnp.ndarray:
sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5
sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples)
sqrt_one_minus_alpha_prod = (1 - self.alphas_cumprod[timesteps]) ** 0.5
sqrt_one_minus_alpha_prod = self.match_shape(sqrt_one_minus_alpha_prod, original_samples)
noisy_samples = sqrt_alpha_prod * original_samples + sqrt_one_minus_alpha_prod * noise
return noisy_samples
def __len__(self):
return self.config.num_train_timesteps

View File

@ -148,7 +148,7 @@ class DDPMScheduler(SchedulerMixin, ConfigMixin):
if variance_type is None:
variance_type = self.config.variance_type
# hacks - were probs added for training stability
# hacks - were probably added for training stability
if variance_type == "fixed_small":
variance = self.clip(variance, min_value=1e-20)
# for rl-diffuser https://arxiv.org/abs/2205.09991
@ -187,7 +187,6 @@ class DDPMScheduler(SchedulerMixin, ConfigMixin):
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`torch.FloatTensor` or `np.ndarray`):
current instance of sample being created by diffusion process.
eta (`float`): weight of noise for added noise in diffusion step.
predict_epsilon (`bool`):
optional flag to use when model predicts the samples directly instead of the noise, epsilon.
generator: random number generator.

View File

@ -0,0 +1,277 @@
# Copyright 2022 UC Berkely Team 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.
# DISCLAIMER: This file is strongly influenced by https://github.com/ermongroup/ddim
import math
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import flax
import jax.numpy as jnp
from jax import random
from ..configuration_utils import ConfigMixin, register_to_config
from .scheduling_utils import SchedulerMixin, SchedulerOutput
def betas_for_alpha_bar(num_diffusion_timesteps, max_beta=0.999) -> jnp.ndarray:
"""
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].
Contains a function alpha_bar that takes an argument t and transforms it to the cumulative product of (1-beta) up
to that part of the diffusion process.
Args:
num_diffusion_timesteps (`int`): the number of betas to produce.
max_beta (`float`): the maximum beta to use; use values lower than 1 to
prevent singularities.
Returns:
betas (`jnp.ndarray`): the betas used by the scheduler to step the model outputs
"""
def alpha_bar(time_step):
return math.cos((time_step + 0.008) / 1.008 * math.pi / 2) ** 2
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 jnp.array(betas, dtype=jnp.float32)
@flax.struct.dataclass
class DDPMSchedulerState:
# setable values
timesteps: jnp.ndarray
num_inference_steps: Optional[int] = None
@classmethod
def create(cls, num_train_timesteps: int):
return cls(timesteps=jnp.arange(0, num_train_timesteps)[::-1])
@dataclass
class FlaxSchedulerOutput(SchedulerOutput):
state: DDPMSchedulerState
class FlaxDDPMScheduler(SchedulerMixin, ConfigMixin):
"""
Denoising diffusion probabilistic models (DDPMs) explores the connections between denoising score matching and
Langevin dynamics sampling.
[`~ConfigMixin`] takes care of storing all config attributes that are passed in the scheduler's `__init__`
function, such as `num_train_timesteps`. They can be accessed via `scheduler.config.num_train_timesteps`.
[`~ConfigMixin`] also provides general loading and saving functionality via the [`~ConfigMixin.save_config`] and
[`~ConfigMixin.from_config`] functions.
For more details, see the original paper: https://arxiv.org/abs/2006.11239
Args:
num_train_timesteps (`int`): number of diffusion steps used to train the model.
beta_start (`float`): the starting `beta` value of inference.
beta_end (`float`): the final `beta` value.
beta_schedule (`str`):
the beta schedule, a mapping from a beta range to a sequence of betas for stepping the model. Choose from
`linear`, `scaled_linear`, or `squaredcos_cap_v2`.
trained_betas (`np.ndarray`, optional):
option to pass an array of betas directly to the constructor to bypass `beta_start`, `beta_end` etc.
variance_type (`str`):
options to clip the variance used when adding noise to the denoised sample. Choose from `fixed_small`,
`fixed_small_log`, `fixed_large`, `fixed_large_log`, `learned` or `learned_range`.
clip_sample (`bool`, default `True`):
option to clip predicted sample between -1 and 1 for numerical stability.
tensor_format (`str`): whether the scheduler expects pytorch or numpy arrays.
"""
@register_to_config
def __init__(
self,
num_train_timesteps: int = 1000,
beta_start: float = 0.0001,
beta_end: float = 0.02,
beta_schedule: str = "linear",
trained_betas: Optional[jnp.ndarray] = None,
variance_type: str = "fixed_small",
clip_sample: bool = True,
):
if trained_betas is not None:
self.betas = jnp.asarray(trained_betas)
elif beta_schedule == "linear":
self.betas = jnp.linspace(beta_start, beta_end, num_train_timesteps, dtype=jnp.float32)
elif beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
self.betas = jnp.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=jnp.float32) ** 2
elif beta_schedule == "squaredcos_cap_v2":
# Glide cosine schedule
self.betas = betas_for_alpha_bar(num_train_timesteps)
else:
raise NotImplementedError(f"{beta_schedule} does is not implemented for {self.__class__}")
self.alphas = 1.0 - self.betas
self.alphas_cumprod = jnp.cumprod(self.alphas, axis=0)
self.one = jnp.array(1.0)
