+hq/ S SK r S SKrS SKrS SKJr S SKJr S SKJrJ r J
r
S SK r S SK J
r S SKJr SSKJrJr Sr " S S
\5 r\ " S S \5 5 r " S
S\5 rS\R0 S\ \ S\R0 4S jrS\R6 4S\S\R0 4S jjr \ R: R " S S5 5 r S\ S\R0 S\R0 S\R0 4S jr S\ S\R0 S\R0 S\R0 4S jr S\ S \R0 S\R0 S\R0 4S jr!g) N) dataclass)Enum)OptionalTupleUnion)validate_hf_hub_args )
BaseOutputPushToHubMixinzscheduler_config.jsonc , \ rS rSrSrSrSrSrSrSr Sr
g )
FlaxKarrasDiffusionSchedulers# r N) __name__
__module__ __qualname____firstlineno__FlaxDDIMSchedulerFlaxDDPMSchedulerFlaxPNDMSchedulerFlaxLMSDiscreteScheduler FlaxDPMSolverMultistepSchedulerFlaxEulerDiscreteScheduler__static_attributes__r d/home/james-whalen/.local/lib/python3.13/site-packages/diffusers/schedulers/scheduling_utils_flax.pyr
r
# s& &'#!" r r
c 8 \ rS rSr% Sr\R \S' Srg)FlaxSchedulerOutput, a&
Base 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.
prev_sampler N) r r r r __doc__jnpndarray__annotations__r r r r! r# r# , s r r# c \ rS rSrSr\rS/r/ rSr \
\ SS\ \
\\R 4 S\ \ 4S jj5 5 rSS \
\\R 4 S
\4S jjr\S 5 r\
S
5 rSrg)FlaxSchedulerMixin: aA
Mixin containing common functions for the schedulers.
Class attributes:
- **_compatibles** (`List[str]`) -- A list of classes that are compatible with the parent class, so that
`from_config` can be used from a class different than the one used to save the config (should be overridden
by parent class).
dtypeTN pretrained_model_name_or_path subfolderc U R " SUUSS.UD6u pTU R " U4SS0UD6u pg[ US5 ( a" [ USS5 ( a UR 5 nU( a UWU4$ UW4$ )a
Instantiate a Scheduler class from a pre-defined JSON-file.
Parameters:
pretrained_model_name_or_path (`str` or `os.PathLike`, *optional*):
Can be either:
- A string, the *model id* of a model repo on huggingface.co. Valid model ids should have an
organization name, like `google/ddpm-celebahq-256`.
- A path to a *directory* containing model weights saved using [`~SchedulerMixin.save_pretrained`],
e.g., `./my_model_directory/`.
subfolder (`str`, *optional*):
In case the relevant files are located inside a subfolder of the model repo (either remote in
huggingface.co or downloaded locally), you can specify the folder name here.
return_unused_kwargs (`bool`, *optional*, defaults to `False`):
Whether kwargs that are not consumed by the Python class should be returned or not.
cache_dir (`Union[str, os.PathLike]`, *optional*):
Path to a directory in which a downloaded pretrained model configuration should be cached if the
standard cache should not be used.
force_download (`bool`, *optional*, defaults to `False`):
Whether or not to force the (re-)download of the model weights and configuration files, overriding the
cached versions if they exist.
proxies (`Dict[str, str]`, *optional*):
A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128',
'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request.
output_loading_info(`bool`, *optional*, defaults to `False`):
Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages.
local_files_only(`bool`, *optional*, defaults to `False`):
Whether or not to only look at local files (i.e., do not try to download the model).
token (`str` or *bool*, *optional*):
The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated
when running `transformers-cli login` (stored in `~/.huggingface`).
revision (`str`, *optional*, defaults to `"main"`):
The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a
git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any
identifier allowed by git.
It is required to be logged in (`hf auth login`) when you want to use private or [gated
models](https://huggingface.co/docs/hub/models-gated#gated-models).
Activate the special ["offline-mode"](https://huggingface.co/transformers/installation.html#offline-mode) to
use this method in a firewalled environment.
T)r. r/ return_unused_kwargsr1 create_state has_stateFr ) load_config from_confighasattrgetattrr2 ) clsr. r/ r1 kwargsconfig scheduler
unused_kwargsstates r! from_pretrained"FlaxSchedulerMixin.from_pretrainedI s ~
*G !%
$'??6#_PT#_X^#_ 9n - -')[RW2X2X **,E e]2 2% r save_directory push_to_hubc , U R " SXS.UD6 g)a
Save a scheduler configuration object to the directory `save_directory`, so that it can be re-loaded using the
[`~FlaxSchedulerMixin.from_pretrained`] class method.
Args:
save_directory (`str` or `os.PathLike`):
Directory where the configuration JSON file will be saved (will be created if it does not exist).
push_to_hub (`bool`, *optional*, defaults to `False`):
Whether or not to push your model to the Hugging Face Hub after saving it. You can specify the
repository you want to push to with `repo_id` (will default to the name of `save_directory` in your
namespace).
kwargs (`Dict[str, Any]`, *optional*):
Additional keyword arguments passed along to the [`~utils.PushToHubMixin.push_to_hub`] method.
