3ja4SSKJr SSKrSSKJr SSKJrJr SSK J r SSK J r J r JrJrJrJr \ R$"\5r\R*R"SS 55r\"S S \55r"S S \\5rg)) dataclassN) ConfigMixinregister_to_config)logging)CommonSchedulerStateFlaxKarrasDiffusionSchedulersFlaxSchedulerMixinFlaxSchedulerOutputadd_noise_commonget_velocity_commonc \rSrSr%\\S'\R\S'\R\S'\R\S'Sr\ \S'\ S\S\RS\RS\R4Sj5r S r g) DDIMSchedulerState&commonfinal_alpha_cumprodinit_noise_sigma timestepsNnum_inference_stepscU"UUUUS9$)Nrrrr)clsrrrrs c/home/wildlama/miniconda3/lib/python3.13/site-packages/diffusers/schedulers/scheduling_ddim_flax.pycreateDDIMSchedulerState.create0s 3-   r) __name__ __module__ __qualname____firstlineno__r __annotations__jnpndarrayrint classmethodr__static_attributes__rrrrr&sw  $kk!{{##  $  ![[  ++  ;;    rrc \rSrSr%\\S'Srg)FlaxDDIMSchedulerOutput@staterN)rr r!r"rr#r(rrrr*r*@s rr*c\rSrSr%Sr\VVs/sHoR PM snnr\R\ S'\ S5r \ SSSSS S S S S S \R4 S\S\S\S\S\R$S -S\S\S\S\S\S\R4Sjj5rS/S\S -S\4SjjrS/S\S\R$S\S -S\R$4SjjrS0S\S \S!\S\4S"jjrS\4S#jrS1S\S$\R$S\S\R$S%\S&\S\\-4S'jjrS\S(\R$S)\R$S*\R$S\R$4 S+jrS\S\R$S)\R$S*\R$S\R$4 S,jrS-r S.r!g s snnf)2FlaxDDIMSchedulerEa 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`. [`SchedulerMixin`] provides general loading and saving functionality via the [`SchedulerMixin.save_pretrained`] and [`~SchedulerMixin.from_pretrained`] functions. For more details, see the original paper: https://huggingface.co/papers/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 for numerical stability. The clip range is determined by `clip_sample_range`. clip_sample_range (`float`, default `1.0`): the maximum magnitude for sample clipping. Valid only when `clip_sample=True`. set_alpha_to_one (`bool`, default `True`): each diffusion step uses the value of alphas product at that step and at the previous one. For the final step there is no previous alpha. When this option is `True` the previous alpha product is fixed to `1`, otherwise it uses the value of alpha at step 0. steps_offset (`int`, default `0`): An offset added to the inference steps, as required by some model families. prediction_type (`str`, default `epsilon`): indicates whether the model predicts the noise (epsilon), or the samples. One of `epsilon`, `sample`. `v-prediction` is not supported for this scheduler. dtype (`jnp.dtype`, *optional*, defaults to `jnp.float32`): the `dtype` used for params and computation. dtypecg)NTrselfs r has_stateFlaxDDIMScheduler.has_statepsrig-C6?g{Gz?linearNT?repsilonnum_train_timesteps beta_startbeta_end beta_schedule trained_betas clip_sampleclip_sample_rangeset_alpha_to_one steps_offsetprediction_typec :[RS5 Xlg)NzFlax classes are deprecated and will be removed in Diffusers v1.0.0. We recommend migrating to PyTorch classes or pinning your version of Diffusers.)loggerwarningr0) r3r9r:r;r<r=r>r?r@rArBr0s r__init__FlaxDDIMScheduler.__init__ts  [  rrreturncUc[R"U5nURR(a[R "SUR S9OURSn[R "SUR S9n[R"SURR5R5SSS2n[RUUUUS9$)Nr7r0rr) r rconfigr@r$arrayr0alphas_cumprodaranger9roundr)r3rrrrs r create_stateFlaxDDIMScheduler.create_states >)006F15 0L0LCIIc ,RXRgRghiRj  99S ;JJq$++"A"ABHHJ4R4P !