self.state = DDPMSchedulerState.create(num_train_timesteps=num_train_timesteps)
self.variance_type = variance_type
def set_timesteps(self, state: DDPMSchedulerState, num_inference_steps: int) -> DDPMSchedulerState:
"""
Sets the discrete timesteps used for the diffusion chain. Supporting function to be run before inference.
Args:
state (`DDIMSchedulerState`):
the `FlaxDDPMScheduler` state data class instance.
num_inference_steps (`int`):
the number of diffusion steps used when generating samples with a pre-trained model.
"""
num_inference_steps = min(self.config.num_train_timesteps, num_inference_steps)
timesteps = jnp.arange(
0, self.config.num_train_timesteps, self.config.num_train_timesteps // num_inference_steps
)[::-1]
return state.replace(num_inference_steps=num_inference_steps, timesteps=timesteps)
def _get_variance(self, t, predicted_variance=None, variance_type=None):
alpha_prod_t = self.alphas_cumprod[t]
alpha_prod_t_prev = self.alphas_cumprod[t - 1] if t > 0 else self.one
# For t > 0, compute predicted variance βt (see formula (6) and (7) from https://arxiv.org/pdf/2006.11239.pdf)
# and sample from it to get previous sample
# x_{t-1} ~ N(pred_prev_sample, variance) == add variance to pred_sample
variance = (1 - alpha_prod_t_prev) / (1 - alpha_prod_t) * self.betas[t]
if variance_type is None:
variance_type = self.config.variance_type
# hacks - were probably added for training stability
if variance_type == "fixed_small":
variance = jnp.clip(variance, a_min=1e-20)
# for rl-diffuser https://arxiv.org/abs/2205.09991
elif variance_type == "fixed_small_log":
variance = jnp.log(jnp.clip(variance, a_min=1e-20))
elif variance_type == "fixed_large":
variance = self.betas[t]
elif variance_type == "fixed_large_log":
# Glide max_log
variance = jnp.log(self.betas[t])
elif variance_type == "learned":
return predicted_variance
elif variance_type == "learned_range":
min_log = variance
max_log = self.betas[t]
frac = (predicted_variance + 1) / 2
variance = frac * max_log + (1 - frac) * min_log
return variance
def step(
self,
state: DDPMSchedulerState,
model_output: jnp.ndarray,
timestep: int,
sample: jnp.ndarray,
key: random.KeyArray,
predict_epsilon: bool = True,
return_dict: bool = True,
) -> Union[FlaxSchedulerOutput, Tuple]:
"""
Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion
process from the learned model outputs (most often the predicted noise).
Args:
state (`DDPMSchedulerState`): the `FlaxDDPMScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`jnp.ndarray`):
current instance of sample being created by diffusion process.
key (`random.KeyArray`): a PRNG key.
predict_epsilon (`bool`):
optional flag to use when model predicts the samples directly instead of the noise, epsilon.
return_dict (`bool`): option for returning tuple rather than SchedulerOutput class
Returns:
[`FlaxSchedulerOutput`] or `tuple`: [`FlaxSchedulerOutput`] if `return_dict` is True, otherwise a `tuple`.
When returning a tuple, the first element is the sample tensor.
"""
t = timestep
if model_output.shape[1] == sample.shape[1] * 2 and self.variance_type in ["learned", "learned_range"]:
model_output, predicted_variance = jnp.split(model_output, sample.shape[1], axis=1)
else:
predicted_variance = None
# 1. compute alphas, betas
alpha_prod_t = self.alphas_cumprod[t]
alpha_prod_t_prev = self.alphas_cumprod[t - 1] if t > 0 else self.one
beta_prod_t = 1 - alpha_prod_t
beta_prod_t_prev = 1 - alpha_prod_t_prev
# 2. compute predicted original sample from predicted noise also called
# "predicted x_0" of formula (15) from https://arxiv.org/pdf/2006.11239.pdf
if predict_epsilon:
pred_original_sample = (sample - beta_prod_t ** (0.5) * model_output) / alpha_prod_t ** (0.5)
else:
pred_original_sample = model_output
# 3. Clip "predicted x_0"
if self.config.clip_sample:
pred_original_sample = jnp.clip(pred_original_sample, -1, 1)
# 4. Compute coefficients for pred_original_sample x_0 and current sample x_t
# See formula (7) from https://arxiv.org/pdf/2006.11239.pdf
pred_original_sample_coeff = (alpha_prod_t_prev ** (0.5) * self.betas[t]) / beta_prod_t
current_sample_coeff = self.alphas[t] ** (0.5) * beta_prod_t_prev / beta_prod_t
# 5. Compute predicted previous sample µ_t
# See formula (7) from https://arxiv.org/pdf/2006.11239.pdf
pred_prev_sample = pred_original_sample_coeff * pred_original_sample + current_sample_coeff * sample
# 6. Add noise
variance = 0
if t > 0:
key = random.split(key, num=1)
noise = random.normal(key=key, shape=model_output.shape)
variance = (self._get_variance(t, predicted_variance=predicted_variance) ** 0.5) * noise
pred_prev_sample = pred_prev_sample + variance
if not return_dict:
return (pred_prev_sample, state)
return FlaxSchedulerOutput(prev_sample=pred_prev_sample, state=state)
def add_noise(
self,
original_samples: jnp.ndarray,
noise: jnp.ndarray,
timesteps: jnp.ndarray,
) -> jnp.ndarray:
sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5
sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples)
sqrt_one_minus_alpha_prod = (1 - self.alphas_cumprod[timesteps]) ** 0.5
sqrt_one_minus_alpha_prod = self.match_shape(sqrt_one_minus_alpha_prod, original_samples)
noisy_samples = sqrt_alpha_prod * original_samples + sqrt_one_minus_alpha_prod * noise
return noisy_samples
def __len__(self):
return self.config.num_train_timesteps

