Ngx\ddlmZddlZddlZddlZddlmZddlmZddlm Z m Z ddl Z ddl Z ddl m Z ddlmZmZmZmZmZddlmZdd lmZejeZddZGdde jZdS)) annotationsNfnmatch)Path)AnyCallable)nn) AutoConfig AutoModel AutoTokenizer MT5ConfigT5Config)is_peft_available)find_adapter_config_file_save_pretrained_fnr subfolderstrreturnCallable[..., None]cdfd }|S)Nsave_directory str | PathrNonectjt|z dt|z fi|S)NT)exist_ok)osmakedirsr)rkwargsrrs d/var/www/html/ai-engine/env/lib/python3.11/site-packages/sentence_transformers/models/Transformer.pywrapperz)_save_pretrained_wrapper..wrappersK D((94tDDDD""4#7#7)#CNNvNNN)rrrr)rrr s`` r_save_pretrained_wrapperr#s5OOOOOOO Nr!ceZdZUdZdZded< d:d;fd Zdd%Z d?d&Z d=d'Zd=d(Zd@d)ZdAd,ZdBd.Z dCdDd3ZdEd4ZdCdFd7ZedGd9ZxZS)H TransformeraHugging Face AutoModel to generate token embeddings. Loads the correct class, e.g. BERT / RoBERTa etc. Args: model_name_or_path: Hugging Face models name (https://huggingface.co/models) max_seq_length: Truncate any inputs longer than max_seq_length model_args: Keyword arguments passed to the Hugging Face Transformers model tokenizer_args: Keyword arguments passed to the Hugging Face Transformers tokenizer config_args: Keyword arguments passed to the Hugging Face Transformers config cache_dir: Cache dir for Hugging Face Transformers to store/load models do_lower_case: If true, lowercases the input (independent if the model is cased or not) tokenizer_name_or_path: Name or path of the tokenizer. When None, then model_name_or_path is used backend: Backend used for model inference. Can be `torch`, `onnx`, or `openvino`. Default is `torch`. Tbool save_in_rootNFtorchmodel_name_or_pathrmax_seq_length int | None model_argsdict[str, Any] | Nonetokenizer_args config_args cache_dir str | None do_lower_casetokenizer_name_or_pathbackendrrc dtddg|_||_| |_|i}|i}|i}|||| |} |j|| || fi|| d|vr||d<tj||n|fd|i||_ |mt|j drXt|j j dr>t|j dr)t|j j j|j j}||_|"|j jj|j j _dSdS)Nr*r2model_max_lengthr0configmax_position_embeddings)super__init__ config_keysr2r4 _load_config _load_modelr from_pretrained tokenizerhasattr auto_modelr7minr8r6r* __class____name__tokenizer_class) selfr)r*r,r.r/r0r2r3r4r7rCs rr:zTransformer.__init__6s ,o>*  J  !N  K""#5y';WW+VYVV:VVV  %*?? v "%T_%;%SUYUcUt!u!u, ! -59^5M5VDO " 2 2 2 . -r!dict[str, Any]c ^t||d|d|dd Jtstd|dkrt d d d lm}|j|fi|d |iStj|fi|d |iS) z"Loads the configuration of a modeltokenrevisionlocal_files_onlyF)rIrJrKNzgLoading a PEFT model requires installing the `peft` package. You can install it via `pip install peft`.r(aPEFT models can currently only be loaded with the `torch` backend. To use other backends, load the model with `backend="torch"`, call `model[0].auto_model.merge_and_unload()`, save that model with `model.save_pretrained()` and then load the model with the desired backend.r) PeftConfigr0) rgetr Exception ValueErrorpeftrLr>r )rFr)r0r4r/rLs rr<zTransformer._load_configfs %"!oog..