(( 3- )  rr,sampletimestepcU$)z Args: state (`PNDMSchedulerState`): the `FlaxPNDMScheduler` state data class instance. sample (`jnp.ndarray`): input sample timestep (`int`, optional): current timestep Returns: `jnp.ndarray`: scaled input sample r)r3r,rSrTs rscale_model_input#FlaxDDIMScheduler.scale_model_inputs  rrshapecURRU-n[R"SU5U-R 5SSS2URR -nUR UUS9$)aM 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. rNrK)rr)rLr9r$rOrPrAreplace)r3r,rrX step_ratiors r set_timestepsFlaxDDIMScheduler.set_timestepssr[[448KK ZZ#67*DKKMdPRdSVZVaVaVnVnn }} 3  rcURRUn[R"US:URRUUR5nSU- nSU- nXv- SXE- - -nU$)Nrr)rrNr$wherer) r3r,rT prev_timestep alpha_prod_talpha_prod_t_prev beta_prod_tbeta_prod_t_prevvariances r _get_varianceFlaxDDIMScheduler._get_variancesz||228< II Q  LL ' ' 6  % %  ,& 00$2q<;[7[\r model_outputeta return_dictcURc [S5eX0RRUR-- nURR nUR n Xn [R"US:XU 5n SU - n URRS:XaXLS-U-- U S-- n UnOURRS:XaUn XJS-U -- U S-- nO`URRS:Xa#U S-U-U S-U-- n U S-U-U S-U--nO#[SURRS 35eURR(a;U RURR*URR5n URXU5nX_S--nSU - US -- S-U-nU S-U -U-nU(dUU4$[UUS 9$) a/ Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion process from the learned model outputs (most often the predicted noise). Args: state (`DDIMSchedulerState`): the `FlaxDDIMScheduler` state data class instance. model_output (`jnp.ndarray`): direct output from learned diffusion model. timestep (`int`): current discrete timestep in the diffusion chain. sample (`jnp.ndarray`): current instance of sample being created by diffusion process. return_dict (`bool`): option for returning tuple rather than FlaxDDIMSchedulerOutput class Returns: [`FlaxDDIMSchedulerOutput`] or `tuple`: [`FlaxDDIMSchedulerOutput`] if `return_dict` is True, otherwise a `tuple`. When returning a tuple, the first element is the sample tensor. zaNumber of inference steps is 'None', you need to run 'set_timesteps' after creating the schedulerrrr8g?rS v_predictionzprediction_type given as z6 must be one of `epsilon`, `sample`, or `v_prediction`r) prev_sampler,)r ValueErrorrLr9rrNrr$r_rBr>clipr?rfr*)r3r,rhrTrSrirjr`rNrrarbrcpred_original_sample pred_epsilonre std_dev_tpred_sample_directionrms rstepFlaxDDIMScheduler.steps'4  $ $ ,s  !;;#B#BeF_F_#__ 44#77&/ IImq&8.:WYlm,&  ;; & &) 3$*S-AL-P$PT`ehTi#i 'L [[ ( (H 4#/ "c%:=Q%QQU`ehUiiL [[ ( (N :$0#$5#?;PSCSWcBc#c (#-=cAQU[@[+DKK,G,G+HI""  ;; " "#7#<#<... 0M0M$ %%e}Es++ "#%6!6A!E3 OR^ ^(C03GGJ__ ' '&;eLLroriginal_samplesnoiserc0[URX#U5$N)r r)r3r,rvrwrs r add_noiseFlaxDDIMScheduler.add_noise3s  .>yQQrc0[URX#U5$ry)rr)r3r,rSrwrs r get_velocityFlaxDDIMScheduler.get_velocity<s#5<< JJrc.URR$ry)rLr9r2s r__len__FlaxDDIMScheduler.__len__Es{{...rrJry)r)gT)"rr r!r"__doc__r name _compatiblesr$r0r#propertyr4rfloat32r&floatstrr%boolrFr rrQrVtupler\rfr*rtrzr}rr().0es00rr.r.Es$L%BB$AqFF$ABL 99 $("%,0 #&!%(;;     {{T) !yy( #7$#> J\ 8 $ ! *   $SU ' >A JO  , #5 ( [M![Mkk[M [M  [M  [M[M !5 ([MzR!R++R{{ R ;; R  RK!K K{{ K ;; K  K/sCsE;r.) dataclassesrflax jax.numpynumpyr$configuration_utilsrrutilsrscheduling_utils_flaxr r r r r r get_loggerrrDstructrr*r.rrrrs$" A   H %   2 1 A/*KA/r