3jecSSKrSSKJr SSKJr SSKrSSKJrJr SSK J r J r SSK J r SS KJr \ R "\5r\"S S \ 55r"S S \\5rg)N) dataclass)Literal) ConfigMixinregister_to_config) BaseOutputlogging) randn_tensor)SchedulerMixinc`\rSrSr%Sr\R \S'Sr\R S-\S'Sr g)EDMEulerSchedulerOutputaM Output class for the scheduler's `step` function output. Args: prev_sample (`torch.Tensor` 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. pred_original_sample (`torch.Tensor` of shape `(batch_size, num_channels, height, width)` for images): The predicted denoised sample `(x_{0})` based on the model output from the current timestep. `pred_original_sample` can be used to preview progress or for guidance. prev_sampleNpred_original_sample) __name__ __module__ __qualname____firstlineno____doc__torchTensor__annotations__r__static_attributes__rc/home/wildlama/miniconda3/lib/python3.13/site-packages/diffusers/schedulers/scheduling_edm_euler.pyrrs' 04%,,-4rrcR\rSrSrSr/rSr\SS\ S \\R0-S!\R$\\-S-4S"jjrS?S#\R$S\S-S\S-S\R$4S$jjrS?S#\R$S\S-S\S-S\R$4S%jjrS@S\\R$-S&\R$S-S\ 4S'jjrS\\R$-SS4S(jrS)S)\"S*5S+SS,S4S\R$S\\R$-S\R$S-\S.\S/\S0\S1\R>S-S2\ S3\R$S-S\!\"-4S4jjr#S5\R$S6\R$S7\R$S\R$4S8jr$S\\R$-S\\R$-4S9jr%S\ 4S:jr&S;r'g)AEDMEulerScheduler1a Implements the Euler scheduler in EDM formulation as presented in Karras et al. 2022 [1]. [1] Karras, Tero, et al. "Elucidating the Design Space of Diffusion-Based Generative Models." https://huggingface.co/papers/2206.00364 This model inherits from [`SchedulerMixin`] and [`ConfigMixin`]. Check the superclass documentation for the generic methods the library implements for all schedulers such as loading and saving. Args: sigma_min (`float`, *optional*, defaults to `0.002`): Minimum noise magnitude in the sigma schedule. This was set to 0.002 in the EDM paper [1]; a reasonable range is [0, 10]. sigma_max (`float`, *optional*, defaults to `80.0`): Maximum noise magnitude in the sigma schedule. This was set to 80.0 in the EDM paper [1]; a reasonable range is [0.2, 80.0]. sigma_data (`float`, *optional*, defaults to `0.5`): The standard deviation of the data distribution. This is set to 0.5 in the EDM paper [1]. sigma_schedule (`Literal["karras", "exponential"]`, *optional*, defaults to `"karras"`): Sigma schedule to compute the `sigmas`. By default, we use the schedule introduced in the EDM paper (https://huggingface.co/papers/2206.00364). The `"exponential"` schedule was incorporated in this model: https://huggingface.co/stabilityai/cosxl. num_train_timesteps (`int`, *optional*, defaults to `1000`): The number of diffusion steps to train the model. prediction_type (`Literal["epsilon", "v_prediction"]`, *optional*, defaults to `"epsilon"`): Prediction type of the scheduler function. `"epsilon"` predicts the noise of the diffusion process, and `"v_prediction"` (see section 2.4 of [Imagen Video](https://huggingface.co/papers/2210.02303) paper). rho (`float`, *optional*, defaults to `7.0`): The rho parameter used for calculating the Karras sigma schedule, which is set to 7.0 in the EDM paper [1]. final_sigmas_type (`Literal["zero", "sigma_min"]`, *optional*, defaults to `"zero"`): The final `sigma` value for the noise schedule during the sampling process. If `"sigma_min"`, the final sigma is the same as the last sigma in the training schedule. If `"zero"`, the final sigma is set to 0. r sigma_min sigma_max sigma_datasigma_schedulekarras exponentialnum_train_timestepsprediction_type)epsilon v_predictionrhofinal_sigmas_type)zeror!returnNc rUS;a[SU<S35eSUl[RRR 5(a[R O[Rn [R"US-U S9U- n US:XaURU 5n OUS:XaURU 5n U R[R 5n URU 5Ul URRS:XaU S n O?URRS :XaS n O"[S URR35e[R "U [R""S XR$S9/5UlSUlSUlSUlUR&RS5Ulg)Nr%z-Wrong value for provided for `sigma_schedule=z`.