h4,SSKrSSKrSSKJr SSKJrJrJrJrJ r J r J r SSK J r JrJrJr SSKJr SSKJr SSKJr SS KJrJrJr SS KJr S S KJr S r\ "5RA\5Sr!"SS\5r""SS\5r#S\#S\$4Sjr%Sr&Sr'g)N)islice)AnyCallableDictIterableListOptionalSequence)ConfigModel Optimizerset_dropout_rate)Errors)Language)Doc)Examplevalidate_examplesvalidate_get_examples)Vocab) TrainablePipez [model] @architectures = "spacy.HashEmbedCNN.v2" pretrained_vectors = null width = 96 depth = 4 embed_size = 2000 window_size = 1 maxout_pieces = 3 subword_features = true modelc V\rSrSrSrS"S\S\S\SS4Sjjr\ S\ S 4S j5r \ S\ \4S j5r S S S \SS4Sjr S S S \S\4SjrS#SjrS\\4SjrS\\SS4SjrSSSS.S\\S\S\\S\\\\44SjjrS#SjrSS.S\/\\4S\\4SjjrS r S!r!g)$Tok2VecaApply a "token-to-vector" model and set its outputs in the doc.tensor attribute. This is mostly useful to share a single subnetwork between multiple components, e.g. to have one embedding and CNN network shared between a parser, tagger and NER. In order to use the `Tok2Vec` predictions, subsequent components should use the `Tok2VecListener` layer as the tok2vec subnetwork of their model. This layer will read data from the `doc.tensor` attribute during prediction. During training, the `Tok2Vec` component will save its prediction and backprop callback for each batch, so that the subsequent components can backpropagate to the shared weights. This implementation is used because it allows us to avoid relying on object identity within the models to achieve the parameter sharing. vocabrnamereturnNcDXlX lX0l0Ul0Ulg)a[Initialize a tok2vec component. vocab (Vocab): The shared vocabulary. model (thinc.api.Model[List[Doc], List[Floats2d]]): The Thinc Model powering the pipeline component. It should take a list of Doc objects as input, and output a list of 2d float arrays. name (str): The component instance name. DOCS: https://spacy.io/api/tok2vec#init N)rrr listener_mapcfg)selfrrrs P/home/james-whalen/.local/lib/python3.13/site-packages/spacy/pipeline/tok2vec.py__init__Tok2Vec.__init__-s"   @B#%Tok2VecListenercvURVVs/sHoRUHo"PM M snn$s snnf)zeRETURNS (List[Tok2VecListener]): The listener models listening to this component. Usually internals. )listening_componentsr!)r#cms r$ listenersTok2Vec.listeners>s7  44S4a>O>OPQ>R>R4SSSs!5cH[URR55$)z_RETURNS (List[str]): The downstream components listening to this component. Usually internals. )listr!keys)r#s r$r*Tok2Vec.listening_componentsEs D%%**,--r'listenercomponent_namecURRU/5 XRU;aURURU5 gg)z=Add a listener for a downstream component. Usually internals.N)r! setdefaultappendr#r3r4s r$ add_listenerTok2Vec.add_listenerLsG $$^R8 ,,^< <   n - 4 4X > =r'cX R;aRXRU;a@URURU5 URU(d URU gg)z@Remove a listener for a downstream component. Usually internals.TF)r!remover8s r$remove_listenerTok2Vec.remove_listenerRsZ .. .,,^<<!!.188B((8)).9r'c,SUR4n[[USS5[5(afURR 5HGn[U[ 5(dMURU;dM,URX1R5 MI gg)a|Walk over a model of a processing component, looking for layers that are Tok2vecListener subclasses that have an upstream_name that matches this component. Listeners can also set their upstream_name attribute to the wildcard string '*' to match any `Tok2Vec`. You're unlikely to ever need multiple `Tok2Vec` components, so it's fine to leave your listeners upstream_name on '*'. *rN) r isinstancegetattrr rwalkr( upstream_namer9)r# componentnamesnodes r$find_listenersTok2Vec.find_listeners]sqdii  gi$7 ? ?!,,.dO449K9Ku9T%%dNN;/ @r'docsc[SU55(dRURRS5nUVs/sH)o0RRR SU45PM+ sn$URR U5nU$s snf)aApply the pipeline's model to a batch of docs, without modifying them. Returns a single tensor for a batch of documents. docs (Iterable[Doc]): The documents to predict. RETURNS: Vector representations for each token in the documents. DOCS: https://spacy.io/api/tok2vec#predict c38# UHn[U5v M g7fN)len.0docs r$ "Tok2Vec.predict..us,t3s88tnOr)anyrget_dimopsallocpredict)r#rJwidthrQtokvecss r$rZTok2Vec.predictlsr,t,,,JJ&&t,E@DEJJNN((!U4E E**$$T*Fs0Bct[X5H)up4URS[U5:XdeXClM+ g)zModify a batch of documents, using pre-computed scores. docs (Iterable[Doc]): The documents to modify. tokvecses: The tensors to set, produced by Tok2Vec.predict. DOCS: https://spacy.io/api/tok2vec#set_annotations rN)zipshaperNtensor)r#rJ tokvecsesrQr\s r$set_annotationsTok2Vec.set_annotations|s5 0LC==#s3x/ // J1r')dropsgdlossesexamplesrfrgrhc^^^^ ^ ^ ^Tc0m[US5 UVs/sHoURPM nn[TRU5 TRR U5umm TVs/sH0nTRR R "UR6PM2 snm TRTRS5 U UUU4Sjm U U U UU4Sjn[RU5n TRSSHn U RU TT 5 M TR(a TRSRU TU5 T$s snfs snf)aLearn from a batch of documents and gold-standard information, updating the pipe's model. examples (Iterable[Example]): A batch of Example objects. drop (float): The dropout rate. sgd (thinc.api.Optimizer): The optimizer. losses (Dict[str, float]): Optional record of the loss during training. Updated using the component name as the key. RETURNS (Dict[str, float]): The updated losses dictionary. DOCS: https://spacy.io/api/tok2vec#update NzTok2Vec.updaterecD>[[U55HEnTU==X- ss'TTR==[XS-R 55- ss'MG TVs/sH0nTR R R"UR6PM2 sn$s snf)zAccumulate tok2vec loss and gradient. This is passed as a callback to all but the last listener. Only the last one does the backprop. r) rangerNrfloatsumrrXalloc2fr`) one_d_tokvecsit2v d_tokvecsrhr#r\s r$accumulate_gradient+Tok2Vec.update..accumulate_gradients 3}-.! 00 tyy!UM,<,A+F+F+H%II!/CJJ'3DJJNN**CII6'J JJs#7BcP>T"U5 T"T5nTbTRT5 U$)z>Callback to actually do the backprop. Passed to last listener.) finish_update)rpd_docsrt bp_tokvecsrsr#rgs r$backprop Tok2Vec.update..backprops-  . *F""3'Mr')r predictedrr begin_updaterXror`r6rr( get_batch_idr-receive)r#rirfrgrhegrJrrrzbatch_idr3rtryrsr\s` `` @@@@r$updateTok2Vec.updates( >F($45'/0x x0T*"jj55d;CJK7CTZZ^^++SYY77K $))S) K K  #//5s+H   Xw0C D, >> NN2  & &x( C ;1Ls E,7E cgrM)r#riscoress r$get_lossTok2Vec.get_losss r')nlp get_examplesrc[US5 /n[U"5S5HnURUR5 M U(d,[R R URS95eURRUS9 g)aDInitialize the pipe for training, using a representative set of data examples. get_examples (Callable[[], Iterable[Example]]): Function that returns a representative sample of gold-standard Example objects. nlp (Language): The current nlp object the component is part of. DOCS: https://spacy.io/api/tok2vec#initialize zTok2Vec.initialize r)XN) rrr7xrE923formatrr initialize)r#rr doc_sampleexamples r$rTok2Vec.initializesq l,@A lnb1G   gii (2=6;;--499-==z  +r'c[erM)NotImplementedError)r#labels r$ add_labelTok2Vec.add_labels!!r')r"r!rrr)tok2vec)rN)"__name__ __module__ __qualname____firstlineno____doc__rr strr%propertyrr-r*r9boolr=rHrrrZr rcrrmr r rrrrrrr__static_attributes__rr'r$rrsX &e&E&&T&"T4 12TT .d3i.. ?