)r@ rA Nr ) save_config)selfr@ rA r9 s r! save_pretrained"FlaxSchedulerMixin.save_pretrained s
ZZSYZr c " U R 5 $ )z
Returns all schedulers that are compatible with this scheduler
Returns:
`List[SchedulerMixin]`: List of compatible schedulers
)_get_compatibles)rD s r! compatibles FlaxSchedulerMixin.compatibles s $$&&r c [ [ U R /U R - 5 5 n[ R
" [ R
S5 S 5 nU Vs/ s H n[ X#5 ( d M [ X#5 PM" nnU$ s snf )N.r ) listsetr _compatibles importlib
import_modulesplitr6 r7 )r8 compatible_classes_strdiffusers_libraryccompatible_classess r! rH #FlaxSchedulerMixin._get_compatibles s} !%c3<<.3;K;K*K&L!M % 3 3HNN34G4J K3I
3IaWUfMj )G% )3I
"!
s B5Br )NNF)F)r r r r r& SCHEDULER_CONFIG_NAME config_nameignore_for_configrO has_compatibles classmethodr r r strosPathLiker> boolrE propertyrI rH r r r r! r+ r+ : s (K LO LP#' " K '/c2;;6F0G'HK C=K K Z[eC4D.E [TX [" ' ' " "r r+ xshapereturnc [ U5 U R : d e[ R " U R U R
S[ U5 U R -
- - 5 U5 $ )N)r )lenndimr' broadcast_toreshaperc )rb rc s r! broadcast_to_shape_from_leftrj sO u: AIIaggE
QVV8K0L&L Mu UUr g+?num_diffusion_timestepsc
S n/ n[ U 5 H< nXP- nUS- U - nUR [ SU" U5 U" U5 - -
U5 5 M> [ R " XBS9$ )aN
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
c h [ R " U S- S- [ R - S- 5 S- $ )NgMb?gT㥛 ?r )mathcospi) time_steps r! alpha_bar&betas_for_alpha_bar..alpha_bar s. xxU*e3dgg=ABaGGr r r- )rangeappendminr' array)rk max_betar- rr betasit1t2s r! betas_for_alpha_barr~ sl $H
E
*
+
(!e.
.
SYr]Yr] ::HEF , 99U ((r c \ rS rSr% \R
\S' \R
\S' \R
\S' \S 5 rSr g)CommonSchedulerState alphasrz alphas_cumprodc UR nUR b+ [ R " UR UR S9nGOUR
S:X a@ [ R " UR UR UR UR S9nOUR
S:X aI [ R " UR S- UR S- UR UR S9S- nO^UR
S:X a [ UR UR S9nO/[ SUR
SUR R 35 eS U-
n[ R " US
S 9nU " UUUS 9$ )
Nrt linear
scaled_linear ?r squaredcos_cap_v2zbeta_schedule z" is not implemented for scheduler g ?r )axis)r rz r )r:
trained_betasr' asarrayr-
beta_schedulelinspace
beta_startbeta_endnum_train_timestepsr~ NotImplementedError __class__r cumprod)r8 r; r: rz r r s r! createCommonSchedulerState.create sL !! +KK 4 4IOOLE
!
!X
-LL!2!2FOOVE_E_gpgvgvwE
!
!_
4 %%s*FOOS,@&B\B\dmdsds
!
!%8
8'(B(B)//ZE% !5!5 66XYbYlYlYuYuXvw
u V!4)
r r N)
r r r r r' r( r) r\ r r r r r! r r s2 KK ;;KK
r r r= original_samplesnoise timestepsc U R nXC S- nUR 5 n[ XQR 5 nSXC -
S- nUR 5 n[ XaR 5 nXV4$ )Nr r )r flattenrj rc )r= r r r r sqrt_alpha_prodsqrt_one_minus_alpha_prods r! get_sqrt_alpha_prodr sw ))N$/36O%--/O2?DZDZ[O!"^%>!>3 F 9 A A C <=VXnXn o 55r c 4 [ XX#5 u pEXA- XR- - nU$ Nr )r= r r r r r
noisy_sampless r! add_noise_commonr s, 2EU^c1o.O#69R9ZZM r samplec 4 [ XX#5 u pEXB- XQ- -
nU$ r r )r= r r r r r velocitys r! get_velocity_commonr s( 1DUTY1e.O &)B)KKH Or )"rP rn r^ dataclassesr enumr typingr r r flax jax.numpynumpyr' huggingface_hub.utilsr utilsr
r rX r
r# r+ r( intrj float32r~ structr r r r r r r! r su ! ) ) 6 . 0 #D #
*
@" @"FVCKK Vc
Vs{{ V
@ECKK ) )\_\g\g ): %
%
%
P
6
636;;
6GJ{{
6_b_j_j
6 36;;GJ{{_b_j_j 3 S[[ QTQ\Q\ ilitit r