>> from transformers import pipeline >>> captioner = pipeline(model="ydshieh/vit-gpt2-coco-en") >>> captioner("https://huggingface.co/datasets/Narsil/image_dummy/raw/main/parrots.png") [{'generated_text': 'two birds are standing next to each other '}] ``` Learn more about the basics of using a pipeline in the [pipeline tutorial](../pipeline_tutorial) This image to text pipeline can currently be loaded from pipeline() using the following task identifier: "image-to-text". See the list of available models on [huggingface.co/models](https://huggingface.co/models?pipeline_tag=image-to-text). TF)max_new_tokensc>[TU]"U0UD6 [US5 URURS:Xa [ 5 g[ 5 g)Nvisiontf)super__init__r check_model_type frameworkrr)selfargskwargs __class__s \/var/www/html/shao/venv/lib/python3.13/site-packages/transformers/pipelines/image_to_text.pyr ImageToTextPipeline.__init__TsD $)&)$) 7;~~7M 3 Sw c0n0nUbX6S'UbXFS'UbXS'Ub%UbSU;a [S5eURU5 URbURUS'URbURUS'URUS'Xe04$)Nprompttimeoutrzp`max_new_tokens` is defined both as an argument and inside `generate_kwargs` argument, please use only 1 versionassistant_model tokenizerassistant_tokenizer) ValueErrorupdater-r/r.)r#rgenerate_kwargsr+r,forward_paramspreprocess_paramss r'_sanitize_parameters(ImageToTextPipeline._sanitize_parameters[s  *0h '  +2i (  %/=+ ,  &).>/.Q &  ! !/ 2    +040D0DN, -  # # /*...N; '484L4LN0 1 "44r)inputsz Image.Imager%returnc gNr#r7r%s r'__call__ImageToTextPipeline.__call__vsber)c gr:r;r<s r'r=r>ystwr)c p>SU;aURS5nUc [S5e[TU] "U40UD6$)a' Assign labels to the image(s) passed as inputs. Args: inputs (`str`, `list[str]`, `PIL.Image` or `list[PIL.Image]`): The pipeline handles three types of images: - A string containing a HTTP(s) link pointing to an image - A string containing a local path to an image - An image loaded in PIL directly The pipeline accepts either a single image or a batch of images. max_new_tokens (`int`, *optional*): The amount of maximum tokens to generate. By default it will use `generate` default. generate_kwargs (`Dict`, *optional*): Pass it to send all of these arguments directly to `generate` allowing full control of this function. timeout (`float`, *optional*, defaults to None): The maximum time in seconds to wait for fetching images from the web. If None, no timeout is set and the call may block forever. Return: A list or a list of list of `dict`: Each result comes as a dictionary with the following key: - **generated_text** (`str`) -- The generated text. imageszBCannot call the image-to-text pipeline without an inputs argument!)popr0rr=)r#r7r%r&s r'r=r>|sA< v ZZ)F >ab bw1&11r)c[XS9nUGb[RS5 [U[5(d[ S[ U5S35eURRRnUS:XaURXRS9nURS:XaURUR5nURUSS 9RnURR /U-n["R$"U5R'S 5nUR)S U05 GOUS :XaFURXURS 9nURS:XaURUR5nOUS:waoURXRS9nURS:XaURUR5nURX RS9nUR)U5 OS[ SUS35eURXRS9nURS:XaURUR5nURRRS:XaUcSUS 'U$)N)r,uPassing `prompt` to the `image-to-text` pipeline is deprecated and will be removed in version 4.48 of 🤗 Transformers. Use the `image-text-to-text` pipeline insteadz&Received an invalid text input, got - zy - but expected a single string. Note also that one single text can be provided for conditional image to text generation.git)rAreturn_tensorsptF)textadd_special_tokensr input_ids pix2struct)rA header_textrEzvision-encoder-decoder)rEz Model type z- does not support conditional text generation)rlogger warning_once isinstancestrr0typemodelconfig model_typeimage_processorr"to torch_dtyper.rI cls_token_idtorchtensor unsqueezer1)r#imager+r,rS model_inputsrI text_inputss r' preprocessImageToTextPipeline.preprocesss252     W fc** >T)#/??43C3C#DL NN5NQ[[ !^^889IE !LL3==a@ ##[)$<=|+#335eieses3t >>T)#/??43C3C#DL77#335Q_Q_3` >>T)#/??43C3C#DL"nnVNNnS ##K0!;zl:g!hii //u^^/\L~~%+t/?/?@ ::   ' '5 0V^(,L %r)c 2SU;a7[US[5(a[SUS55(aSUS'SU;aURUS'UR UR R 5nUR R"U40UDUD6nU$)NrIc3&# UHoSLv M g7fr:r;).0xs r' /ImageToTextPipeline._forward..sA'@!I'@sgeneration_config)rNlistallrfrBrQmain_input_namegenerate)r#r\r2r7 model_outputss r'_forwardImageToTextPipeline._forwards < '< 4d;;A|K'@AAA(,L % o 5373I3IO/ 0 !!$**"<"<= ++FVlVoV r)cv/nUH0nSURRUSS90nURU5 M2 U$)Ngenerated_textT)skip_special_tokens)r.decodeappend)r#rkrecords output_idsrecords r' postprocessImageToTextPipeline.postprocesssL'J $.."7"7(,#8#F NN6 "(r)r;)NNNN)NN)__name__ __module__ __qualname____firstlineno____doc___pipeline_calls_generate_load_processor_load_image_processor_load_feature_extractor_load_tokenizerr_default_generation_configr r5rrrOrrgdictr=r^rlrv__static_attributes__ __classcell__)r&s@r'rr.s4 $O #O!1" 56euS-%78eCeDQUVY[^V^Q_L`ee wuT#Y]0C%CDwPSwX\]abfgjlogobp]qXrww"2uS$s)]DDW%WX"2H1f,  r)r)typingrrr generationrutilsrr r r r r baserrPILr image_utilsrmodels.auto.modeling_tf_autorrXmodels.auto.modeling_autor get_loggerrxrLrr;r)r'rs (')5(VP   H %,4UYZ[D(D\Dr)