`r dtyper&r'r!r.rC`final_sigmas_type` must be one of 'zero', or 'sigma_min', but got r  fill_valuedeviceFcpu) ValueErrornum_inference_stepsrbackendsmps is_availablefloat32float64arange_compute_karras_sigmas_compute_exponential_sigmastoprecondition_noise timestepsconfigr-catfullr8sigmasis_scale_input_called _step_index _begin_index) selfr!r"r#r$r(r)r,r- sigmas_dtyperJ sigma_lasts r__init__EDMEulerScheduler.__init__Wso !: :Mn=NcRS S$( (-(:(:(G(G(I(Iu}}u}} 1A5\JM`` X %008F } ,55f=F5==)008 ;; ( (K 7J [[ * *f 4JUVZVaVaVsVsUtu iiDZXeXe)f gh %*" kknnU+ rc@URRS-S-S-$)z Return the standard deviation of the initial noise distribution. Returns: `float`: The initial noise sigma value computed as `(sigma_max**2 + 1) ** 0.5`. rr ?)rGr"rNs rinit_noise_sigma"EDMEulerScheduler.init_noise_sigmas# %%q(1,44rcUR$)z Return the index counter for the current timestep. The index will increase by 1 after each scheduler step. Returns: `int` or `None`: The current step index, or `None` if not yet initialized. )rLrUs r step_indexEDMEulerScheduler.step_indexsrcUR$)z Return the index for the first timestep. This should be set from the pipeline with the `set_begin_index` method. Returns: `int` or `None`: The begin index, or `None` if not yet set. rMrUs r begin_indexEDMEulerScheduler.begin_indexs   rr]cXlg)z Sets the begin index for the scheduler. This function should be run from pipeline before the inference. Args: begin_index (`int`, defaults to `0`): The begin index for the scheduler. Nr\)rNr]s rset_begin_index!EDMEulerScheduler.set_begin_indexs (rsamplesigmac0URU5nX-nU$)a9 Precondition the input sample by scaling it according to the EDM formulation. Args: sample (`torch.Tensor`): The input sample tensor to precondition. sigma (`float` or `torch.Tensor`): The current sigma (noise level) value. Returns: `torch.Tensor`: The scaled input sample. )_get_conditioning_c_in)rNrbrcc_in scaled_samples rprecondition_inputs%EDMEulerScheduler.precondition_inputss!**51 rc[U[R5(d[R"U/5nS[R"U5-nU$)a Precondition the noise level by applying a logarithmic transformation. Args: sigma (`float` or `torch.Tensor`): The sigma (noise level) value to precondition. Returns: `torch.Tensor`: The preconditioned noise value computed as `0.25 * log(sigma)`. g?) isinstancerrtensorlog)rNrcc_noises rrE$EDMEulerScheduler.precondition_noises=%..LL%)E5))r model_outputchURRnUS-US-US--- nURRS:XaX4-US-US--S-- nOSURRS:XaU*U-US-US--S-- nO#[SURRS35eXQ-Xb--nU$)a Precondition the model outputs according to the EDM formulation. Args: sample (`torch.Tensor`): The input sample tensor. model_output (`torch.Tensor`): The direct output from the learned diffusion model. sigma (`float` or `torch.Tensor`): The current sigma (noise level) value. Returns: `torch.Tensor`: The denoised sample computed by combining the skip connection and output scaling. rr*rTr+zPrediction type z is not supported.)rGr#r)r:)rNrbrprcr#c_skipc_outdenoiseds rprecondition_outputs&EDMEulerScheduler.precondition_outputss*[[++ Q%(Z]":; ;; & &) 3&%(Z]*Bs)JJE [[ ( (N :FZ'5!8j!m+C*KKE/ 0K0K/LL^_` `?U%99rtimestepcURcURU5 URURnURX5nSUlU$)a Scale the denoising model input to match the Euler algorithm. Ensures interchangeability with schedulers that need to scale the denoising model input depending on the current timestep. Args: sample (`torch.Tensor`): The input sample tensor. timestep (`float` or `torch.Tensor`): The current timestep in the diffusion chain. Returns: `torch.Tensor`: A scaled input sample. T)rY_init_step_indexrJrhrK)rNrbrwrcs rscale_model_input#EDMEulerScheduler.scale_model_inputsK ?? "  ! !( + DOO,))&8%)" rr;r8rJc Xl[RRR 5(a[R O[R nUc"[R"SSURUS9nO;[U[5(a[R"X4S9nOURU5nURRS:XaURU5nO+URRS:XaURU5nUR[R US9nUR!U5UlURR$S:XaUS nO?URR$S :XaSnO"['S URR$35e[R("U[R*"S XSR,S 9/5UlSUlSUlUR.RS5Ulg)aP Sets the discrete timesteps used for the diffusion chain (to be run before inference). Args: num_inference_steps (`int`, *optional*): The number of diffusion steps used when generating samples with a pre-trained model. device (`str` or `torch.device`, *optional*): The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. sigmas (`torch.Tensor | list[float]`, *optional*): Custom sigmas to use for the denoising process. If not defined, the default behavior when `num_inference_steps` is passed will be used. Nrr r1r&r')r2r8r!r3r.r4r5r6r9)r;rr<r=r>r?r@linspacerkfloatrlrDrGr$rBrCrErFr-r:rHrIr8rJrLrM)rNr;r8rJrOrPs r set_timestepsEDMEulerScheduler.set_timestepss$$7 (-(:(:(G(G(I(Iu}}u}} >^^Aq$*B*B,WF  & &\\&=FYY|,F ;; % % 1008F [[ ' '= 855f=Fv>008 ;; ( (K 7J [[ * *f 4JUVZVaVaVsVsUtu iiDZXeXe)f gh  kknnU+ rrampcU=(d URRnU=(d URRnURRnUSU- -nUSU- -nXaXV- --U-nU$)a Construct the noise schedule of [Karras et al. (2022)](https://huggingface.co/papers/2206.00364). Args: ramp (`torch.Tensor`): A tensor of values in [0, 1] representing the interpolation positions. sigma_min (`float`, *optional*): Minimum sigma value. If `None`, uses `self.config.sigma_min`. sigma_max (`float`, *optional*): Maximum sigma value. If `None`, uses `self.config.sigma_max`. Returns: `torch.Tensor`: The computed Karras sigma schedule. r )rGr!r"r,)rNrr!r"r, min_inv_rho max_inv_rhorJs rrB(EDMEulerScheduler._compute_karras_sigmasCso*6!6!6 6!6!6 kkooAG, AG,  (A BBsJ rcNU=(d URRnU=(d URRn[R"[ R "U5[ R "U5[U55R5RS5nU$)aT Compute the exponential sigma schedule. Implementation closely follows k-diffusion: https://github.com/crowsonkb/k-diffusion/blob/6ab5146d4a5ef63901326489f31f1d8e7dd36b48/k_diffusion/sampling.py#L26 Args: ramp (`torch.Tensor`): A tensor of values representing the interpolation positions. sigma_min (`float`, *optional*): Minimum sigma value. If `None`, uses `self.config.sigma_min`. sigma_max (`float`, *optional*): Maximum sigma value. If `None`, uses `self.config.sigma_max`. Returns: `torch.Tensor`: The computed exponential sigma schedule. r) rGr!r"rr}mathrmlenexpflip)rNrr!r"rJs rrC-EDMEulerScheduler._compute_exponential_sigmasasn,6!6!6 6!6!6  3TXXi5H#d)TXXZ__`ab rschedule_timestepscUc URnX!:HR5n[U5S:aSOSnX4R5$)a Find the index of a given timestep in the timestep schedule. Args: timestep (`float` or `torch.Tensor`): The timestep value to find in the schedule. schedule_timesteps (`torch.Tensor`, *optional*): The timestep schedule to search in. If `None`, uses `self.timesteps`. Returns: `int`: The index of the timestep in the schedule. For the very first step, returns the second index if multiple matches exist to avoid skipping a sigma when starting mid-schedule (e.g., for image-to-image). r r)rFnonzeroritem)rNrwrindicesposs rindex_for_timestep$EDMEulerScheduler.index_for_timestep}sJ"  %!% %1::< w'>FVF\F\]  " " ' '5 0U5L5LY5W5W!%!2!23C3J3JRWR_R_!2!` ! %5%<%G:@*,*,*, *,  78 *, ! *,!!:;*,*,##67*, *,*,X5%55 C   !S ! !(3(t(%,,uu||?SX]XdXd$ (<&! !ll!u||# !  !F  @TY^YeYe4$(%)48 /, /,ell"/, tE{*T1 /,j#'"& ll4<4<   B#'"& ll4<4<   :Y]# ,#BG,,QUBU# #> 1)= 1$ 1(e ,0 48jklljk%,,&jk jk  jk  jkjkjk??T)jkjk$llT1jk !5 (jkZ.,,.||.<< .  .` EELL,@ UU\\EY //rr)r dataclassesrtypingrrconfiguration_utilsrrutilsrr utils.torch_utilsr scheduling_utilsr get_loggerrrrrrrrrs\ ! A',,   H % 5j5 5"g/ g/r