View File

@ -105,7 +105,10 @@ class KarrasVeScheduler(SchedulerMixin, ConfigMixin):
self.num_inference_steps = num_inference_steps
self.timesteps = np.arange(0, self.num_inference_steps)[::-1].copy()
self.schedule = [
(self.sigma_max * (self.sigma_min**2 / self.sigma_max**2) ** (i / (num_inference_steps - 1)))
(
self.config.sigma_max
* (self.config.sigma_min**2 / self.config.sigma_max**2) ** (i / (num_inference_steps - 1))
)
for i in self.timesteps
]
self.schedule = np.array(self.schedule, dtype=np.float32)
@ -121,13 +124,13 @@ class KarrasVeScheduler(SchedulerMixin, ConfigMixin):
TODO Args:
"""
if self.s_min <= sigma <= self.s_max:
gamma = min(self.s_churn / self.num_inference_steps, 2**0.5 - 1)
if self.config.s_min <= sigma <= self.config.s_max:
gamma = min(self.config.s_churn / self.num_inference_steps, 2**0.5 - 1)
else:
gamma = 0
# sample eps ~ N(0, S_noise^2 * I)
eps = self.s_noise * torch.randn(sample.shape, generator=generator).to(sample.device)
eps = self.config.s_noise * torch.randn(sample.shape, generator=generator).to(sample.device)
sigma_hat = sigma + gamma * sigma
sample_hat = sample + ((sigma_hat**2 - sigma**2) ** 0.5 * eps)

