h"LSSKrSSKJr SSKJrJrJr SSKrSSKJ r J r J r J r J r SSKJr SSKJr SSKJr SS KJr SS KJr SS KJr S S KJrJrJrJrJrJrJ r \RB"SSSS.S9\"SSSSS9\"SSS9\"SSSSS9\"SSSSS9\"SSS S!S9\"SS"S#S$S9\"SS%S&S'S9\"SS(S)S*S9\"SS+S,S-S9\"SS.S/S0S9\"SS1S2S3S9\"S4S5S6S7S94 S8\RDS9\S:\#S;\#S<\$S=\$S>\$S?\$S@\$SA\$SB\$SC\$SD\%4SEjj5r&SSF.SG\\\#\44SHjjr'SIr(SOSJ\#4SKjjr)SLr*SMr+SNr,g)PN)Path)AnyDictOptional)Modeldata_validationfix_random_seedset_dropout_rateset_gpu_allocator)msg)Example)resolve_dot_names)util)ConfigSchemaTraining)registry)ArgOpt debug_cliparse_config_overrides setup_gpushow_validation_errorstring_to_listmodelT)allow_extra_argsignore_unknown_options)context_settings.zPath to config file)helpexists allow_dashzDName of the pipeline component of which the model should be analysed)rz--layersz-lz+Comma-separated names of layer IDs to printFz --dimensionsz-DIMzShow dimensionsz --parametersz-PARzShow parametersz --gradientsz-GRADzShow gradientsz --attributesz-ATTRzShow attributesz --print-step0z-P0zPrint model before trainingz --print-step1z-P1z Print model after initializationz --print-step2z-P2zPrint model after trainingz --print-step3z-P3zPrint final predictionsz--gpu-idz-gzGPU ID or -1 for CPUctx config_path componentlayers dimensions parameters gradients attributesP0P1P2P3use_gpuc [U 5 [USS9nUUUUUUU U U S. n [UR5n[ U5 [ R "XSS9nSSS5 WR5nUSSnU S :aU(a [U5 [ U5 [ R"U5nURR5n[R"US[S 9nSSS5 WS nUb$[R"S U35 [!U5 WR#U5n[%UUUUU S 9 g!,(df  N=f!,(df  Nj=f)z Analyze a Thinc model implementation. Includes checks for internal structure and activations during training. DOCS: https://spacy.io/api/cli#debug-model T)intify) r(r)r*r+r'print_before_trainingprint_after_initprint_after_trainingprint_predictionF) overrides interpolateNtraining gpu_allocatorr)schemaseedzFixing random seed: print_settings)rrrargsrr load_configr8r load_model_from_configconfigrresolverr infor get_pipe debug_model)r$r%r&r'r(r)r*r+r,r-r.r/r0r>config_overrides raw_configrB allocatornlpTr<pipes O/home/james-whalen/.local/lib/python3.13/site-packages/spacy/cli/debug_model.pydebug_model_clirN sA6g F4 0F   !# " N.chh7 { +%%   , # # %Fz"?3I!| )$ { +))*5'')   VJ/8L M , V9D  'v./ << "D3^D% , + , +sD;A E ; E  Er=r>c >^^ ^ [US5(d[R"SUS3SS9 URn[ U[ 5(d"[R"S[ U5S3SS9 Uc0n[R"SUR35 URS 5(a![R"S 5 [XT5 [S 5 US /n[5 [X5um TRUU 4S j5 SSS5 [R"S5 [!["R$"T "T5S55m SSS5 URS5(a![R"S5 [XT5 [3US5 SnUR5S5(a0SUR7S5R8;aTR;S5nUR5S5(a0SUR7S5R8;aTR;S5n[=S5H,n U(aUR?T 5 UR?T 5 M. URS5(a![R"S5 [XT5 URAT V s/sHoRBPM sn 5n URS5(a5[R"S 5 [R"[EU 55 [RF"S!5 g!,(df  GN#=f![&a [)SUS9 [5 [+5Vs/sHn[,R."U05PM Os snfsnm TRU 4Sj5 SSS5 O!,(df  O=f[R"S5 GN![0a [R"SSS9 GNf=ff=f!,(df  GN=fs sn f)"NrzThe component 'z0' does not specify an object that holds a Model.r)exitsz1Requires a Thinc Model to be