+hlSSKJrJrJrJrJr SSKrSSKJrJ r SSK J r SSK J r Jr SSKJr SSKJrJrJrJr SS KJr S S KJrJr \"5(a SSKJs Jr S rOS r\R@"\!5r"Sr#SSjr$"SS\\ 5r%g))CallableDictListOptionalUnionN)T5EncoderModel T5Tokenizer)StableDiffusionLoraLoaderMixin)Kandinsky3UNetVQModel) DDPMScheduler) deprecateis_torch_xla_availableloggingreplace_example_docstring) randn_tensor)DiffusionPipelineImagePipelineOutputTFa Examples: ```py >>> from diffusers import AutoPipelineForText2Image >>> import torch >>> pipe = AutoPipelineForText2Image.from_pretrained( ... "kandinsky-community/kandinsky-3", variant="fp16", torch_dtype=torch.float16 ... ) >>> pipe.enable_model_cpu_offload() >>> prompt = "A photograph of the inside of a subway train. There are raccoons sitting on the seats. One of them is reading a newspaper. The window shows the city in the background." >>> generator = torch.Generator(device="cpu").manual_seed(0) >>> image = pipe(prompt, num_inference_steps=25, generator=generator).images[0] ``` cnXS--nXS--S:waUS- nXS--nXS--S:waUS- nX2-XB-4$)Nrr)heightwidth scale_factor new_height new_widths l/home/james-whalen/.local/lib/python3.13/site-packages/diffusers/pipelines/kandinsky3/pipeline_kandinsky3.pydownscale_height_and_widthr 1s[?*J a1$a q(I Q!#Q  $i&> >>c%^\rSrSrSr/SQrS\S\S\S\ S\ 4 U4S jjr S r \ R"5S+S\\ R S\\ R S\\ R S\\ R 4Sjj5rSrS,Sjr\S5r\S5r\S5r\ R"5\"\5S SSS S SSS S S S S SS S S S/4S\\\\4S\S\S \\\\\4S!\\S"\\S#\\S$\\\ R>\\ R>4S\\ R S\\ R S\\ R S\\ R S%\\S&\ S'\\!\\\"/S 4S(\\4 S)jj55r#S*r$U=r%$)-Kandinsky3Pipeline;ztext_encoder->unet->movq)latents prompt_embedsnegative_prompt_embedsnegative_attention_maskattention_mask tokenizer text_encoderunet schedulermovqcD>[TU]5 URXX4US9 g)N)r*r+r,r-r.)super__init__register_modules)selfr*r+r,r-r. __class__s rr1Kandinsky3Pipeline.__init__Es+  ae  r!cU(aX[R"XS:H5XS:H'URS5R5S-nUSS2SU24nUSS2SU24nX4$)Nrr)torch zeros_likesummax)r3 embeddingsr) cut_contextmax_seq_lengths rprocess_embeds!Kandinsky3Pipeline.process_embedsSss .3.>.>z\]J]?^._J* ++//3779A=N#A$67J+A,>?N))r!TrNr&r'r)r(c Ub>Ub;[U5[U5La$[S[U5S[U5S35eUc URnUb[U[5(aSn O3Ub![U[ 5(a [ U5n OURSn Sn UcURUSU S S S 9n U RRU5nU RRU5n URUU S 9nUSnURXiU5upiXiRS 5-nURbURRnOSnURXS9nURunnnUR!SUS5nUR#UU-US5nU R!US5n U(Ga+UGc'UcS/U -nOK[U[5(aU/nO2U [ U5:wa![%SUS[ U5SUSU S3 5eUnUbURUSSS S S S9nURRU5nURRU5n URUU S 9nUSnUSS2SURS24nU SS2SURS24n XzRS 5-nO,[&R("U5n[&R("U 5n U(asURSnURXS9nURUR:wa:UR!SUS5nUR#X-US5nU R!US5n OSnSn XgX4$)a Encodes the prompt into text encoder hidden states. Args: prompt (`str` or `List[str]`, *optional*): prompt to be encoded device: (`torch.device`, *optional*): torch device to place the resulting embeddings on num_images_per_prompt (`int`, *optional*, defaults to 1): number of images that should be generated per prompt do_classifier_free_guidance (`bool`, *optional*, defaults to `True`): whether to use classifier free guidance or not negative_prompt (`str` or `List[str]`, *optional*): The prompt or prompts not to guide the image generation. If not defined, one has to pass `negative_prompt_embeds`. instead. If not defined, one has to pass `negative_prompt_embeds`. instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`). prompt_embeds (`torch.Tensor`, *optional*): Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not provided, text embeddings will be generated from `prompt` input argument. negative_prompt_embeds (`torch.Tensor`, *optional*): Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input argument. attention_mask (`torch.Tensor`, *optional*): Pre-generated attention mask. Must provide if passing `prompt_embeds` directly. negative_attention_mask (`torch.Tensor`, *optional*): Pre-generated negative attention mask. Must provide if passing `negative_prompt_embeds` directly. Nz?`negative_prompt` should be the same type to `prompt`, but got z != .rr max_lengthTpt)paddingrD truncationreturn_tensors)r)r)dtypedevicer7z`negative_prompt`: z has batch size z, but `prompt`: zT. Please make sure that passed `negative_prompt` matches the batch size of `prompt`.)rFrDrGreturn_attention_maskrH)type TypeError_execution_device isinstancestrlistlenshaper* input_idstor)r+r? unsqueezerIrepeatview ValueErrorr8r9)r3promptdo_classifier_free_guidancenum_images_per_promptrJnegative_promptr&r' _cut_contextr)r( batch_sizerD text_inputstext_input_idsrIbs_embedseq_len_ uncond_tokens uncond_inputs r encode_prompt Kandinsky3Pipeline.encode_prompt[sT  /"=F|4#88UVZ[jVkUlmV ~Q( >++F  *VS"9"9J  Jvt$<$<VJ&,,Q/J  ..$%# )K)2255f=N(77::6BN ---.M*!,M,0,?,? _k,l )M),D,DQ,GGM    (%%++EE%((u(D ,22'1%,,Q0EqI %**86K+KWVXY '../DaH &+A+I&!#z 1 OS11!0 1 s?33 )/)::J3K_J`ax/ |<33 !0 *#~~!("#*.#' .  ".!7!7!:!:6!B*6*E*E*H*H*P')-):):"#:*;*&*@)B&)?C[]EXEXYZE[C[@[)\&*A!E]}GZGZ[\G]E]B]*^')?BcBcdeBf)f&*/)9)9-)H&*/*:*:>*J' &,2215G%;%>%>U%>%Z "%++}/B/BB)?)F)FqJ_ab)c&)?)D)DZEgiprt)u&*A*H*HI^`a*b' &* "&* #n]]r!cUc [XX2S9nO# UHoTR;v M g7fN_callback_tensor_inputs.0kr3s r 2Kandinsky3Pipeline.check_inputs.. F 7Y!-- -7Y2`callback_on_step_end_tensor_inputs` has to be in , but found zCannot forward both `prompt`: z and `prompt_embeds`: z2. Please make sure to only forward one of the two.zeProvide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined.z2`prompt` has to be of type `str` or `list` but is z'Cannot forward both `negative_prompt`: z and `negative_prompt_embeds`: zu`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but got: `prompt_embeds` z != `negative_prompt_embeds` zLPlease provide `negative_attention_mask` along with `negative_prompt_embeds`rz`negative_prompt_embeds` and `negative_attention_mask` must have the same batch_size and token length when passed directly, but got: `negative_prompt_embeds` z != `negative_attention_mask` z:Please provide `attention_mask` along with `prompt_embeds`z`prompt_embeds` and `attention_mask` must have the same batch_size and token length when passed directly, but got: `prompt_embeds` z != `attention_mask` ) rPintrZrMallrsrQrRrT) r3r[callback_stepsr^r&r'"callback_on_step_end_tensor_inputsr)r(rvs ` r check_inputsKandinsky3Pipeline.check_inputss  %z.#/N/NR`deReGGWX(),  . 9#F 7YF C C DTEaEaDbbn|^pH|^vwkokGkGbGpq|^pHoIJ   -";08N}o^00 ^ 5w  FC)@)@TZ\`IaIaQRVW]R^Q_`a a  &+A+M9/9JK*++]_   $)?)K""&<&B&BB --:-@-@,AB.445Q8 " -2I2Qkl l ! -2I2U%++BQ/3J3P3PP 66L6R6RSUTU6V5WX/556a9  $)?YZ Z  $)C""2A&.*>*>> --:-@-@!-D,EF&,,-Q0?