View File

@ -0,0 +1,228 @@
# Copyright 2022 NVIDIA 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 dataclasses import dataclass
from typing import Optional, Tuple, Union
import flax
import jax.numpy as jnp
from jax import random
from ..configuration_utils import ConfigMixin, register_to_config
from ..utils import BaseOutput
from .scheduling_utils import SchedulerMixin
@flax.struct.dataclass
class KarrasVeSchedulerState:
# setable values
num_inference_steps: Optional[int] = None
timesteps: Optional[jnp.ndarray] = None
schedule: Optional[jnp.ndarray] = None # sigma(t_i)
@classmethod
def create(cls):
return cls()
@dataclass
class FlaxKarrasVeOutput(BaseOutput):
"""
Output class for the scheduler's step function output.
Args:
prev_sample (`jnp.ndarray` of shape `(batch_size, num_channels, height, width)` for images):
Computed sample (x_{t-1}) of previous timestep. `prev_sample` should be used as next model input in the
denoising loop.
derivative (`jnp.ndarray` of shape `(batch_size, num_channels, height, width)` for images):
Derivate of predicted original image sample (x_0).
state (`KarrasVeSchedulerState`): the `FlaxKarrasVeScheduler` state data class.
"""
prev_sample: jnp.ndarray
derivative: jnp.ndarray
state: KarrasVeSchedulerState
class FlaxKarrasVeScheduler(SchedulerMixin, ConfigMixin):
"""
Stochastic sampling from Karras et al. [1] tailored to the Variance-Expanding (VE) models [2]. Use Algorithm 2 and
the VE column of Table 1 from [1] for reference.
[1] Karras, Tero, et al. "Elucidating the Design Space of Diffusion-Based Generative Models."
https://arxiv.org/abs/2206.00364 [2] Song, Yang, et al. "Score-based generative modeling through stochastic
differential equations." https://arxiv.org/abs/2011.13456
[`~ConfigMixin`] takes care of storing all config attributes that are passed in the scheduler's `__init__`
function, such as `num_train_timesteps`. They can be accessed via `scheduler.config.num_train_timesteps`.
[`~ConfigMixin`] also provides general loading and saving functionality via the [`~ConfigMixin.save_config`] and
[`~ConfigMixin.from_config`] functions.
For more details on the parameters, see the original paper's Appendix E.: "Elucidating the Design Space of
Diffusion-Based Generative Models." https://arxiv.org/abs/2206.00364. The grid search values used to find the
optimal {s_noise, s_churn, s_min, s_max} for a specific model are described in Table 5 of the paper.
Args:
sigma_min (`float`): minimum noise magnitude
sigma_max (`float`): maximum noise magnitude
s_noise (`float`): the amount of additional noise to counteract loss of detail during sampling.
A reasonable range is [1.000, 1.011].
s_churn (`float`): the parameter controlling the overall amount of stochasticity.
A reasonable range is [0, 100].
s_min (`float`): the start value of the sigma range where we add noise (enable stochasticity).
A reasonable range is [0, 10].
s_max (`float`): the end value of the sigma range where we add noise.
A reasonable range is [0.2, 80].
"""
@register_to_config
def __init__(
self,
sigma_min: float = 0.02,
sigma_max: float = 100,
s_noise: float = 1.007,
s_churn: float = 80,
s_min: float = 0.05,
s_max: float = 50,
):
self.state = KarrasVeSchedulerState.create()
def set_timesteps(self, state: KarrasVeSchedulerState, num_inference_steps: int) -> KarrasVeSchedulerState:
"""
Sets the continuous timesteps used for the diffusion chain. Supporting function to be run before inference.
Args:
state (`KarrasVeSchedulerState`):
the `FlaxKarrasVeScheduler` state data class.
num_inference_steps (`int`):
the number of diffusion steps used when generating samples with a pre-trained model.
"""
timesteps = jnp.arange(0, num_inference_steps)[::-1].copy()
schedule = [
(
self.config.sigma_max
* (self.config.sigma_min**2 / self.config.sigma_max**2) ** (i / (num_inference_steps - 1))
)
for i in timesteps
]
return state.replace(
num_inference_steps=num_inference_steps,
schedule=jnp.array(schedule, dtype=jnp.float32),
timesteps=timesteps,
)
def add_noise_to_input(
self,
state: KarrasVeSchedulerState,
sample: jnp.ndarray,
sigma: float,
key: random.KeyArray,
) -> Tuple[jnp.ndarray, float]:
"""
Explicit Langevin-like "churn" step of adding noise to the sample according to a factor gamma_i 0 to reach a
higher noise level sigma_hat = sigma_i + gamma_i*sigma_i.
TODO Args:
"""
if self.config.s_min <= sigma <= self.config.s_max:
gamma = min(self.config.s_churn / state.num_inference_steps, 2**0.5 - 1)
else:
gamma = 0
# sample eps ~ N(0, S_noise^2 * I)
key = random.split(key, num=1)
eps = self.config.s_noise * random.normal(key=key, shape=sample.shape)
sigma_hat = sigma + gamma * sigma
sample_hat = sample + ((sigma_hat**2 - sigma**2) ** 0.5 * eps)
return sample_hat, sigma_hat
def step(
self,
state: KarrasVeSchedulerState,
model_output: jnp.ndarray,
sigma_hat: float,
sigma_prev: float,
sample_hat: jnp.ndarray,
return_dict: bool = True,
) -> Union[FlaxKarrasVeOutput, Tuple]:
"""
Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion
process from the learned model outputs (most often the predicted noise).
Args:
state (`KarrasVeSchedulerState`): the `FlaxKarrasVeScheduler` state data class.
model_output (`torch.FloatTensor` or `np.ndarray`): direct output from learned diffusion model.
sigma_hat (`float`): TODO
sigma_prev (`float`): TODO
sample_hat (`torch.FloatTensor` or `np.ndarray`): TODO
return_dict (`bool`): option for returning tuple rather than SchedulerOutput class
Returns:
[`~schedulers.scheduling_karras_ve_flax.FlaxKarrasVeOutput`] or `tuple`: Updated sample in the diffusion
chain and derivative. [`~schedulers.scheduling_karras_ve_flax.FlaxKarrasVeOutput`] if `return_dict` is
True, otherwise a `tuple`. When returning a tuple, the first element is the sample tensor.
"""
pred_original_sample = sample_hat + sigma_hat * model_output
derivative = (sample_hat - pred_original_sample) / sigma_hat
sample_prev = sample_hat + (sigma_prev - sigma_hat) * derivative
if not return_dict:
return (sample_prev, derivative, state)
return FlaxKarrasVeOutput(prev_sample=sample_prev, derivative=derivative, state=state)
def step_correct(
self,
state: KarrasVeSchedulerState,
model_output: jnp.ndarray,
sigma_hat: float,
sigma_prev: float,
sample_hat: jnp.ndarray,
sample_prev: jnp.ndarray,
derivative: jnp.ndarray,
return_dict: bool = True,
) -> Union[FlaxKarrasVeOutput, Tuple]:
"""
Correct the predicted sample based on the output model_output of the network. TODO complete description
Args:
state (`KarrasVeSchedulerState`): the `FlaxKarrasVeScheduler` state data class.
model_output (`torch.FloatTensor` or `np.ndarray`): direct output from learned diffusion model.
sigma_hat (`float`): TODO
sigma_prev (`float`): TODO
sample_hat (`torch.FloatTensor` or `np.ndarray`): TODO
sample_prev (`torch.FloatTensor` or `np.ndarray`): TODO
derivative (`torch.FloatTensor` or `np.ndarray`): TODO
return_dict (`bool`): option for returning tuple rather than SchedulerOutput class
Returns:
prev_sample (TODO): updated sample in the diffusion chain. derivative (TODO): TODO
"""
pred_original_sample = sample_prev + sigma_prev * model_output
derivative_corr = (sample_prev - pred_original_sample) / sigma_prev
sample_prev = sample_hat + (sigma_prev - sigma_hat) * (0.5 * derivative + 0.5 * derivative_corr)
if not return_dict:
return (sample_prev, derivative, state)
return FlaxKarrasVeOutput(prev_sample=sample_prev, derivative=derivative, state=state)
def add_noise(self, original_samples, noise, timesteps):
raise NotImplementedError()

