+ jc%^RIt^RIHtHtHt^RIHtHt]P!] 4t !RR]RR7t ]!RR ]44t R .tR#) N)MultiModalDataProcessingKwargsProcessorMixin)auto_docstringloggingc2]tRt^tRRRRRRR/RRR//tR tR #) Glm46VProcessorKwargs text_kwargspaddingFreturn_token_type_idsreturn_mm_token_type_idsT videos_kwargsreturn_metadataN)__name__ __module__ __qualname____firstlineno__ _defaults__static_attributes__rm/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/glm46v/processing_glm46v.pyr r s- u #U & +T2 Irr F)totalcaa]tRt^)to]tRV3RlltV3RlRltV3RlRltRRlt RRlt ] V3Rl4t V3R lR lt RV3R lR lltR tVtV;t#)Glm46VProcessorc <\VR4'gRM VPVn\VR4'gRM VPVn\VRR4'd VPMVP VP4Vn\VRR4'd VP MVP VP4Vn\SV`!WW4R7VP R4Vn VP R 4Vn R#) image_tokenz <|image|> video_tokenz <|video|>image_token_idNvideo_token_id) chat_templatez<|begin_of_video|>z<|end_of_video|>) hasattrrrgetattrrconvert_tokens_to_idsr super__init__video_start_id video_end_id)selfimage_processor tokenizervideo_processorr!kwargs __class__s&&&&&,rr&Glm46VProcessor.__init__-s.5i.O.O;U^UjUj.5i.O.O;U^UjUjy"2D99  $ $001A1AB y"2D99  $ $001A1AB  _b'==>RS%;;s#3333SV3rc VPP^,pVR,V,P4V,pVPV,#)r1image_grid_thw)r* merge_sizeprodr)r)r2r3r- merge_lengthnum_image_tokenss&&&, rreplace_image_token#Glm46VProcessor.replace_image_token>sI++669 '(89)DIIK|["222rc,<V^8dQhRS[RS[RS[/#)r1 video_inputs video_idxr4r5)r9r:s"rr;r<Cs#SVrc VPP^,pVR,V,^,pVR,V,P4V,V,pVR,V,pRpVPf\P R4VPf^M VPVnVP R,p .p \^\V 44Fp V PW,4K V RVp \V 4V8d%T PV 'd V R,M^4K4\V4F%p W,pVPWR7pW, pK' V#) r1video_grid_thwvideo_metadataNaGLM46V requires frame timestamps to construct prompts, but the `fps` of the input video could not be inferred. Probably `video_metadata` was missing from inputs and you passed pre-sampled frames. Defaulting to `fps=24`. Please provide `video_metadata` for more accurate results.:NNr1)rB) r,r?r@fpslogger warning_once timestampsrangelenappendreplace_frame_token_id)r)rFrGr-rA num_framesrBmetadatavideo_structurerPunique_timestampsidxselected_timestamps frame_idx timestamp_secframe_structures&&&, rreplace_video_token#Glm46VProcessor.replace_video_tokenCsI++669 !"23I>qA '(89)DIIK|[_ii 01)< <<    e  &\\1rx|| ((- C O,C  $ $Z_ 5-0 <%&3  & &BU':2'>[\ ]z*I/:M"99-9kO  .O+ rc /pVe\PPR/4pVPV4VPRR4;'gVPP pVUu.uF"pVPP !.VOVN5!NK$ ppVU u.uFqV^,,NK p p VPRV RV/4Ve\PPR/4p V PV4VU u.uF"p VPP!.V OV N5!NK$ p p V U u.uFqX^,,NK pp WR&\R/VB#uupiuup iuup iuup i) 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. video_sizes (`list[list[int]]`, *optional*): The input sizes formatted as (num_frames, height, width) per each video. Returns: `MultiModalData`: A `MultiModalData` object holding number of tokens per each of the provided input modalities, along with other useful data. N images_kwargsr?rBnum_image_patchesrnum_video_tokensr) r rgetupdater*r?get_number_of_image_patchesr,get_number_of_video_patchesr)r) image_sizes video_sizesr- vision_datarar? image_sizerb num_patchesrBr video_sizenum_video_patchesrcs&&&, r_get_num_multimodal_tokens*Glm46VProcessor._get_num_multimodal_tokensbs  "1;;??QSTM   (&**<>aa$BVBVBaBaJ#.!"-J$$@@\*\m\"- !SddRc; A !=!=Rc d    24DFY[lm n  "1;;??QSTM   (#.!"-J$$@@\*\m\"- !SddRc; A !=!=Rc d.>* +, ,,#! e ! es,(EE(E#0E(c DVPP!V3RVRV/VB#)a* Post-process the output of the model to decode the text. Args: generated_outputs (`torch.Tensor` or `np.ndarray`): The output of the model `generate` function. The output is expected to be a tensor of shape `(batch_size, sequence_length)` or `(sequence_length,)`. skip_special_tokens (`bool`, *optional*, defaults to `True`): Whether or not to remove special tokens in the output. Argument passed to the tokenizer's `batch_decode` method. clean_up_tokenization_spaces (`bool`, *optional*, defaults to `False`): Whether or not to clean up the tokenization spaces. Argument passed to the tokenizer's `batch_decode` method. **kwargs: Additional arguments to be passed to the tokenizer's `batch_decode method`. Returns: `list[str]`: The decoded text. skip_special_tokensclean_up_tokenization_spaces)r+ batch_decode)r)generated_outputsrrrsr-s&&&&,rpost_process_image_text_to_text/Glm46VProcessor.post_process_image_text_to_texts;(~~**  3 *F    rc(<\SV`R.,#)mm_token_type_ids)r%model_input_names)r)r.s&rrz!Glm46VProcessor.model_input_namessw(,?+@@@rcF<V^8dQhRS[RS[S[S[,,/#)r1 input_idsr4)listr7)r9r:s"rr;r<s#!!$!4S ?!rc.pVFp\P!V4p\P!V4p\P!W@P8H^R7p\P!W@P 8H^R7pWg8p^WTVP 8HV,&^WTVP 8HV(,&VPVP44K V#)r)axis) nparray zeros_likecumsumr'r(rrStolist) r)r}ryinput array_idsmm_token_typesstartsendsis_video_modalitys && rcreate_mm_token_type_ids(Glm46VProcessor.create_mm_token_type_idssEI]]51N YYy,?,??aHF99Y*;*;;!DD & UVN)<)<<@QQ RXYN)<)<z<|end_of_image|>z.1fz seconds)r)r)r\rBs&&&rrT&Glm46VProcessor.replace_frame_token_ids+#D$4$47G$G#HHXYfgjXkksttr)rrr(r'rr )NNNN)NN)TF))rrrrr valid_processor_kwargsr&rCr^rorvpropertyrzrrTr__classdictcell__ __classcell__)r.r:s@@rrr)sb2P"33 >$-L 6AA!!*uuurr)numpyrprocessing_utilsrrrutilsrr get_loggerrrNr r__all__rrrrs`*PP,   H %,ESunSuSul  r