+ j~!\Rt^RIHt^RIHt^RIHt.ROt]!RR]44tR.t R#)z Processor class for EVOLLA. ) BatchFeature)ProcessorMixin)auto_docstringcaa]tRt^toRV3RlltRRltRV3RlRllt]RV3RlRll4tRV3RlRllt RV3R lR llt R t R t R t VtV;t#)EvollaProcessorc <Vf \R4hVf \R4h\SV` W4RVPnW0nW@nR#)aQ protein_tokenizer (`EsmTokenizer`): An instance of [`EsmTokenizer`]. The protein tokenizer is a required input. protein_max_length (`int`, *optional*, defaults to 1024): The maximum length of the sequence to be generated. text_max_length (`int`, *optional*, defaults to 512): The maximum length of the text to be generated. Nz+You need to specify an `protein_tokenizer`.z"You need to specify a `tokenizer`.z<|reserved_special_token_0|>) ValueErrorsuper__init__ tokenizer pad_tokenprotein_max_lengthtext_max_length)selfprotein_tokenizerr r rkwargs __class__s&&&&&,m/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/evolla/processing_evolla.pyr EvollaProcessor.__init__sO  $JK K  AB B *6#A "4.c \.pVFpVPR4pVPR4pRP\WV4UUu.uF*wrxVP4VP 4,NK, upp4p VP V 4K VP VRRVRR7p V #uuppi)aa_seqfoldseekptT)return_tensors truncation max_lengthpadding)getjoinzipupperlowerappendr) rproteinsr sa_sequencesproteinrrsf sa_sequence sa_tokenss &&& rprocess_proteins EvollaProcessor.process_proteins0s G[[*F{{:.H''SEZ"[EZTQ1779qwwy#8#8EZ"[\K    ,  ** $K]gk+  #\s0B( c <V^8dQhRS[/#)r)int)format __classdict__s"r __annotate__EvollaProcessor.__annotate__=src .pVF2pVPPVRRR7pVPV4K4 VPVRRRRVR7pV#)FT)tokenizeadd_generation_promptrlongest)add_special_tokensrrrr)r apply_chat_templater$)rtextsrpromptsmessagesprompt prompt_inputss&&& r process_textEvollaProcessor.process_text=sr H^^77&*8F NN6 "  $& ' rc <V^8dQhRS[S[,S[,R,RS[S[S[,,S[S[,,R,RS[R,RS[R,/#)r/r%N messages_listr r)listdictr0)r1r2s"rr3r4Vs`0 0 t*t#d*0 DJ'$t*4t;0  $J 0 t 0 rc \VeVf \R4hVeTM VPpVeTM VPpVPW4wrVP W4VP W4pVP W$4p\RVR,RVR,RVR,RVR,/R7#)a proteins (`Union[List[dict], dict]`): A list of dictionaries or a single dictionary containing the following keys: - `"aa_seq"` (`str`) -- The amino acid sequence of the protein. - `"foldseek"` (`str`) -- The foldseek string of the protein. messages_list (`Union[List[List[dict]], List[dict]]`): A list of lists of dictionaries or a list of dictionaries containing the following keys: - `"role"` (`str`) -- The role of the message. - `"content"` (`str`) -- The content of the message. protein_max_length (`int`, *optional*, defaults to 1024): The maximum length of the sequence to be generated. text_max_length (`int`, *optional*, defaults to 512): The maximum length of the text. Return: a dict with following keys: - `protein_input_ids` (`torch.Tensor` of shape `(batch_size, sequence_length)`) -- The input IDs for the protein sequence. - `protein_attention_mask` (`torch.Tensor` of shape `(batch_size, sequence_length)`) -- The attention mask for the protein sequence. - `text_input_ids` (`torch.Tensor` of shape `(batch_size, sequence_length)`) -- The input IDs for the text sequence. - `text_attention_mask` (`torch.Tensor` of shape `(batch_size, sequence_length)`) -- The attention mask for the text sequence. z3You need to specify `messages_list` and `proteins`.protein_input_ids input_idsprotein_attention_maskattention_mask)data)rr rprepare_inputs_layoutvalidate_inputsr,r@r)rr%rCr rrr+ text_tokenss&&&&&, r__call__EvollaProcessor.__call__Us>  }4RS S3E3Q/W[WnWn-<-H/dNbNb"&"<""?   rc<V^8dQhRS[S[,S[,R,RS[S[S[,,S[S[,,R,/#r/r%NrCrDrE)r1r2s"rr3r4sD ' 't*t#d* 'DJ'$t*4t; 'rc \V\4'dV.p\V\\34'd'\V^,\\34'gV.pW3#)) isinstancerErDtuple)rr%rCrs&&&,rrL%EvollaProcessor.prepare_inputs_layoutsR h % % zH mdE] 3 3J}UVGWZ^`eYf2EvollaProcessor.validate_inputs..s:aX`ST:a;N;NX`FTzUThe proteins should be a list of dictionaries, but not all elements are dictionaries.c3"TFUp\;QJd*RVP44F 'dK RM RM!RVP444xKW R#5i)c32"TF q\9xK R#5ir\)PROTEIN_VALID_KEYS)r_ks& rra..s:A''sFTN)allkeysr^s& rrarbs=; DLqCC::CCC:: : :HsAA+Az2There should be a list of dictionaries with keys: z, z for each protein.But got: z;Each messages in messages_list should be a list instead of .c3B"TFp\V\4xK R#5ir\r]r_ms& rrarbsA1:a..rczfEach message in messages_list should be a list of dictionaries, but not all elements are dictionaries.c3Z"TF!p\VP44^8gxK# R#5i)r/N)lenrjrms& rrarbs<8as1668})8s)+c3^"TF#p\VP44RR08gxK% R#5i)rolecontentN)setrjrms& rrarbs(DBJQCMfi%88(s+-zlEach message in messages_list should be a list of dictionaries with two keys: 'role' and 'content'.But got: zFThe messages_list should be a list of lists of dictionaries, but it's N) rVrDrWrirr rf TypeErrortypeany)rr%rCrr=s&&&, rrMEvollaProcessor.validate_inputss hu . .ss:aX`:asss:aX`:a7a7atu u hu . .ss; DL; sss; DL; 8 8 D99/012$:'  mdE] 3 3)!(T5M::#&abfgobpaqqr$sttsAAsssAAAA$A3<8<333<8<<<DBJDDBJDAA%$$,:/*XY]^kYlXmmno rc:VPP!V/VB#r\)r batch_decoderargsrs&*,rprotein_batch_decode$EvollaProcessor.protein_batch_decodes%%22DCFCCrc:VPP!V/VB#r\)rdecoder{s&*,rprotein_decodeEvollaProcessor.protein_decodes%%,,d=f==r)r r)N)r)r)NNNN)NN)__name__ __module__ __qualname____firstlineno__r r,r@rrOrLrMr}r__static_attributes____classdictcell__ __classcell__)rr2s@@rrrsX/( 00 0 0 d ' '$$LD>>rrN)rrmsa) __doc__feature_extraction_utilsrprocessing_utilsrutilsrrfr__all__rrrsF5.#3e>ne>e>P  r