+hKSSKrSSKJrJrJr SSKrSSKrSSKJ r SSK J r J r JrJr SSKJrJr SSKJr SSKJrJr SS KJr S S KJrJrJr S S KJr \"5(a SSK J!s J"r# Sr$OSr$\RJ"\&5r'"SS\\5r(g)N)ListOptionalUnion) functional)CLIPImageProcessorCLIPTextModelWithProjection CLIPTokenizerCLIPVisionModelWithProjection)UNet2DConditionModel UNet2DModel)UnCLIPScheduler)is_torch_xla_availablelogging) randn_tensor)DeprecatedPipelineMixinDiffusionPipelineImagePipelineOutput)UnCLIPTextProjModelTFc`^\rSrSr%SrSr\\S'\\S'\ \S'\ \S'\ \S'\ \S '\ \S '\ \S '\\S '\\S 'SrS\S\ S\ S\S\ S \ S \ S \ S \S \4U4SjjrSrSrS S\\R*4Sjjr\R."5S!S\\\R4R4\\R4R4\R*4S\S\S\S\\R:S\\R*S\\R*S\\R*S\S\\S\ 4Sjj5r!Sr"U=r#$)"UnCLIPImageVariationPipeline.aR Pipeline to generate image variations from an input image using UnCLIP. This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, running on a particular device, etc.). Args: text_encoder ([`~transformers.CLIPTextModelWithProjection`]): Frozen text-encoder. tokenizer ([`~transformers.CLIPTokenizer`]): A `CLIPTokenizer` to tokenize text. feature_extractor ([`~transformers.CLIPImageProcessor`]): Model that extracts features from generated images to be used as inputs for the `image_encoder`. image_encoder ([`~transformers.CLIPVisionModelWithProjection`]): Frozen CLIP image-encoder ([clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14)). text_proj ([`UnCLIPTextProjModel`]): Utility class to prepare and combine the embeddings before they are passed to the decoder. decoder ([`UNet2DConditionModel`]): The decoder to invert the image embedding into an image. super_res_first ([`UNet2DModel`]): Super resolution UNet. Used in all but the last step of the super resolution diffusion process. super_res_last ([`UNet2DModel`]): Super resolution UNet. Used in the last step of the super resolution diffusion process. decoder_scheduler ([`UnCLIPScheduler`]): Scheduler used in the decoder denoising process (a modified [`DDPMScheduler`]). super_res_scheduler ([`UnCLIPScheduler`]): Scheduler used in the super resolution denoising process (a modified [`DDPMScheduler`]). z0.33.1decoder text_proj text_encoder tokenizerfeature_extractor image_encodersuper_res_firstsuper_res_lastdecoder_schedulersuper_res_schedulerzPtext_encoder->image_encoder->text_proj->decoder->super_res_first->super_res_lastc R>[T U]5 URUUUUUUUUU U S9 g)N) rrrrrr r!r"r#r$)super__init__register_modules) selfrrrrrr r!r"r#r$ __class__s t/home/james-whalen/.local/lib/python3.13/site-packages/diffusers/pipelines/unclip/pipeline_unclip_image_variation.pyr'%UnCLIPImageVariationPipeline.__init__ZsA  %/'+)/ 3  cUc [XX2S9nO) isinstancelistlenrmodel_max_length input_idsattention_maskboolr4r text_embedslast_hidden_staterepeat_interleaver2repeatviewtorchcat)r)promptr0num_images_per_promptdo_classifier_free_guidance batch_size text_inputstext_input_ids text_masktext_encoder_output prompt_embedstext_encoder_hidden_states uncond_tokensr; uncond_inputuncond_text_mask*negative_prompt_embeds_text_encoder_outputnegative_prompt_embeds!uncond_text_encoder_hidden_statesseq_lens r+_encode_prompt+UnCLIPImageVariationPipeline._encode_promptsi$.vt$<$>$%# *L ,::??ADDVL 9=9J9J)r0r1rr?) nextr parametersr1rDrPTensorr pixel_valuesr4 image_embedsrM)r)imager0rSrer1s r+ _encode_image*UnCLIPImageVariationPipeline._encode_imagesT''2245;;  #eU\\22..eD.Q^^HHFH8E#11%8EE +==>SYZ=[r-rmrSdecoder_num_inference_stepssuper_res_num_inference_stepsr/decoder_latentssuper_res_latentsdecoder_guidance_scale output_type return_dictc P Ub][U[RR5(aSn O@[U[5(a [ U5n OUR Sn OUR Sn S/U -n UR nX-n U S:nURXX/5unnnURXX(5nURUUUUS9unnURS:XakUR[R5n[R"UURRS4SS9nUR[R 5nO,[R"UURRS4S S9nUR"R%X>S 9 UR"R&nUR(R*R,nUR(R*R.nUR(R*R.nUc.UR1U UUU4UR2UUUUR"5n[5UR7U55GH unnU(a[R8"U/S -5OUnUR)UUUUUS 9R:nU(avUR=S 5unnUR?UR SSS 9unnUR?UR SSS 9unn UU UU- --n[R8"UU /SS 9nUS-UR S:XaSn!OUUS-n!UR"RAUUUU!US9RBnGM URESS5nUn"URFR%XNS 9 URFR&n#URHR*R,S -n$URHR*R.nURHR*R.nUc.UR1U U$UU4U"R2UUUURF5nURS:Xa[RJ"U"UU/S9n%OV0n&S[LRN"[RJ5RP;aS U&S'[RJ"U"4UU/SSS.U&D6n%[5UR7U#55HunnUU#R SS- :Xa URRn'O URHn'[R8"UU%/SS 9nU'"UUS9R:nUS-U#R S:XaSn!OU#US-n!URFRAUUUU!US9RBn[T(dM[VRX"5 M UnUR[5 US-S-nURESS5nUR]5R_SS SS5Ra5Rc5nU S:XaUReU5nU (dU4$[gUS9$)a The call function to the pipeline for generation. Args: image (`PIL.Image.Image` or `List[PIL.Image.Image]` or `torch.Tensor`): `Image` or tensor representing an image batch to be used as the starting point. If you provide a tensor, it needs to be compatible with the [`CLIPImageProcessor`] [configuration](https://huggingface.co/fusing/karlo-image-variations-diffusers/blob/main/feature_extractor/preprocessor_config.json). Can be left as `None` only when `image_embeddings` are passed. num_images_per_prompt (`int`, *optional*, defaults to 1): The number of images to generate per prompt. decoder_num_inference_steps (`int`, *optional*, defaults to 25): The number of denoising steps for the decoder. More denoising steps usually lead to a higher quality image at the expense of slower inference. super_res_num_inference_steps (`int`, *optional*, defaults to 7): The number of denoising steps for super resolution. More denoising steps usually lead to a higher quality image at the expense of slower inference. generator (`torch.Generator`, *optional*): A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make generation deterministic. decoder_latents (`torch.Tensor` of shape (batch size, channels, height, width), *optional*): Pre-generated noisy latents to be used as inputs for the decoder. super_res_latents (`torch.Tensor` of shape (batch size, channels, super res height, super res width), *optional*): Pre-generated noisy latents to be used as inputs for the decoder. decoder_guidance_scale (`float`, *optional*, defaults to 4.0): A higher guidance scale value encourages the model to generate images closely linked to the text `prompt` at the expense of lower image quality. Guidance scale is enabled when `guidance_scale > 1`. image_embeddings (`torch.Tensor`, *optional*): Pre-defined image embeddings that can be derived from the image encoder. Pre-defined image embeddings can be passed for tasks like image interpolations. `image` can be left as `None`. output_type (`str`, *optional*, defaults to `"pil"`): The output format of the generated image. Choose between `PIL.Image` or `np.array`. return_dict (`bool`, *optional*, defaults to `True`): Whether or not to return a [`~pipelines.ImagePipelineOutput`] instead of a plain tuple. Returns: [`~pipelines.ImagePipelineOutput`] or `tuple`: If `return_dict` is `True`, [`~pipelines.ImagePipelineOutput`] is returned, otherwise a `tuple` is returned where the first element is a list with the generated images. NrrrAg?)rerZr[rTmps)valueT)r0r)sampletimestepencoder_hidden_states class_labelsrIr?) prev_timestepr/rB)size antialiasbicubicF)rmode align_corners)rzr{g?r pil)rg)4rDPILImagerErFr2_execution_devicercrnrtyperPintFpadclip_extra_context_tokensrJr# set_timesteps timestepsrconfig in_channels sample_sizer8r1 enumerate progress_barrQrzchunksplitstep prev_sampleclampr$r! interpolateinspect signaturerir" XLA_AVAILABLExm mark_stepmaybe_free_model_hookscpupermutefloatnumpy numpy_to_pilr)(r)rmrSrprqr/rrrsrertrurvrUrRr0rTrZr[rXadditive_clip_time_embeddingsdecoder_text_maskdecoder_timesteps_tensornum_channels_latentsheightwidthitlatent_model_input noise_prednoise_pred_uncondnoise_pred_text_predicted_variancer~ image_smallsuper_res_timesteps_tensorchannelsimage_upscaledinterpolate_antialiasunets( r+__call__%UnCLIPImageVariationPipeline.__call__s5n  %11 E4(( Z "[[^ )//2J "''7 &>$$00 ##//  ""22165A*00&& Od//0HIJDAqE`O+'>?Q?W?WXY?Z`a'>'b$!16E6K6KL^LdLdefLgmn6K6o3!3.1G?]nKn1oo "YY 4F'GQO 1u066q99 $ 8Q ? #4499AmW`:k 3K:*//A6%   ../L.\%)%=%=%G%G"''..::a?%%,,88$$++77  $ $ 4 4Xvu5!!!(( !  ;;% ]];fe_MN$& !g// >III59%k2]]#)5/ QVZoNd//0JKLDAq.44Q7!;;**++!&,=~+NTU!V )f  1u288;; $ :1q5 A !% 8 8 = =A0 Yb!>!k } 5M8" ##% c! Aq! ##Aq!Q/557==? % %%e,E8O"%00r-)N) NrNNNNg @rT)$__name__ __module__ __qualname____firstlineno____doc___last_supported_versionr __annotations__rrr rr r rmodel_cpu_offload_seqr'r8rcrrPrjrnno_gradrrrrr GeneratorrstrrJr__static_attributes__ __classcell__)r*s@r+rr.s:' !!""--))00  &&((n % 2 !  '  .  5 % $ + - : =D~  T\]b]i]iTj   ]]_X\%&+--./3264837(+%* _1ciiootCIIOO/DellRST_1 #_1&) _1 (+ _1 EOO, _1"%,,/_1$ELL1_1#5<<0_1!&_1c]_1_1_1r-r))rtypingrrr PIL.ImagerrPtorch.nnrr transformersrrr r modelsr r schedulersrutilsrrutils.torch_utilsrpipeline_utilsrrrrrtorch_xla.core.xla_modelcore xla_modelrr get_loggerrloggerrrr-r+rsw(( $8)4-\\*))MM   H %A1#: