+ jzBRt^RIHt^RIHt^RIHt^RIHtH t H t H t ^RI H t Ht^RIHtHt]!4'd^RItRt!R R ] R R 7t!R R]R R 7tRRltRtRtRt]!RR] 44tR.tR#)z Processor class for IDEFICS. )urlparse) BatchFeature) ImageInput)ProcessingKwargsProcessorMixin TextKwargsUnpack)PreTokenizedInput TextInput)auto_docstringis_torch_availableNc0a]tRt^'toRtV3RltRtVtR#)IdeficsTextKwargsa7 add_eos_token (`bool`, *optional*, defaults to `False`): Whether to add an end-of-sequence token at the end of the text input. When enabled, an EOS token is appended to mark the end of the text sequence, which is useful for generation tasks. add_end_of_utterance_token (`bool`, *optional*): Whether to add an end-of-utterance token to mark the end of a user's message in conversational contexts. This token helps the model distinguish between different utterances in a multi-turn conversation and is particularly important for chat-based models. cN<V^8dQh/S[R,;R&S[R,;R&#)N add_eos_tokenadd_end_of_utterance_token)bool)format __classdict__s"o/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/idefics/processing_idefics.py __annotate__IdeficsTextKwargs.__annotate__'s'$;!%t +N)__name__ __module__ __qualname____firstlineno____doc____annotate_func____static_attributes____classdictcell__rs@rrr'srrF)totalcHa]tRt^6toRRRRRRR/RRR //tV3R ltR tVtR #) IdeficsProcessorKwargs text_kwargsadd_special_tokensFpaddinglongestr common_kwargsreturn_tensorsptc&<V^8dQh/S[;R&#)rr()r)rrs"rr#IdeficsProcessorKwargs.__annotate__6s""rrN)rrrr _defaultsr!r"r#r$s@rr'r'6s9  % y U *D1 Irr'cVR8wdVR8XdRWV8&VR8Xd:VR8Hp^W&\PPPWR7p^WCR3&X#)r. num_classes:NNN)torchnn functionalone_hot)incremental_maskr-r5 negatives attn_masks&&& r$incremental_to_binary_attention_maskr>Csmb T !@B < =$* &'#HH''//0@/Z "# Q, rc*VR8Xd \W4#R#)r.N),image_attention_mask_for_packed_input_ids_pt) input_ids tokenizerr-s&&&r)image_attention_mask_for_packed_input_idsrCTs;IQQrc\P!VRR7p\P!VRR7pVP\4pVPp\ VP ^44FfpRpRp\W,4FJwrW8XdV^, pWrV,V &RpM WrV,V &V'd RW&,V &W8XgKHRpKL Kh \ VP ^44FpRpRp\ W,P ^4^, RR4FWp W,V ,p W8XdV^, pWsV,V &RpM WsV,V &W8XdRpV'gKLRW6,V &KY W6,R8gp W6,V ;;,V,uu&W6,V ;;,R,uu&K W#3#)r3) fill_valueFTr6)r7 full_likeconvert_tokens_to_ids IMAGE_TOKEN eos_token_idrangesize enumerate) rArBimage_attention_masknext_image_attention_maskimage_token_id eod_token_id batch_idxcountseen_eodidxtoken_idnon_negative_indicess && rr@r@Ys ??9D % b I44[AN))L9>>!,- &y';>!,- -221592rBC +C0H) )4S9C 9CrI!,-ABeKB!,-ABbHB)., ::rclRV9dR#\V4p\VPVP.4#)zChecks if the passed string contains a valid url and nothing else. e.g. if space is included it's immediately invalidated the url F)rallschemenetloc)stringresults& ris_urlr^s0 f} f F  v}}- ..rcjaa]tRt^to]tRV3Rllt]RV3RlRll4t] R4t Rt Vt V;t #)IdeficsProcessorc ~<\SV`W4\VR4'd VPMVP \ 4VnVP PVP PVP P3Vn RVPPPR.49Vn R#)z image_size (int, *optional*, defaults to 224): The size of the image to be processed. add_end_of_utterance_token (bool, *optional*, defaults to None): Whether to add the end of utterance token to the text. image_tokenadditional_special_tokensN)super__init__hasattrrOrGrHimage_processorimage_num_channels image_sizedefault_image_dimsrBspecial_tokens_mapget1tokenizer_was_trained_with_end_of_utterance_token)selfrhrBrjrkwargs __class__s&&&&&,rrfIdeficsProcessor.__init__s 4y-00  $ $00=   3 3  + +  + +#  !DNN$E$E$I$IJegi$j j >rcr<V^8dQhRS[S[S[,,S[,S[S[,,S[S[S[,,,RS[S[,S[S[,,S[S[,,S[S[S[,,,S[S[S[,,,RS[S[,RS[/#)rimagestextrpreturn)rliststrr r rr'r)rrs"rrIdeficsProcessor.__annotate__sk k T*--3d3i?$tCy/Qk   y/  !" tI   t%& ' (k /0k  k rc  a(a)VfVf \R4hVfTpEMVEe\V\\34'gV.p\V\4'dV.p\V\\34'd%\ V4\ V48wd \R4h\ ;QJdRV4F 'dK RM RM !RV44'g \R4h\V^,\\34'd#\W4UUu.uF wrVV.VONK pppM\\W44pVP!\3RVPP/VBpVR ,PR R4pVR ,PR R4p V f VPp \;QJdR X4F 'gK RM RM !R X44'gV.pR o(Ro)Rp V(V)3Rlp .p .p VEFypVPP p.pRpRp\!V4FwppV^8d\#V'*4p\V\4'dVP%R4p\'V4'd?VP(P+V4pW!V4, pVP-V4RpKV 'dV'd W, pVV, pRpKW!V4, pVP-V4RpK V'dWPP., p\ V4^8dVP(!V3/VR,BpV P-V4V P-V4EK| VR ,PRR4pVP!V 3/VR ,BpVR,pVR,p\1RV 44p\1^V4p\3RV 44^8p.p.p.p\VVV 4EFwppp Tp!V!P5VP64p"\9V"V4p#V RV#p$\ V$4^8dJVR8XdB\:P<!V.V$P?4R,O5!p%V$V%RV$P?^4%M)VR8Xd#\:P<!V.VP@O5!p%VP-X%4VR8XgKVP-\:PB!V!44VP-\:PB!V44EK VR8XdC\:PD!V4p\:PD!V4p\:PD!V4pV'd*\GVVPV4wp&p'\IV&VVR7p&MQVR8XdK\:P<!VPJ^,VPJ^,^\:P"R7p&\MRVRVRVRX&/R7#uuppi)a4 Returns: a dict with entries: `input_ids`, `attention_mask`, `pixel_values`, `image_attention_mask` which can be directly passed to `model.generate` Detailed explanation: Each entry in `text` is either a text to be passed as is or an image that will be processed. An image can be either an image object (`PIL.Image`) or a url from which the image can be retrieved. When the processor encounters an image it'll inject `` entry into the prompt. Example: ```python checkpoint = "HuggingFaceM4/idefics-9b" processor = AutoProcessor.from_pretrained(checkpoint) url = "https://hips.hearstapps.com/hmg-prod/images/cute-photos-of-cats-in-grass-1593184777.jpg" img = processor.image_processor.fetch_images([url])[0] prompts = [ "User:", img, "Describe this image.\nAssistant: An image of two kittens in grass.\n", "User:", "https://hips.hearstapps.com/hmg-prod/images/dog-puns-1581708208.jpg", "Describe this image.\nAssistant:", ] inputs = processor(text=prompts, return_tensors="pt") generated_ids = model.generate(**inputs, max_length=100) generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0] ``` In this example the `prompts` will be converted into: ``` User:Describe this image. Assistant: An image of two kittens in grass. User:Describe this image. Assistant:' ``` and the two images will be massaged using [`IdeficsImageProcessor.__call__`] method and placed inside the `pixel_values` dict entry of the return value. This example also exemplifies that images can be passed as objects or as text urls. It can be seen that the first image is passed as object and the second one as a url. To do training do: ```python image_transform = transforms.Compose( [ transforms.RandomResizedCrop( (w, h), scale=(0.9, 1.0), interpolation=transforms.InterpolationMode.BICUBIC ), transforms.ToTensor(), transforms.Normalize(mean=self.image_mean, std=self.image_std), ] ) inputs = processor(text=prompts, transform=image_transform, return_tensors="pt") ``` In order to help debug prompt generation enable `debug=True` which will show you what's happening. Nz9You need to specify either `text` or `images` and `text`.aWhen providing both images and text arguments, the number of text prompts should be the same as the number of images.If you want to have several images per prompt, images should be nested as such: images=[[img1, img2], [img3, img4], ...] for text=[prompt1, prompt2, ...].c3B"TFp\V\4xK R#5iN) isinstancerx.0is& r ,IdeficsProcessor.__call__..s84az!S))4sFTzQWhen using the image-text-to-text behavior, the prompts should only contain text.tokenizer_init_kwargsr(rrc3N"TFp\V\\34xK R#5ir|)r}rwtupler~s& rrr'sAA:a$//s#%zr rccF<V'd SS,#SS,S,#r|r)last_was_image fake_tokenrbs&r image_tokens/IdeficsProcessor.__call__..image_tokens.s!"Z//!K/*<shaper)*rortrurpprompts image_listsample output_kwargsrrend_of_utterance_tokenr all_prompts all_images full_text image_objectsr last_was_textritemimager- text_encoding all_textsall_attention_masksmax_num_imagesat_least_one_imageoutput_input_ids output_imagesoutput_attention_masks text_singlerextracted_imagespadded_input_ids image_countlocal_max_num_imagescurrent_imagespadded_image_tensorrM_rrbs*&&&, @@r__call__IdeficsProcessor.__call__sb >dlXY Y >G  ftUm44 $$$v$u ..3t9F 3K q 384833384888 !tuu&)dE]33KNvK\]K\5GZF0Z0K\]s601** " "&.."<"<  &m488%P %2=%A%E%EFbdh%i" & -)-)_)_ &sAAsssAAAAiG0  !5 =   F>>334IM"N!M$V,4q5$(^);$"Q&!T)*/++n*a~GZGZG\]_G`*a'DR'(@.*=*=a*@A!T)*/++n*_tG^G^*_'  !4 5% '' 5E(FG&--ell>.JK#>m& T !${{+;< !KK 6M%*[[1G%H " &O $...' # !$H$n.$ %',{{$**1-/?/E/Ea/H!SXS]S]($- "8 &(<   y^sXcVPPpVPPp\W,R.,4#)rM)rBmodel_input_namesrhrw)rotokenizer_input_namesimage_processor_input_namess& rr"IdeficsProcessor.model_input_namess< $ @ @&*&:&:&L&L#)GKaJbbccr)rkrOrn)NN)NN)rrrrr'valid_processor_kwargsrfr rpropertyrr"r# __classcell__)rqrs@@rr`r`sI3 0k k k Zdddrr`)r6)r urllib.parserfeature_extraction_utilsr image_utilsrprocessing_utilsrrrrtokenization_utils_baser r utilsr r r7rHrr'r>rCr@r^r`__all__rrrrs"4% D7  , % , -U  "R ,;^/Md~MdMd`  r