c H dZddlZddlZddlZddlZddlZddlmZddlZedkrej dej ej dd ej ejej dZeej d kr4eed ezejd ejej d Zejej d ZejejcZZeej d krJeej d Zeej eeZeej eeZej!"d eeZ#ej!"deeZ$de%de$j&Dzee$ezz Z'ej deee#j(d ddddddde#j)f D]gZ*e*e#_*eZ+e#D]Z,ee+z Z-ej.deeze*z Z/ej de*e-eee-z e/e-z hde#_0ej dd ddddddde#j)f D]nZ*e*e#_*eZ+de#DZ1ee+z Z-ej.deeze*z Z/ej de*e-eee-z e/e-z ode#_0ej deee$j(e'd ddd dd!d"e$j)fD]gZ*e*e$_*eZ+e$D]Z,ee+z Z-ej.deeze*z Z/ej de*e-eee-z e/e-z hde$_0ej d#d ddd dd!d"e$j)fD]gZ*e*e$_*eZ+e$D]Z,ee+z Z-ej.deeze*z Z/ej de*e-eee-z e/e-z hde$_0eej ed"Z2ej d$ee2ee#e#j(d%D]Z*eZ+e*d kr)gZ1ej34e2e*D] Z5e#e5Z6 n e2D] Z7e#e7Z6 ee+z Z-ej.dee2ze*z Z/ej de*e-ee2e-z e/e-z eej ed"Z2ej d&ee2eee$j(e'd'D]Z*eZ+e*d kr)gZ1ej34e2e*D] Z5e$e5Z6 n e2D] Z7e$e7Z6 ee+z Z-ej.dee2ze*z Z/ej de*e-ee2e-z e/e-z ej d(edSdS))at USAGE: %(program)s CORPUS_DENSE.mm CORPUS_SPARSE.mm [NUMDOCS] Run speed test of similarity queries. Only use the first NUMDOCS documents of each corpus for testing (or use all if no NUMDOCS is given). The two sample corpora can be downloaded from http://nlp.fi.muni.cz/projekty/gensim/wikismall.tgz Example: ./simspeed2.py wikismall.dense.mm wikismall.sparse.mm N)time__main__z)%(asctime)s : %(levelname)s : %(message)s)formatlevelz running %s __doc__z /tmp/tstdensez/tmp/tstsparsegY@c#$K|] }|jV dS)N)num_nnz).0shards 5lib/python3.11/site-packages/gensim/test/simspeed2.py r0s$IIE%-IIIIIIzKtest 1 (dense): similarity of all vs. all (%i documents, %i dense features) @g?z6chunksize=%i, time=%.4fs (%.2f docs/s, %.2f queries/s) zTtest 2 (dense): as above, but only ask for the top-10 most similar for each documentcg|]}|Sr)rsims r rGs++++++rzVtest 3 (sparse): similarity of all vs. all (%i documents, %i features, %.2f%% density)dzUtest 4 (sparse): as above, but only ask for the top-10 most similar for each documentzStest 5 (dense): dense corpus of %i docs vs. index (%i documents, %i dense features))r rrrrrrrzftest 6 (sparse): sparse corpus of %i docs vs. sparse index (%i documents, %i features, %.2f%% density))r rrr r!r"zfinished running %s)8r loggingsys itertoolsosmathrgensim__name__ basicConfigINFOinfojoinargvpathbasenameprogramlenprintglobalslocalsexitcorporaMmCorpus corpus_dense corpus_sparse num_termsdense_featuressparse_featuresintNUMDOCSlistislice similarities Similarity index_dense index_sparsesumshardsdensity num_features shardsize chunksizestartrtakenceilqueriesnum_bestsimsqueryutilschunkize_serialchunk_vecrrrrXsg    zE1GJRYR^____GLsxx11222gsx{++G s38}}q ggii "VVXX-... >**38A;77LN++CHQK88M&2&U#NO s38}}qG#chqk""t,I,\7CCDD -Y-mWEEFF %00,P^__K&112BMSbccLccII\5HIIIIIISSQ]M^M^apMpqGGLU L;3BCc49NO   )   C $)C##l"3"33i?@@ D ucc,//%75    KGLghhhBCc49NO   ) ++{+++$)C##l"3"33i?@@ D ucc,//%75     KGL` ML5w BS#t\5KL   !*   C $)C##m"4"44y@AA D ucc-0058'E/    LGLhiiiBS#t\5KL   !*   C $)C##m"4"44y@AA D ucc-0058'E/    !L D!!,55 6 6EGL] E CC $$k&>9   q= %D55eYGG ' '& ' % %$$)C##e**,y899 D ucc%jj50'E/    D!!,55 6 6EGLp E CC && (A70   q= &D55eYGG ( ( ' ( & & %$)C##e**,y899 D ucc%jj50'E/    GL&00000KE1E1r