View File

@ -113,7 +113,7 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
the number of diffusion steps used when generating samples with a pre-trained model.
"""
self.num_inference_steps = num_inference_steps
self.timesteps = np.linspace(self.num_train_timesteps - 1, 0, num_inference_steps, dtype=float)
self.timesteps = np.linspace(self.config.num_train_timesteps - 1, 0, num_inference_steps, dtype=float)
low_idx = np.floor(self.timesteps).astype(int)
high_idx = np.ceil(self.timesteps).astype(int)

View File

@ -0,0 +1,207 @@
# Copyright 2022 Katherine Crowson 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 dataclasses import dataclass
from typing import Optional, Tuple, Union
import flax
import jax.numpy as jnp
from scipy import integrate
from ..configuration_utils import ConfigMixin, register_to_config
from .scheduling_utils import SchedulerMixin, SchedulerOutput
@flax.struct.dataclass
class LMSDiscreteSchedulerState:
# setable values
num_inference_steps: Optional[int] = None
timesteps: Optional[jnp.ndarray] = None
sigmas: Optional[jnp.ndarray] = None
derivatives: jnp.ndarray = jnp.array([])
@classmethod
def create(cls, num_train_timesteps: int, sigmas: jnp.ndarray):
return cls(timesteps=jnp.arange(0, num_train_timesteps)[::-1], sigmas=sigmas)
@dataclass
class FlaxSchedulerOutput(SchedulerOutput):
state: LMSDiscreteSchedulerState
class FlaxLMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
"""
Linear Multistep Scheduler for discrete beta schedules. Based on the original k-diffusion implementation by
Katherine Crowson:
https://github.com/crowsonkb/k-diffusion/blob/481677d114f6ea445aa009cf5bd7a9cdee909e47/k_diffusion/sampling.py#L181
[`~ConfigMixin`] takes care of storing all config attributes that are passed in the scheduler's `__init__`
function, such as `num_train_timesteps`. They can be accessed via `scheduler.config.num_train_timesteps`.
[`~ConfigMixin`] also provides general loading and saving functionality via the [`~ConfigMixin.save_config`] and
[`~ConfigMixin.from_config`] functions.
Args:
num_train_timesteps (`int`): number of diffusion steps used to train the model.
beta_start (`float`): the starting `beta` value of inference.
beta_end (`float`): the final `beta` value.
beta_schedule (`str`):
the beta schedule, a mapping from a beta range to a sequence of betas for stepping the model. Choose from
`linear` or `scaled_linear`.
trained_betas (`jnp.ndarray`, optional):
option to pass an array of betas directly to the constructor to bypass `beta_start`, `beta_end` etc.
options to clip the variance used when adding noise to the denoised sample. Choose from `fixed_small`,
`fixed_small_log`, `fixed_large`, `fixed_large_log`, `learned` or `learned_range`.
"""
@register_to_config
def __init__(
self,
num_train_timesteps: int = 1000,
beta_start: float = 0.0001,
beta_end: float = 0.02,
beta_schedule: str = "linear",
trained_betas: Optional[jnp.ndarray] = None,
):
if trained_betas is not None:
self.betas = jnp.asarray(trained_betas)
if beta_schedule == "linear":
self.betas = jnp.linspace(beta_start, beta_end, num_train_timesteps, dtype=jnp.float32)
elif beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
self.betas = jnp.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=jnp.float32) ** 2
else:
raise NotImplementedError(f"{beta_schedule} does is not implemented for {self.__class__}")
self.alphas = 1.0 - self.betas
self.alphas_cumprod = jnp.cumprod(self.alphas, axis=0)
self.state = LMSDiscreteSchedulerState.create(
num_train_timesteps=num_train_timesteps, sigmas=((1 - self.alphas_cumprod) / self.alphas_cumprod) ** 0.5
)
def get_lms_coefficient(self, state, order, t, current_order):
"""
Compute a linear multistep coefficient.
Args:
order (TODO):
t (TODO):
current_order (TODO):
"""
def lms_derivative(tau):
prod = 1.0
for k in range(order):
if current_order == k:
continue
prod *= (tau - state.sigmas[t - k]) / (state.sigmas[t - current_order] - state.sigmas[t - k])
return prod
integrated_coeff = integrate.quad(lms_derivative, state.sigmas[t], state.sigmas[t + 1], epsrel=1e-4)[0]
return integrated_coeff
def set_timesteps(self, state: LMSDiscreteSchedulerState, num_inference_steps: int) -> LMSDiscreteSchedulerState:
"""
Sets the timesteps used for the diffusion chain. Supporting function to be run before inference.
Args:
state (`LMSDiscreteSchedulerState`):
the `FlaxLMSDiscreteScheduler` state data class instance.
num_inference_steps (`int`):
the number of diffusion steps used when generating samples with a pre-trained model.
"""
timesteps = jnp.linspace(self.config.num_train_timesteps - 1, 0, num_inference_steps, dtype=jnp.float32)
low_idx = jnp.floor(timesteps).astype(int)
high_idx = jnp.ceil(timesteps).astype(int)
frac = jnp.mod(timesteps, 1.0)
sigmas = jnp.array(((1 - self.alphas_cumprod) / self.alphas_cumprod) ** 0.5)
sigmas = (1 - frac) * sigmas[low_idx] + frac * sigmas[high_idx]
sigmas = jnp.concatenate([sigmas, jnp.array([0.0])]).astype(jnp.float32)
return state.replace(
num_inference_steps=num_inference_steps,
timesteps=timesteps,
derivatives=jnp.array([]),
sigmas=sigmas,
)
def step(
self,
state: LMSDiscreteSchedulerState,
model_output: jnp.ndarray,
timestep: int,
sample: jnp.ndarray,
order: int = 4,
return_dict: bool = True,
) -> Union[SchedulerOutput, Tuple]:
"""
Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion
process from the learned model outputs (most often the predicted noise).
Args:
state (`LMSDiscreteSchedulerState`): the `FlaxLMSDiscreteScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`jnp.ndarray`):
current instance of sample being created by diffusion process.
order: coefficient for multi-step inference.
return_dict (`bool`): option for returning tuple rather than SchedulerOutput class
Returns:
[`FlaxSchedulerOutput`] or `tuple`: [`FlaxSchedulerOutput`] if `return_dict` is True, otherwise a `tuple`.
When returning a tuple, the first element is the sample tensor.
"""
sigma = state.sigmas[timestep]
# 1. compute predicted original sample (x_0) from sigma-scaled predicted noise
pred_original_sample = sample - sigma * model_output
# 2. Convert to an ODE derivative
derivative = (sample - pred_original_sample) / sigma
state = state.replace(derivatives=state.derivatives.append(derivative))
if len(state.derivatives) > order:
state = state.replace(derivatives=state.derivatives.pop(0))
# 3. Compute linear multistep coefficients
order = min(timestep + 1, order)
lms_coeffs = [self.get_lms_coefficient(state, order, timestep, curr_order) for curr_order in range(order)]
# 4. Compute previous sample based on the derivatives path
prev_sample = sample + sum(
coeff * derivative for coeff, derivative in zip(lms_coeffs, reversed(state.derivatives))
)
if not return_dict:
return (prev_sample, state)
return FlaxSchedulerOutput(prev_sample=prev_sample, state=state)
def add_noise(
self,
state: LMSDiscreteSchedulerState,
original_samples: jnp.ndarray,
noise: jnp.ndarray,
timesteps: jnp.ndarray,
) -> jnp.ndarray:
sigmas = self.match_shape(state.sigmas[timesteps], noise)
noisy_samples = original_samples + noise * sigmas
return noisy_samples
def __len__(self):
return self.config.num_train_timesteps