$44!,1CU!K!K       %&& }'!! w ( ' ' ' ' '-:-.@eeKee[deee e)*<aa aaW`aaaar!c |dkrxt|tr|j|||fi|n@t|tr|j|||fi|nt j|f||d||_|j|||fi|dS|dkr|j |||fi|dS|dkr|j |||fi|dStd|d)zLoads the transformer modelr(r7r0onnxopenvinozUnsupported backend 'z6'. `backend` should be `torch`, `onnx`, or `openvino`.N) isinstancer_load_t5_modelr _load_mt5_modelr r>rA_load_peft_model_load_onnx_model_load_openvino_modelrO)rFr)r7r0r4r,s rr=zTransformer._load_modelsI g  &(++ ##$6 XXZXXXXFI.. $$%7YYjYYYY"+";&#/5##NX## "D !"4fi V V: V V V V V    !D !"4fi V V: V V V V V  " " %D %&8&) Z Zz Z Z Z Z ZtWtttuu ur!c tr6ddlm}m}t ||r |j|j|f||d||_dSdSdS)Nr)rL PeftModelrR)rrPrLr\rUr>rA)rFr)r7r0r,rLr\s rrXzTransformer._load_peft_models     2 2 2 2 2 2 2 2&*-- ";)";O%7#@FR[##_i##     r!c jt|tst|trtd ddlm}ddlm}n#t$rtdwxYwt|}| }d} d} | ||||| | \} }| r| ddd |vr|d } t| tspt|  std t| d 5} t!j| |d <dddn #1swxYwYni|d <|j|f||| d ||_t)|jj|j|j_| r|||| dSdS)Nz8T5 models are not yet supported by the OpenVINO backend.r)OVModelForFeatureExtraction)OV_XML_FILE_NAMEzUsing the OpenVINO backend requires installing Optimum and OpenVINO. You can install them with pip: `pip install optimum[openvino]`.OpenVINOz openvino*.xml file_name ov_configzXov_config should be a dictionary or a path to a .json file containing an OpenVINO configzutf-8)encodingr7r0export)rUrr rO optimum.intelr^optimum.intel.openvinor_ModuleNotFoundErrorrNrexists_backend_should_exportpopdictopenjsonloadr>rAr#_save_pretrainedr4_backend_warn_to_save)rFr)r7r0r,r^r_ load_pathis_local backend_nametarget_file_globrerbfs rrZz Transformer._load_openvino_modelsy fh ' ' Y:fi+H+H YWXX X  A A A A A A ? ? ? ? ? ? ?"   R   +,, ##%%! *"88 x-=?OQ]    . NN; - - - * $ $";/Ii.. ;I--//$r)g666;!.2illJ{+;;;;;;;;;;;;;;;')J{ #8c7R7b 8  8 8  8 8 ,DDODdfjfr+s+s(  S  & &'98\ R R R R R S Ss AA"#EE E c , ddl}ddlm}m}n#t$rt dwxYw|d|d|d<t|}| } d} d} | || ||| | \} }| r|dd|j |f||| d||_ t|j j|j|j _| r||| | dSdS) Nr)ONNX_WEIGHTS_NAMEORTModelForFeatureExtractionzUsing the ONNX backend requires installing Optimum and ONNX Runtime. You can install them with pip: `pip install optimum[onnxruntime]` or `pip install optimum[onnxruntime-gpu]`providerONNXz*.onnxrard) onnxruntimeoptimum.onnxruntimerxryrhrNrkget_available_providersrrirjr>rAr#rpr4rq) rFr)r7r0r,ortrxryrrrsrtrures rrYzTransformer._load_onnx_models  % % % % [ [ [ [ [ [ [ [ ["   <  ", C@PR^    . NN; - - -9e8T8d 9  9 9  9 9 ,DDODdfjfr+s+s(  S  & &'98\ R R R R R S Ss )rrrrstarget_file_namerurttuple[bool, dict[str, Any]]c ~|dd}|r||fS|d|}|dd} | r"t| |n t|} | r(t| |j|n&t|j|} | r| d|nd||r"fdD} n`t jd|d d|d d } fd | D} | | v}|sid|vre| | v}|r_t| d kr=d|vr9t d|dd| d| d |j|d<||d<|| }t|j }t|d krI|d|d<t|ddg|ddR|d<|rbt d|dd|d| r*t d| d|d| dd||fS)a Determines whether the model should be exported to the backend, or if it can be loaded