3j/SSKJr SSKrSSKrSSKJrJr SSKJ r SSK J r J r SSK Jr \ R"\5r\"SS \ 55r"S S \ \5rg) ) dataclassN) ConfigMixinregister_to_config)SchedulerMixin) BaseOutputlogging) randn_tensorc`\rSrSr%Sr\R \S'Sr\R S-\S'Sr g)SCMSchedulerOutput aM 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__r]/home/wildlama/miniconda3/lib/python3.13/site-packages/diffusers/schedulers/scheduling_scm.pyr r s' 04%,,-4rr c\rSrSrSrSr\SS\S\S\ 4Sjj5r \ S5r \ S 5r S S \4S jjrS!S \S\R S\\R"-S\ S\ 4 SjjrS\ \R -SS 4SjrS"S\ \R -S\R S -S\4SjjrS#S\R*S\ S\R*S\R,S\S\\-4 SjjrSrSrg )$ SCMScheduler3a `SCMScheduler` extends the denoising procedure introduced in denoising diffusion probabilistic models (DDPMs) with non-Markovian guidance. 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: num_train_timesteps (`int`, defaults to 1000): The number of diffusion steps to train the model. prediction_type (`str`, defaults to `trigflow`): Prediction type of the scheduler function. Currently only supports "trigflow". sigma_data (`float`, defaults to 0.5): The standard deviation of the noise added during multi-step inference. num_train_timestepsprediction_type sigma_datacSUlSUl[R"[R "SU5SSS2R 5R[R55Ul SUl SUl g)a Initialize the SCM scheduler. Args: num_train_timesteps (`int`, defaults to 1000): The number of diffusion steps to train the model. prediction_type (`str`, defaults to `trigflow`): Prediction type of the scheduler function. Currently only supports "trigflow". sigma_data (`float`, defaults to 0.5): The standard deviation of the noise added during multi-step inference. g?Nr) init_noise_sigmanum_inference_stepsr from_numpynparangecopyastypeint64 timesteps _step_index _begin_index)selfr r!r"s r__init__SCMScheduler.__init__Esh&!$$( ))"))A7J*KDbD*Q*V*V*X*_*_`b`h`h*ij rcUR$N)r.r0s r step_indexSCMScheduler.step_indexasrcUR$r4r/r5s r begin_indexSCMScheduler.begin_indexes   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. Nr9)r0r:s rset_begin_indexSCMScheduler.set_begin_indexjs (rNr&r-device max_timestepsintermediate_timestepsc pXRR:a=[SUSURRSURRS35eUb[U5US-:wa [S5eUbUb [S5eUcUc [S 5eUbUS :wa [S 5eXlUb[ U[ 5(a([R"X#S 9R5Ul O[ U[R5(a%URU5R5Ul Or[S [U535eUb+[R"XES/US 9R5Ul O-[R"USUS-US 9R5Ul SUlSUlg)a] Sets the discrete timesteps used for the diffusion chain (to be run before inference). Args: num_inference_steps (`int`): The number of diffusion steps used when generating samples with a pre-trained model. timesteps (`torch.Tensor`, *optional*): Custom timesteps to use for the denoising process. max_timesteps (`float`, defaults to 1.57080): The maximum timestep value used in the SCM scheduler. intermediate_timesteps (`float`, *optional*, defaults to 1.3): The intermediate timestep value used in SCM scheduler (only used when num_inference_steps=2). z`num_inference_steps`: z6 cannot be larger than `self.config.train_timesteps`: zG as the unet model trained with this scheduler can only handle maximal z timesteps.NrzWIf providing custom timesteps, `timesteps` must be of length `num_inference_steps + 1`.zFIf providing custom timesteps, `max_timesteps` should not be provided.z5Should provide either `timesteps` or `max_timesteps`.rzNIntermediate timesteps for SCM is not supported when num_inference_steps != 2.)r?zUnsupported timesteps type: r)configr ValueErrorlenr& isinstancelistrtensorfloatr-rtotypelinspacer.r/)r0r&r-r?r@rAs r set_timestepsSCMScheduler.set_timestepsts* !@!@ @)*=)>?KK3345 KK;;ef f  !6TU U ! -2E2Jmn n#6  )T**!&i!G!M!M!OIu||44!*f!5!;!;!= #?Y?P!QRR # /"\\=RS*T]cdjjlDN#^^M1>QTU>U^dekkmDN rtimestepreturncURc[[U[R5(a%UR UR R 5nURU5UlgURUlg)z Initialize the step index for the scheduler based on the given timestep. Args: timestep (`float` or `torch.Tensor`): The current timestep to initialize the step index from. N) r:rFrrrJr-r?index_for_timestepr.r/)r0rOs r_init_step_indexSCMScheduler._init_step_indexsZ    #(ELL11#;;t~~'<'<=#66x@D #00D rschedule_timestepscUc URnX!:HR5n[U5S:aSOSnX4R5$)a Find the index of a given timestep in the timestep schedule. Args: timestep (`float` or `torch.Tensor`): The timestep value to find in the schedule. schedule_timesteps (`torch.Tensor`, *optional*): The timestep schedule to search in. If `None`, uses `self.timesteps`. Returns: `int`: The index of the timestep in the schedule. For the very first step, returns the second index if multiple matches exist to avoid skipping a sigma when starting mid-schedule (e.g., for image-to-image). rr)r-nonzerorEitem)r0rOrUindicesposs rrRSCMScheduler.index_for_timestepsJ"  %!% %1::< wN=OPQ Q t~~  "\// 8K8KW`a++(()  ))A,0599Q<%3GGK!K A) )!kXXrc.URR$r4)rCr r5s r__len__SCMScheduler.__len__s{{...r)r/r.r%r&r-)irag?)r)NNgH.!?g?r4)NT)rrrrrorderrintstrrIr1propertyr6r:r=rrr?rMrSrR FloatTensor Generatorboolr tuplerjrmrrrrrr3s  E$() ! !! !!6  !!(3(#'%)&(+ 8! 8!<<8!ell" 8!  8! !& 8!v 1)= 1$ 1"Y]# ,#BG,,QUBU# #F&* >Y''>Y>Y!! >Y ?? >Y  >Y e #>Y@/rr) dataclassesrnumpyr(rconfiguration_utilsrrschedulers.scheduling_utilsrutilsrr utils.torch_utilsr get_loggerrloggerr rrrrrsY$" A8',   H % 55 5"k/>;k/r