View File

@ -108,8 +108,6 @@ class PNDMScheduler(SchedulerMixin, ConfigMixin):
self.alphas = 1.0 - self.betas
self.alphas_cumprod = np.cumprod(self.alphas, axis=0)
self.one = np.array(1.0)
# For now we only support F-PNDM, i.e. the runge-kutta method
# For more information on the algorithm please take a look at the paper: https://arxiv.org/pdf/2202.09778.pdf
# mainly at formula (9), (12), (13) and the Algorithm 2.

View File

@ -25,7 +25,7 @@ from ..configuration_utils import ConfigMixin, register_to_config
from .scheduling_utils import SchedulerMixin, SchedulerOutput
def betas_for_alpha_bar(num_diffusion_timesteps: int, max_beta=0.999):
def betas_for_alpha_bar(num_diffusion_timesteps: int, max_beta=0.999) -> jnp.ndarray:
"""
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].
@ -40,7 +40,7 @@ def betas_for_alpha_bar(num_diffusion_timesteps: int, max_beta=0.999):
prevent singularities.
Returns:
betas (`jnp.array`): the betas used by the scheduler to step the model outputs
betas (`jnp.ndarray`): the betas used by the scheduler to step the model outputs
"""
def alpha_bar(time_step):
@ -56,36 +56,23 @@ def betas_for_alpha_bar(num_diffusion_timesteps: int, max_beta=0.999):
@flax.struct.dataclass
class PNDMSchedulerState:
betas: jnp.array
# setable values
_timesteps: jnp.array
_timesteps: jnp.ndarray
num_inference_steps: Optional[int] = None
_offset: int = 0
prk_timesteps: Optional[jnp.array] = None
plms_timesteps: Optional[jnp.array] = None
timesteps: Optional[jnp.array] = None
prk_timesteps: Optional[jnp.ndarray] = None
plms_timesteps: Optional[jnp.ndarray] = None
timesteps: Optional[jnp.ndarray] = None
# running values
cur_model_output: Optional[jnp.ndarray] = None
counter: int = 0
cur_sample: Optional[jnp.ndarray] = None
ets: jnp.array = jnp.array([])
@property
def alphas(self) -> jnp.array:
return 1.0 - self.betas
@property
def alphas_cumprod(self) -> jnp.array:
return jnp.cumprod(self.alphas, axis=0)
ets: jnp.ndarray = jnp.array([])
@classmethod
def create(cls, betas: jnp.array, num_train_timesteps: int):
return cls(
betas=betas,
_timesteps=jnp.arange(0, num_train_timesteps)[::-1],
)
def create(cls, num_train_timesteps: int):
return cls(_timesteps=jnp.arange(0, num_train_timesteps)[::-1])
@dataclass
@ -112,7 +99,7 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
beta_schedule (`str`):
the beta schedule, a mapping from a beta range to a sequence of betas for stepping the model. Choose from
`linear`, `scaled_linear`, or `squaredcos_cap_v2`.
trained_betas (`np.ndarray`, optional):
trained_betas (`jnp.ndarray`, optional):
option to pass an array of betas directly to the constructor to bypass `beta_start`, `beta_end` etc.
skip_prk_steps (`bool`):
allows the scheduler to skip the Runge-Kutta steps that are defined in the original paper as being required
@ -126,28 +113,31 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
beta_start: float = 0.0001,
beta_end: float = 0.02,
beta_schedule: str = "linear",
trained_betas: Optional[jnp.array] = None,
trained_betas: Optional[jnp.ndarray] = None,
skip_prk_steps: bool = False,
):
if trained_betas is not None:
betas = jnp.asarray(trained_betas)
self.betas = jnp.asarray(trained_betas)
if beta_schedule == "linear":
betas = jnp.linspace(beta_start, beta_end, num_train_timesteps, dtype=jnp.float32)
self.betas = jnp.linspace(beta_start, beta_end, num_train_timesteps, dtype=jnp.float32)
elif beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
betas = jnp.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=jnp.float32) ** 2
self.betas = jnp.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=jnp.float32) ** 2
elif beta_schedule == "squaredcos_cap_v2":
# Glide cosine schedule
betas = betas_for_alpha_bar(num_train_timesteps)
self.betas = betas_for_alpha_bar(num_train_timesteps)
else:
raise NotImplementedError(f"{beta_schedule} does is not implemented for {self.__class__}")
self.alphas = 1.0 - self.betas
self.alphas_cumprod = jnp.cumprod(self.alphas, axis=0)
# For now we only support F-PNDM, i.e. the runge-kutta method
# For more information on the algorithm please take a look at the paper: https://arxiv.org/pdf/2202.09778.pdf
# mainly at formula (9), (12), (13) and the Algorithm 2.
self.pndm_order = 4
self.state = PNDMSchedulerState.create(betas=betas, num_train_timesteps=num_train_timesteps)
self.state = PNDMSchedulerState.create(num_train_timesteps=num_train_timesteps)
def set_timesteps(
self, state: PNDMSchedulerState, num_inference_steps: int, offset: int = 0
@ -157,7 +147,7 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
Args:
state (`PNDMSchedulerState`):
the PNDMScheduler state data class instance.
the `FlaxPNDMScheduler` state data class instance.
num_inference_steps (`int`):
the number of diffusion steps used when generating samples with a pre-trained model.
offset (`int`):
@ -165,7 +155,7 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
"""
step_ratio = self.config.num_train_timesteps // num_inference_steps
# creates integer timesteps by multiplying by ratio
# casting to int to avoid issues when num_inference_step is power of 3
# rounding to avoid issues when num_inference_step is power of 3
_timesteps = (jnp.arange(0, num_inference_steps) * step_ratio).round()[::-1]
_timesteps = _timesteps + offset
@ -212,7 +202,7 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
This function calls `step_prk()` or `step_plms()` depending on the internal variable `counter`.
Args:
state (`PNDMSchedulerState`): the PNDMScheduler state data class instance.
state (`PNDMSchedulerState`): the `FlaxPNDMScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`jnp.ndarray`):
@ -246,7 +236,7 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
solution to the differential equation.
Args:
state (`PNDMSchedulerState`): the PNDMScheduler state data class instance.
state (`PNDMSchedulerState`): the `FlaxPNDMScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`jnp.ndarray`):
@ -268,24 +258,24 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
timestep = state.prk_timesteps[state.counter // 4 * 4]
if state.counter % 4 == 0:
state.replace(
state = state.replace(
cur_model_output=state.cur_model_output + 1 / 6 * model_output,
ets=state.ets.append(model_output),
cur_sample=sample,
)
elif (self.counter - 1) % 4 == 0:
state.replace(cur_model_output=state.cur_model_output + 1 / 3 * model_output)
state = state.replace(cur_model_output=state.cur_model_output + 1 / 3 * model_output)
elif (self.counter - 2) % 4 == 0:
state.replace(cur_model_output=state.cur_model_output + 1 / 3 * model_output)
state = state.replace(cur_model_output=state.cur_model_output + 1 / 3 * model_output)
elif (self.counter - 3) % 4 == 0:
model_output = state.cur_model_output + 1 / 6 * model_output
state.replace(cur_model_output=0)
state = state.replace(cur_model_output=0)
# cur_sample should not be `None`
cur_sample = state.cur_sample if state.cur_sample is not None else sample
prev_sample = self._get_prev_sample(cur_sample, timestep, prev_timestep, model_output, state=state)
state.replace(counter=state.counter + 1)
state = state.replace(counter=state.counter + 1)
if not return_dict:
return (prev_sample, state)
@ -305,7 +295,7 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
times to approximate the solution.
Args:
state (`PNDMSchedulerState`): the PNDMScheduler state data class instance.
state (`PNDMSchedulerState`): the `FlaxPNDMScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`jnp.ndarray`):
@ -333,18 +323,18 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
prev_timestep = max(timestep - self.config.num_train_timesteps // state.num_inference_steps, 0)
if state.counter != 1:
state.replace(ets=state.ets.append(model_output))
state = state.replace(ets=state.ets.append(model_output))
else:
prev_timestep = timestep
timestep = timestep + self.config.num_train_timesteps // state.num_inference_steps
if len(state.ets) == 1 and state.counter == 0:
model_output = model_output
state.replace(cur_sample=sample)
state = state.replace(cur_sample=sample)
elif len(state.ets) == 1 and state.counter == 1:
model_output = (model_output + state.ets[-1]) / 2
sample = state.cur_sample
state.replace(cur_sample=None)
state = state.replace(cur_sample=None)
elif len(state.ets) == 2:
model_output = (3 * state.ets[-1] - state.ets[-2]) / 2
elif len(state.ets) == 3:
@ -355,7 +345,7 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
)
prev_sample = self._get_prev_sample(sample, timestep, prev_timestep, model_output, state=state)
state.replace(counter=state.counter + 1)
state = state.replace(counter=state.counter + 1)
if not return_dict:
return (prev_sample, state)
@ -375,8 +365,8 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
# sample -> x_t
# model_output -> e_θ(x_t, t)
# prev_sample -> x_(tδ)
alpha_prod_t = state.alphas_cumprod[timestep + 1 - state._offset]
alpha_prod_t_prev = state.alphas_cumprod[timestep_prev + 1 - state._offset]
alpha_prod_t = self.alphas_cumprod[timestep + 1 - state._offset]
alpha_prod_t_prev = self.alphas_cumprod[timestep_prev + 1 - state._offset]
beta_prod_t = 1 - alpha_prod_t
beta_prod_t_prev = 1 - alpha_prod_t_prev
@ -400,14 +390,13 @@ class FlaxPNDMScheduler(SchedulerMixin, ConfigMixin):
def add_noise(
self,
state: PNDMSchedulerState,
original_samples: jnp.ndarray,
noise: jnp.ndarray,
timesteps: jnp.ndarray,
) -> jnp.ndarray:
sqrt_alpha_prod = state.alphas_cumprod[timesteps] ** 0.5
sqrt_alpha_prod = self.alphas_cumprod[timesteps] ** 0.5
sqrt_alpha_prod = self.match_shape(sqrt_alpha_prod, original_samples)
sqrt_one_minus_alpha_prod = (1 - state.alphas_cumprod[timesteps]) ** 0.5
sqrt_one_minus_alpha_prod = (1 - self.alphas_cumprod[timesteps]) ** 0.5
sqrt_one_minus_alpha_prod = self.match_shape(sqrt_one_minus_alpha_prod, original_samples)
noisy_samples = sqrt_alpha_prod * original_samples + sqrt_one_minus_alpha_prod * noise

