^j#dZddlmZddlmZddlmZmZmZm Z m Z ddl m Z m Z ddlmZGdde d ZGd d ed ZeGd deZdgZy)z Processor class for Chameleon. ) BatchFeature) ImageInput)MultiModalDataProcessingKwargsProcessorMixin TextKwargsUnpack)PreTokenizedInput TextInput)auto_docstringceZdZUdZeed<y)ChameleonTextKwargsa return_for_text_completion (`bool`, *optional*, defaults to `False`): Whether the processed text is intended for text completion tasks. When `True`, the processor does not append the separator token (`sep_token`) to the end of the prompt, which is typically used for chat mode. When `False`, the separator token is appended for proper chat formatting. return_for_text_completionN)__name__ __module__ __qualname____doc__bool__annotations__}/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/transformers/models/chameleon/processing_chameleon.pyrrs!%$rrF)totalc.eZdZUeed<ddddddidZy)ChameleonProcessorKwargs text_kwargsF)paddingrreturn_mm_token_type_idsreturn_tensorspt)r common_kwargsN)rrrrr _defaultsrrrrr*s+$$*/(- d Irrc eZdZeZddedeffd Zede efdZ e dde dzde eze e ze ezdzd eedeffd Zd ed edefd ZddZxZS)ChameleonProcessorimage_seq_length image_tokenct|||||_t|dr |jn||_|j |j|_t|dr |jnd|_t|dr |jnd|_ |j |j|_|j |j|_ |j |j|_ y)a image_seq_length (`int`, *optional*, defaults to 1024): Sequence length of one image embedding. image_token (`str`, *optional*, defaults to `""`): The special token used to indicate image in the text. r& boi_tokenz eoi_tokenzN) super__init__r%hasattrr&convert_tokens_to_idsimage_token_idr(image_start_tokenr)image_end_tokenimage_start_token_idimage_end_token_id)selfimage_processor tokenizerr%r& __class__s rr+zChameleonProcessor.__init__<s )4 04;I}4U900[f'==d>N>NO#*9k#BI   7>i6Uy22[c'==d>N>NO$-$C$CDDZDZ$[!"+"A"A$BVBV"WrreturncH|j|j|jgSN)r.r1r2)r3s rimage_token_idsz"ChameleonProcessor.image_token_idsPs!##T%>%>@W@WXXrNimagestextkwargsc t|tr|g}|Dcgc]}||jj}}t |d||d|}|Scc}w)aA Returns: [`BatchFeature`]: A [`BatchFeature`] with the following fields: - **input_ids** -- List of token ids to be fed to a model. Returned when `text` is not `None`. - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not `None`). - **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`. )r;r<r) isinstancestrr5 sep_tokenr*__call__)r3r;r<r=sample model_inputsr6s rrBzChameleonProcessor.__call__Tsf" dC 6DEII&6(4>>3345IIw'KvDKFK Js!A image_inputs image_idxcl|j|j|jzz|jz}|Sr9)r/r&r%r0)r3rErFone_img_tokenss rreplace_image_tokenz&ChameleonProcessor.replace_image_tokenms5//43C3CdF[F[3[\_c_s_ssrc i}|?|jdzgt|z}dgt|z}|j||dtdi|S)a Computes the number of placeholder tokens needed for multimodal inputs with the given sizes. Args: image_sizes (`list[list[int]]`, *optional*): The input sizes formatted as (height, width) per each image. Returns: `MultiModalData`: A `MultiModalData` object holding number of tokens per each of the provided input modalities, along with other useful data. )num_image_tokensnum_image_patchesr)r%lenupdater)r3 image_sizesr= vision_datarMrNs r_get_num_multimodal_tokensz-ChameleonProcessor._get_num_multimodal_tokensqsb  " $ 5 5 9:S=MM !"c+&6 6    4D[lm n, ,,r)iz)NNr9)rrrrvalid_processor_kwargsintr@r+propertylistr:r rr r r rrBdictrIrS __classcell__)r6s@rr$r$8s5XSX^aX(YcYY%)aeT!++d9o=EV@WWZ^^12   0-rr$N)rfeature_extraction_utilsr image_utilsrprocessing_utilsrrrrr tokenization_utils_baser r utilsr rrr$__all__rrrr`sk5%D#%*E% /u M-M-M-` r