%6??PT? (9 3 SW  <HSM !HSM ! ! #'-1 47#4 4 i 4 c5j)* 4l #' ,r8G#445,h  ,,"r'rct\rSrSrSrSrS\S\SS4Sjr\ S \ \ S\4S j5r S \SS4S jr S\4S jrSrg)r(aA layer that gets fed its answers from an upstream connection, for instance from a component earlier in the pipeline. The Tok2VecListener layer is used as a sublayer within a component such as a parser, NER or text categorizer. Usually you'll have multiple listeners connecting to a single upstream Tok2Vec component, that's earlier in the pipeline. The Tok2VecListener layers act as proxies, passing the predictions from the Tok2Vec component into downstream components, and communicating gradients back upstream. ztok2vec-listenerrDr[rNc[R"XR[SU0S9 XlSUlSUlSUlg)a upstream_name (str): A string to identify the 'upstream' Tok2Vec component to communicate with. The upstream name should either be the wildcard string '*', or the name of the `Tok2Vec` component. You'll almost never have multiple upstream Tok2Vec components, so the wildcard string will almost always be fine. width (int): The width of the vectors produced by the upstream tok2vec component. rU)rforwarddimsN)r r%rrrD _batch_id_outputs _backprop)r#rDr[s r$r%Tok2VecListener.__init__s7 t))WD%=Q*(, r'inputsc&[SU55$)zzCalculate a content-sensitive hash of the batch of documents, to check whether the next batch of documents is unexpected. c3F# UHn[SU55v M g7f)c38# UHoRv M g7frM)orth)rPtokens r$rR9Tok2VecListener.get_batch_id...s3sezzsrTNrnrOs r$rR/Tok2VecListener.get_batch_id..sFv33s333vs!r)clsrs r$rTok2VecListener.get_batch_ids FvFFFr'rc(XlX lX0lg)zStore a batch of training predictions and a backprop callback. The predictions and callback are produced by the upstream Tok2Vec component, and later will be used when the listener's component's model is called. N)rrr)r#routputsrzs r$rTok2VecListener.receives " !r'c URc&URc[[R5eUR U5nX R:wa0[[R RX RS95eg)zOCheck that the batch of Doc objects matches the ones we have a prediction for. )id1id2T)rr ValueErrorrE954rE953r)r#rrs r$ verify_inputsTok2VecListener.verify_inputs sd >> !dmm&;V[[) )((0H>>) !3!3nn!3!UVVr')rrrrD)rrrrrrrintr% classmethodrrrrrrrrr'r$r(r(sq  Dc#$ G(3-GCGG ""4" t r'r(is_traincvU(aURcn/nUH^nURRS:Xa&[[R R SS95eURUR5 M` U[4$URU5 URUR4$/nURS5nUHnnURRS:Xa6URURR[U5U55 MSURUR5 Mp U[4$)z7Supply the outputs from the upstream Tok2Vec component.rrrrU)rrasizerrE203rr7_empty_backproprrrrWrXrorN)rrrrrQr[s r$rrs ?? "G::??a'$V[[%7%7Y%7%GHHNN3::.  O+ +    '>>5??2 2 d#Czz!#uyy00S5ABszz*''r'c/$rMr)dXs r$rr<s Ir'c|US:Xa"[R"S5nUR$[S[SU35e)N make_tok2veczspacy.pipeline.factorieszmodule z has no attribute ) importlib import_modulerAttributeErrorr)rmodules r$ __getattr__rAs@ ~(()CD""" 78*,>tfE FFr')(rsys itertoolsrtypingrrrrrr r thinc.apir r r rerrorsrlanguagertokensrtrainingrrrrrtrainable_piperdefault_model_configfrom_strDEFAULT_TOK2VEC_MODELrr(rrrrrr'r$rs JJJ@@HH) ))*>?H{"m{"|9e9x"(?"(d"(J Gr'