View File

@ -55,6 +55,7 @@ class ScoreSdeVeScheduler(SchedulerMixin, ConfigMixin):
[`~ConfigMixin.from_config`] functions.
Args:
num_train_timesteps (`int`): number of diffusion steps used to train the model.
snr (`float`):
coefficient weighting the step from the model_output sample (from the network) to the random noise.
sigma_min (`float`):

View File

@ -0,0 +1,260 @@
# Copyright 2022 Google Brain 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.
# DISCLAIMER: This file is strongly influenced by https://github.com/yang-song/score_sde_pytorch
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import flax
import jax.numpy as jnp
from jax import random
from ..configuration_utils import ConfigMixin, register_to_config
from .scheduling_utils import SchedulerMixin, SchedulerOutput
@flax.struct.dataclass
class ScoreSdeVeSchedulerState:
# setable values
timesteps: Optional[jnp.ndarray] = None
discrete_sigmas: Optional[jnp.ndarray] = None
sigmas: Optional[jnp.ndarray] = None
@classmethod
def create(cls):
return cls()
@dataclass
class FlaxSdeVeOutput(SchedulerOutput):
"""
Output class for the ScoreSdeVeScheduler's step function output.
Args:
state (`ScoreSdeVeSchedulerState`):
prev_sample (`jnp.ndarray` of shape `(batch_size, num_channels, height, width)` for images):
Computed sample (x_{t-1}) of previous timestep. `prev_sample` should be used as next model input in the
denoising loop.
prev_sample_mean (`jnp.ndarray` of shape `(batch_size, num_channels, height, width)` for images):
Mean averaged `prev_sample`. Same as `prev_sample`, only mean-averaged over previous timesteps.
"""
state: ScoreSdeVeSchedulerState
prev_sample: jnp.ndarray
prev_sample_mean: Optional[jnp.ndarray] = None
class FlaxScoreSdeVeScheduler(SchedulerMixin, ConfigMixin):
"""
The variance exploding stochastic differential equation (SDE) scheduler.
For more information, see the original paper: https://arxiv.org/abs/2011.13456
[`~ConfigMixin`] takes care of storing all config attributes that are passed in the scheduler's `__init__`
function, such as `num_train_timesteps`. They can be accessed via `scheduler.config.num_train_timesteps`.
[`~ConfigMixin`] also provides general loading and saving functionality via the [`~ConfigMixin.save_config`] and
[`~ConfigMixin.from_config`] functions.
Args:
num_train_timesteps (`int`): number of diffusion steps used to train the model.
snr (`float`):
coefficient weighting the step from the model_output sample (from the network) to the random noise.
sigma_min (`float`):
initial noise scale for sigma sequence in sampling procedure. The minimum sigma should mirror the
distribution of the data.
sigma_max (`float`): maximum value used for the range of continuous timesteps passed into the model.
sampling_eps (`float`): the end value of sampling, where timesteps decrease progressively from 1 to
epsilon.
correct_steps (`int`): number of correction steps performed on a produced sample.
"""
@register_to_config
def __init__(
self,
num_train_timesteps: int = 2000,
snr: float = 0.15,
sigma_min: float = 0.01,
sigma_max: float = 1348.0,
sampling_eps: float = 1e-5,
correct_steps: int = 1,
):
state = ScoreSdeVeSchedulerState.create()
self.state = self.set_sigmas(state, num_train_timesteps, sigma_min, sigma_max, sampling_eps)
def set_timesteps(
self, state: ScoreSdeVeSchedulerState, num_inference_steps: int, sampling_eps: float = None
) -> ScoreSdeVeSchedulerState:
"""
Sets the continuous timesteps used for the diffusion chain. Supporting function to be run before inference.
Args:
state (`ScoreSdeVeSchedulerState`): the `FlaxScoreSdeVeScheduler` state data class instance.
num_inference_steps (`int`):
the number of diffusion steps used when generating samples with a pre-trained model.
sampling_eps (`float`, optional): final timestep value (overrides value given at Scheduler instantiation).
"""
sampling_eps = sampling_eps if sampling_eps is not None else self.config.sampling_eps
timesteps = jnp.linspace(1, sampling_eps, num_inference_steps)
return state.replace(timesteps=timesteps)
def set_sigmas(
self,
state: ScoreSdeVeSchedulerState,
num_inference_steps: int,
sigma_min: float = None,
sigma_max: float = None,
sampling_eps: float = None,
) -> ScoreSdeVeSchedulerState:
"""
Sets the noise scales used for the diffusion chain. Supporting function to be run before inference.
The sigmas control the weight of the `drift` and `diffusion` components of sample update.
Args:
state (`ScoreSdeVeSchedulerState`): the `FlaxScoreSdeVeScheduler` state data class instance.
num_inference_steps (`int`):
the number of diffusion steps used when generating samples with a pre-trained model.
sigma_min (`float`, optional):
initial noise scale value (overrides value given at Scheduler instantiation).
sigma_max (`float`, optional): final noise scale value (overrides value given at Scheduler instantiation).
sampling_eps (`float`, optional): final timestep value (overrides value given at Scheduler instantiation).
"""
sigma_min = sigma_min if sigma_min is not None else self.config.sigma_min
sigma_max = sigma_max if sigma_max is not None else self.config.sigma_max
sampling_eps = sampling_eps if sampling_eps is not None else self.config.sampling_eps
if state.timesteps is None:
state = self.set_timesteps(state, num_inference_steps, sampling_eps)
discrete_sigmas = jnp.exp(jnp.linspace(jnp.log(sigma_min), jnp.log(sigma_max), num_inference_steps))
sigmas = jnp.array([sigma_min * (sigma_max / sigma_min) ** t for t in state.timesteps])
return state.replace(discrete_sigmas=discrete_sigmas, sigmas=sigmas)
def get_adjacent_sigma(self, state, timesteps, t):
return jnp.where(timesteps == 0, jnp.zeros_like(t), state.discrete_sigmas[timesteps - 1])
def step_pred(
self,
state: ScoreSdeVeSchedulerState,
model_output: jnp.ndarray,
timestep: int,
sample: jnp.ndarray,
key: random.KeyArray,
return_dict: bool = True,
) -> Union[FlaxSdeVeOutput, Tuple]:
"""
Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion
process from the learned model outputs (most often the predicted noise).
Args:
state (`ScoreSdeVeSchedulerState`): the `FlaxScoreSdeVeScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
timestep (`int`): current discrete timestep in the diffusion chain.
sample (`jnp.ndarray`):
current instance of sample being created by diffusion process.
generator: random number generator.
return_dict (`bool`): option for returning tuple rather than SchedulerOutput class
Returns:
[`FlaxSdeVeOutput`] or `tuple`: [`FlaxSdeVeOutput`] if `return_dict` is True, otherwise a `tuple`. When
returning a tuple, the first element is the sample tensor.
"""
if state.timesteps is None:
raise ValueError(
"`state.timesteps` is not set, you need to run 'set_timesteps' after creating the scheduler"
)
timestep = timestep * jnp.ones(
sample.shape[0],
)
timesteps = (timestep * (len(state.timesteps) - 1)).long()
sigma = state.discrete_sigmas[timesteps]
adjacent_sigma = self.get_adjacent_sigma(state, timesteps, timestep)
drift = jnp.zeros_like(sample)
diffusion = (sigma**2 - adjacent_sigma**2) ** 0.5
# equation 6 in the paper: the model_output modeled by the network is grad_x log pt(x)
# also equation 47 shows the analog from SDE models to ancestral sampling methods
drift = drift - diffusion[:, None, None, None] ** 2 * model_output
# equation 6: sample noise for the diffusion term of
key = random.split(key, num=1)
noise = random.normal(key=key, shape=sample.shape)
prev_sample_mean = sample - drift # subtract because `dt` is a small negative timestep
# TODO is the variable diffusion the correct scaling term for the noise?
prev_sample = prev_sample_mean + diffusion[:, None, None, None] * noise # add impact of diffusion field g
if not return_dict:
return (prev_sample, prev_sample_mean, state)
return FlaxSdeVeOutput(prev_sample=prev_sample, prev_sample_mean=prev_sample_mean, state=state)
def step_correct(
self,
state: ScoreSdeVeSchedulerState,
model_output: jnp.ndarray,
sample: jnp.ndarray,
key: random.KeyArray,
return_dict: bool = True,
) -> Union[SchedulerOutput, Tuple]:
"""
Correct the predicted sample based on the output model_output of the network. This is often run repeatedly
after making the prediction for the previous timestep.
Args:
state (`ScoreSdeVeSchedulerState`): the `FlaxScoreSdeVeScheduler` state data class instance.
model_output (`jnp.ndarray`): direct output from learned diffusion model.
sample (`jnp.ndarray`):
current instance of sample being created by diffusion process.
generator: random number generator.
return_dict (`bool`): option for returning tuple rather than SchedulerOutput class
Returns:
[`FlaxSdeVeOutput`] or `tuple`: [`FlaxSdeVeOutput`] if `return_dict` is True, otherwise a `tuple`. When
returning a tuple, the first element is the sample tensor.
"""
if state.timesteps is None:
raise ValueError(
"`state.timesteps` is not set, you need to run 'set_timesteps' after creating the scheduler"
)
# For small batch sizes, the paper "suggest replacing norm(z) with sqrt(d), where d is the dim. of z"
# sample noise for correction
key = random.split(key, num=1)
noise = random.normal(key=key, shape=sample.shape)
# compute step size from the model_output, the noise, and the snr
grad_norm = jnp.linalg.norm(model_output)
noise_norm = jnp.linalg.norm(noise)
step_size = (self.config.snr * noise_norm / grad_norm) ** 2 * 2
step_size = step_size * jnp.ones(sample.shape[0])
# compute corrected sample: model_output term and noise term
prev_sample_mean = sample + step_size[:, None, None, None] * model_output
prev_sample = prev_sample_mean + ((step_size * 2) ** 0.5)[:, None, None, None] * noise
if not return_dict:
return (prev_sample, state)
return FlaxSdeVeOutput(prev_sample=prev_sample, state=state)
def __len__(self):
return self.config.num_train_timesteps