directly. Also update the `file_name` and `subfolder` model_args if necessary. These are the cases: 1. If export is set in model_args, just return export 2. If `/` exists; set export to False 3. If `/` exists; set export to False and set subfolder to the backend (e.g. "onnx") 4. If `` contains a folder, add those folders to the subfolder and set the file_name to the last part We will warn if: 1. The expected file does not exist in the model directory given the optional file_name and subfolder. If there are valid files for this backend, but they're don't align with file_name, then we give a useful warning. 2. Multiple files are found in the model directory that match the target file name and the user did not specify the desired file name via `model_kwargs={"file_name": ""}` Args: load_path: The model repository or directory, as a Path instance is_local: Whether the model is local or remote, i.e. whether load_path is a local directory model_args: The model_args dictionary. Notable keys are "export", "file_name", and "subfolder" target_file_name: The expected file name in the model directory, e.g. "model.onnx" or "openvino_model.xml" target_file_glob: The glob pattern to match the target file name, e.g. "*.onnx" or "openvino*.xml" backend_name: The human-readable name of the backend for use in warnings, e.g. "ONNX" or "OpenVINO" Returns: Tuple[bool, dict[str, Any]]: A tuple of the export boolean and the updated model_args dictionary. reNrarz/**/z**/c^g|])}|*Sr") relative_toas_posix).0pathrrs r z6Transformer._backend_should_export..8s3ppp4 0 0 ; ; D D F Fpppr!modelrJrI) repo_typerJrIc4g|]}t||Sr"r)rfname glob_patterns rrz6Transformer._backend_should_export..@s(]]]%|@\@\]]]]r!z Multiple z files found in z: z, defaulting to zW. Please specify the desired file name via `model_kwargs={"file_name": ""}`.zNo z found in z. Exporting the model to .z#If you intended to load one of the  zN files, please specify the desired file name via `model_kwargs={"file_name": "rz"}`.) rkrMrrr4globhuggingface_hublist_repo_fileslenloggerwarningparts)rFrrrsr,rrurtrerarprimary_full_pathsecondary_full_pathmodel_file_names all_files model_foundfile_name_partsrs ` @rrjz"Transformer._backend_should_exportsN$//  &:% %NN;0@AA NN;55 ENnDI66??AAATXYbTcTcTlTlTnTn :DDL) 4 4 = = ? ? ?dlI..7799  @If);;)9;;;NfTdNfNf   ^ppppS\SaSabnSoSoppp  '7""$$!# D99 nnWd33 I ^]]]9]]]  (+;;  4{*<<-1AAK 4'((1,,J1N1NNNsLss)BTBTBVBVss\lssRsss+/, ;'*3 ;' >$_Fy///   ! # #&5b&9J{ #&*:>>+r+J+J&b_]`^`]`Ma&b&b&b&k&k&m&mJ{ #   NNmimmY-?-?-A-Amm^jmmm     y:Jyy\yy^nop^qyyy z!!r!cpd|d}|r |d|dz }n |d|dz }t|dS)NzSaving the exported zA model is heavily recommended to avoid having to export it again.z# Do so with `model.save_pretrained(z)`.z Do so with `model.push_to_hub(z, create_pr=True)`.)rr)rFr)rsrtto_logs rrqz!Transformer._backend_warn_to_savefscw www  b UrA)rFr)r7r0r,rs rrVzTransformer._load_t5_modelnsR//////=HM98.8  '-  FP  r!c Nddlm}dg|_|j|f||d||_dS)rr)MT5EncoderModelrrRN)rrrr>rA)rFr)r7r0r,rs rrWzTransformer._load_mt5_modelwsR000000>I]:9/9  '-  FP  r!cVd|d|jjjdS)Nz Transformer(z) with Transformer model: r)get_config_dictrArCrDrFs r__repr__zTransformer.