analysed, but found z instead.zAnalysing model with ID r3zSTEP 0 - before trainingF train_corpusc>T"T5$N)rJrQsrMdebug_model..s |C'8z/Initialized the model with the training corpus.)nOrc>T$rSrT)examplessrMrUrVs8rWz&Initialized the model with dummy data.zjCould not initialize the model: you'll have to provide a valid 'train_corpus' argument in the config file.r4zSTEP 1 - after initializationg?tok2vecztok2vec-listenerztransformer-listener transformerr5zSTEP 2 - after trainingr6zSTEP 3 - predictionz.Succesfully ended analysis - model looks good.)$hasattrr failr isinstancertyperDidgetdivider _print_modelrrr initializelist itertoolsislice ValueError_set_output_dim _get_docsr from_dict Exceptionr has_refget_refnamerErangeupdatepredict predictedstrgood)rBresolved_train_configrJrLr>r dot_namesxupstream_componenteex predictionr\rQs ` @@rMrFrF^s& 4 ! ! dV#S T  JJE eU # # ?U }I V HH'z 23122 .0U+   .~>?I&("3F"F89) HHF GI,,\#->BCH *,-- 35U+UC  }}Y$6%-- :R:W:W$W \\)4 i  "emmI&>&C&C C \\-8 1X   % %h / H011 -/U+x@x x@AJ,-- )+ Z!HH=?_)(   1E2*,BK+N+Q 1 1!R 8+NHNN#34-,,AB A   Z AsPL?-$L-AL?/P- L< 7L?? P O N3+!N N3* O3 O =OPP<P=PPPP Pc /SQ$)N)z6Apple is looking at buying U.K. startup for $1 billionz>Autonomous cars shift insurance liability toward manufacturersz8San Francisco considers banning sidewalk delivery robotsz+London is a big city in the United Kingdom.rTrTrWrM _sentencesrs  rWlangc|[R"U5"5n[UR[ 555$rS)rget_lang_classrirLr)rrJs rMrnrns+   d # %C & ''rWcURS5cURSU5 URS5(aDURS5RS5c"URS5RSU5 ggg)NrZ output_layer)has_dimset_dimrqrr)rrZs rMrmrmsk }}T" dB }}^$$ == ( 0 0 6 > MM. ) 1 1$ ; ?%rWc URSS5nURSS5nURSS5nURSS5nURSS5n[UR55GHupxU(aXr;dM[R"SUS UR S UR S 35 U(a>URH.n [R"S U S URU 535 M0 U(aURH|n URU 5(a8[URU 55n [R"SU S U 35 MQ[R"SU S URU 535 M~ U(aURH|n URU 5(a8[URU 55n [R"SU S U 35 MQ[R"SU S URU 535 M~ U(dGMURn U R!5H!up[R"SU S U 35 M# GM g)Nr'r"r)Fr(r*r+zLayer z : model ID z: ''z - dim z: z - param z - grad z - attr )re enumeratewalkr rDrdrs dim_names maybe_get_dim param_names has_param _print_matrix get_paramhas_gradget_gradattrsitems) rr>r'r)r(r*r+inoders print_valuervalues rMrgrgs   " -F##L%8J##L%8J"";6I##L%8JUZZ\* HHvaS DGG9C {!D E NNDHHwtfBt/A/A$/G.HIJ+ ,,D~~d++&3DNN44H&I 9TF"[M!BC9TF"T^^D5I4J!KL -  ,,D}}T**&3DMM$4G&H 8D6K=!AB8D6DMM$4G3H!IJ - z #(;;=KDHHxvRw78$1/+rWcUb[U[5(aU$[UR5S-nUn[ UR S- 5HnUSnM USSnU[U5-nU$)Nz - sample: rrrX)rbboolrxshapertndim)rresult sample_matrixds rMrrsq } 5$//   -FM 5::> "%a( #!!A&M c-( (F MrW)en)-rjpathlibrtypingrrrtyper thinc.apirrr r r wasabir spacy.trainingr spacy.utilrr"rschemasrr_utilrrrrrrrcommandContextrxrintrNrFrrnrmrgrrTrWrMrs&& "(*  *.$OC&;DUYZ#ijb*d1^_5.&?PQ5.&?PQ%>NO5.'@QR5/57TU5/57YZ5/57ST5/57PQr:t2HI7E 7E7E 7E  7E  7E7E7E7E 7E 7E 7E 7E7E 7E@04 O@ T#s(^, O@d(C( <9D rW