View File

@ -11,8 +11,43 @@ class FlaxModelMixin(metaclass=DummyObject):
requires_backends(self, ["flax"])
class FlaxDDIMScheduler(metaclass=DummyObject):
_backends = ["flax"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["flax"])
class FlaxDDPMScheduler(metaclass=DummyObject):
_backends = ["flax"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["flax"])
class FlaxKarrasVeScheduler(metaclass=DummyObject):
_backends = ["flax"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["flax"])
class FlaxLMSDiscreteScheduler(metaclass=DummyObject):
_backends = ["flax"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["flax"])
class FlaxPNDMScheduler(metaclass=DummyObject):
_backends = ["flax"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["flax"])
class FlaxScoreSdeVeScheduler(metaclass=DummyObject):
_backends = ["flax"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["flax"])

View File

@ -814,7 +814,7 @@ class ScoreSdeVeSchedulerTest(unittest.TestCase):
for i, t in enumerate(scheduler.timesteps):
sigma_t = scheduler.sigmas[i]
for _ in range(scheduler.correct_steps):
for _ in range(scheduler.config.correct_steps):
with torch.no_grad():
model_output = model(sample, sigma_t)
sample = scheduler.step_correct(model_output, sample, generator=generator, **kwargs).prev_sample

View File

@ -52,7 +52,7 @@ class TrainingTests(unittest.TestCase):
tensor_format="pt",
)
assert ddpm_scheduler.num_train_timesteps == ddim_scheduler.num_train_timesteps
assert ddpm_scheduler.config.num_train_timesteps == ddim_scheduler.config.num_train_timesteps
# shared batches for DDPM and DDIM
set_seed(0)