__repr__s1ud2244uuPTP_PiPruuuur!featuresdict[str, torch.Tensor]c P|d|dd}d|vr |d|d<|jdi||ddi}|d}trddlm}t |j|rt|jjjrc|d}|d}tj ||jjj |j } tj | |fd |d<||d <|jj jr7t|d kr$d } t|dkrd } || } | |d<|S)z#Returns token_embeddings, cls_token input_idsattention_mask)rrtoken_type_ids return_dictFr)PeftModelForFeatureExtraction)devicer)dimtoken_embeddingsall_layer_embeddingsr")rArrPrrUactive_peft_configis_prompt_learningsizer(onesnum_virtual_tokensrcatr7output_hidden_statesr) rFrrtrans_features output_states output_tokensr batch_sizerprefix_attention_mask all_layer_idx hidden_statess rforwardzTransformer.forwards'/ '..s&4441A 444r!r"rcols rrz(Transformer.tokenize..s'LLL44444LLLr!c&g|]}d|DS)c6g|]}|Sr")lowerrs rrz3Transformer.tokenize...s 333!AGGII333r!r"rs rrz(Transformer.tokenize..s'KKK33s333KKKr! longest_firstpt)r truncationreturn_tensors max_length) rUrrlnextiteritemsappendr2updater?r*) rFrroutput to_tokenizelookuptext_keytextbatch1batch2 text_tuples rtokenizezTransformer.tokenizes eAh $ $ + 'KK a$ ' ' +K"$F;  5 5!%d6<<>>&:&:!;!;$""4((({#**84444&-KKFF# - -  jm,,, jm,,,,!6*KML LLL    LKK{KKKK DN*#.        r!c*fdjDS)Nc,i|]}|j|Sr")__dict__)rkeyrFs r z/Transformer.get_config_dict..s"DDDCT]3'DDDr!)r;rs`rrzTransformer.get_config_dicts DDDD43CDDDDr! output_pathsafe_serializationcR|j|||j|ttj|dd5}tj| |dddddS#1swxYwYdS)N)rsentence_bert_config.jsonwr)indent) rAsave_pretrainedr?rmrrjoinrndumpr)rFrrfOuts rsavezTransformer.saves '' HZ'[[[ &&{333 "',,{,GHH# N N >RV Id**,,d1 = = = = > > > > > > > > > > > > > > > > > >s%*BB #B  input_pathcdD]C}tj||}tj|rnDt |5}t j|}dddn #1swxYwYd|vr%d|dvr|ddd|vr%d|dvr|ddd|vr%d|dvr|dd|dd|i|S)N)rzsentence_roberta_config.jsonzsentence_distilbert_config.jsonzsentence_camembert_config.jsonzsentence_albert_config.jsonz sentence_xlm-roberta_config.jsonzsentence_xlnet_config.jsonr,trust_remote_coder.r/r)r")rrrrirmrnrork)clsr config_namesbert_config_pathfInr7s rrozTransformer.loadsz  K!# Z E E w~~/00  # $ $ $Ys^^F $ $ $ $ $ $ $ $ $ $ $ $ $ $ $ 6 ! !&9VL=Q&Q&Q < $ $%8 9 9 9 v % %*=HXAY*Y*Y # $ ( ()< = = = F " "':f]>S'S'S = ! % %&9 : : :s;;j;F;;;sA77A;>A;)NNNNNFNr()r)rr*r+r,r-r.r-r/r-r0r1r2r&r3rr4rrr)r)rr0r1r4rr/rG)rr)rrrrsr&r,rGrrrurrtrrr)r)rrsrrtrrr)rr)rrrr)rr)T)rrrrrr)rrG)rrrr&rr)rrrr%)rD __module__ __qualname____doc__r'__annotations__r:r<r=rXrZrYrjrqrVrWrrrrrr classmethodro __classcell__)rCs@rr%r%s.L &*,004-1 $#&*.W.W.W.W.W.W.W`bbbb8vvvv&5S5S5S5Sn*S*S*S*SXc"c"c"c"J        vvvv####J2222\`&&&&&PEEEE>>>>><<<[<<<<rsI""""""    RRRRRRRRRRRRRR======BBBBBB  8 $ $Z<Z<Z<Z<Z<")Z<Z<Z<Z<Z