*D $UpHs /HHcUR$rq_guidance_scaler3s rguidance_scale!Kandinsky3Pipeline.guidance_scaleBs###r!c URS:$)Nrrrs rr\.Kandinsky3Pipeline.do_classifier_free_guidanceFs##a''r!cUR$rq)_num_timestepsrs r num_timesteps Kandinsky3Pipeline.num_timestepsJs"""r!g@ipilr%r[num_inference_stepsrr^r]rrrk output_type return_dictcallback_on_step_endrc ^URSS5nURSS5nUb [SSS5 Ub [SSS5 UbX[U4SjU55(d>[STRS UVs/sHnUTR;dMUPM sn35eS nTR nTR UUUU U UU U 5 UTlUb[U[5(aS nO3Ub![U[5(a [U5nOU RS nTRUTRUUUU U UU U S 9 uppTR(a<[R "X/5n [R "X/5R#5n TR$R'UUS9 TR$R(n[+XgS5upgTR-UU-SXg4U R.UUUTR$5n[1TS5(a'TR2bTR2R55 [U5UTR$R6-- n[U5TlTR;US9n[=U5GHunnTR(a[R "U/S-5OUnTR?UUU U SS9S nTR(a"URAS5un n!US-U!-UU -- nTR$RCUUUUS9RDnUb0n"UHn[G5UU"U'M U"TUUU"5n#U#RSU5nU#RSU 5n U#RSU 5n U#RSU 5n U#RSU 5n U[U5S - :Xd)US -U:a`US -TR$R6-S :Xa@URI5 Ub-UU-S :Xa$U[KTR$SS 5-n$U"U$UU5 [L(dGM[NRP"5 GM U S;a[SU 35eU S :XdTRRRUUS S!9S"n%U S#;aXU%S$-S$-n%U%RWS S 5n%U%RY5R[S SS%S 5R]5R_5n%U S&:XaTRaU%5n%OUn%TRc5 U(d U%4sSSS5 $[eU%S'9sSSS5 $s snf!,(df  g=f)(u Function invoked when calling the pipeline for generation. Args: prompt (`str` or `List[str]`, *optional*): The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`. instead. num_inference_steps (`int`, *optional*, defaults to 25): The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. timesteps (`List[int]`, *optional*): Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps` timesteps are used. Must be in descending order. guidance_scale (`float`, *optional*, defaults to 3.0): Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2. of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`, usually at the expense of lower image quality. negative_prompt (`str` or `List[str]`, *optional*): The prompt or prompts not to guide the image generation. If not defined, one has to pass `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`). num_images_per_prompt (`int`, *optional*, defaults to 1): The number of images to generate per prompt. height (`int`, *optional*, defaults to self.unet.config.sample_size): The height in pixels of the generated image. width (`int`, *optional*, defaults to self.unet.config.sample_size): The width in pixels of the generated image. eta (`float`, *optional*, defaults to 0.0): Corresponds to parameter eta (η) in the DDIM paper: https://huggingface.co/papers/2010.02502. Only applies to [`schedulers.DDIMScheduler`], will be ignored for others. generator (`torch.Generator` or `List[torch.Generator]`, *optional*): One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make generation deterministic. prompt_embeds (`torch.Tensor`, *optional*): Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not provided, text embeddings will be generated from `prompt` input argument. negative_prompt_embeds (`torch.Tensor`, *optional*): Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input argument. attention_mask (`torch.Tensor`, *optional*): Pre-generated attention mask. Must provide if passing `prompt_embeds` directly. negative_attention_mask (`torch.Tensor`, *optional*): Pre-generated negative attention mask. Must provide if passing `negative_prompt_embeds` directly. output_type (`str`, *optional*, defaults to `"pil"`): The output format of the generate image. Choose between [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`. return_dict (`bool`, *optional*, defaults to `True`): Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple. callback (`Callable`, *optional*): A function that will be called every `callback_steps` steps during inference. The function will be called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`. callback_steps (`int`, *optional*, defaults to 1): The frequency at which the `callback` function will be called. If not specified, the callback will be called at every step. clean_caption (`bool`, *optional*, defaults to `True`): Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to be installed. If the dependencies are not installed, the embeddings will be created from the raw prompt. cross_attention_kwargs (`dict`, *optional*): A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under `self.processor` in [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py). Examples: Returns: [`~pipelines.ImagePipelineOutput`] or `tuple` callbackNrz1.0.0zhPassing `callback` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`znPassing `callback_steps` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`c3@># UHoTR;v M g7frqrrrts rrw.Kandinsky3Pipeline.__call__..ryrzr{r|Trr)r]rJr^r&r'r_r)r()rJtext_encoder_offload_hook)totalrF)encoder_hidden_statesencoder_attention_maskrg?)rkr%r&r'r)r(order)rEnprlatentzSOnly the output types `pt`, `pil`, `np` and `latent` are supported not output_type=r)force_not_quantizesample)rrg?r r)images)3poprr~rZrsrOrrrPrQrRrSrTrhr\r8catboolr- set_timesteps timestepsr rmrIhasattrroffloadrr progress_bar enumerater,chunkstep prev_samplelocalsupdategetattr XLA_AVAILABLExm mark_stepr.decodeclampcpupermutefloatnumpy numpy_to_pilmaybe_free_model_hooksr)&r3r[rrr^r]rrrkr&r'r)r(rrr%rrkwargsrrrvr=rJr`rnum_warmup_stepsritlatent_model_input noise_prednoise_pred_uncondnoise_pred_textcallback_kwargscallback_outputsstep_idximages&` r__call__Kandinsky3Pipeline.__call__Ns@::j$/$4d;   z   %  A  . 9#F 7YF C C DTEaEaDbbn|^pH|^vwbckokGkGbGpq|^pHoIJ  ''      " .  #  .  *VS"9"9J  Jvt$<$<VJ&,,Q/JZ^YkYk   , ,"7+'#9$)$;Zl Z V ~  + +!II'=&MNM"YY(?'PQVVXN $$%8$HNN,, 36!D && / /F B       NN   44 5 5$:X:X:d  * * 2 2 4y>,?$..BVBV,VV!)n   %8  9\!),1AEAaAaUYYy1}%=gn""YY&*7+9 % '   339C9I9I!9L6%"03"6/!IN]nLn!nJ..--' . + (3&(O?-3Xa[*@';D!Q'X$.229gFG$4$8$8-$XM-=-A-ABZ\r-s*%5%9%9:JN%[N.>.B.BC\^u.v+I**A9I/IqSTuX\XfXfXlXlNlpqNq '')+N0Ba0G#$(K#K 1g6 =LLN[-`"?? ijuivw(* ((T(J8T-/!CK#-E!KK1-E!IIK//1a;AACIIKE%' --e4E  ' ' )xM: 9P'e4Q: 9EpHD: 9s&:T1T18F6T63C"T6T66 U)rr) TrNNNNFNN)NNNNNN)&__name__ __module__ __qualname____firstlineno__model_cpu_offload_seqrsr rr rr r1r?r8no_gradrTensorrhrmrpropertyrr\rrEXAMPLE_DOC_STRINGrrQrr}r Generatorrrrr__static_attributes__ __classcell__)r4s@rr#r#;s6    %    !    * ]]_%)049=15:>S^  -S^!) 6S^!.S^"*%,,!7S^S^j #+/ $DL$$((## ]]_12)-#% #;?/0 $#MQ049=15:>%* KO9B %|5c49n%|5!|5 |5 "%T#Y"78 |5 (} |5 |5}|5E%//43H"HIJ|5  -|5!) 6|5!.|5"*%,,!7|5c]|5|5"'xc40@$0F'GH#|5$-1I%|53|5r!r#)r)&typingrrrrrr8 transformersrr loadersr modelsr r schedulersrutilsrrrrutils.torch_utilsrpipeline_utilsrrtorch_xla.core.xla_modelcore xla_modelrr get_loggerrloggerrr r#rr!rrsv88 45-' .C))MM   H %(?Q5*,JQ5r!