Αi%SrSSKrSSKrSSKrSSKJr /rSrSr Sr Sr Sr S r S rS rS rS rSrSrSrSrSSjr\"SSSSS9S5r\"SSSSS9S5r\"SSSSS9S5r\"SSSSS9S5rg)a^ Conll05 dataset. Paddle semantic role labeling Book and demo use this dataset as an example. Because Conll05 is not free in public, the default downloaded URL is test set of Conll05 (which is public). Users can change URL and MD5 to their Conll dataset. And a pre-trained word vector model based on Wikipedia corpus is used to initialize SRL model. N) deprecatedzBhttp://paddlemodels.bj.bcebos.com/conll05st/conll05st-tests.tar.gz 387719152ae52d60422c016e92a742fcz:http://paddlemodels.bj.bcebos.com/conll05st%2FwordDict.txt ea7fb7d4c75cc6254716f0177a506baaz:http://paddlemodels.bj.bcebos.com/conll05st%2FverbDict.txt 0d2977293bbb6cbefab5b0f97db1e77cz# [R"T5nURT5nURT5n[R"US9n[R"US9n/n/n/n[ X45GHSupUR 5R5nU R 5R5R5n [U 5S:XGa[[US55H(n UV s/sHoU PM n n URU 5 M* [U5S:Gau/n USHn U S:wdM U RU 5 M [USS5GH;upSnSn/nSnUGHnUS:XaU(dURS5 M$US:XaU(aURS U-5 MGUS :XaURS U-5 SnMeURS 5S :waAURS 5S :wa,USURS5nURSU-5 SnMURS 5S :waBURS 5S :Xa-USURS5nURSU-5 SnGM[SU35e X]U U4v GM> /n/n/nGM1URU5 URU 5 GMV SSS5 SSS5 UR5 UR5 UR5 gs sn f!,(df  NL=f!,(df  NU=f7f)N)fileobjrr-rF*r z*)()r TzUnexpected label: )tarfiler extractfilegzipGzipFileziprdecodesplitlenrangeappendrfind RuntimeErrorclose)tfwfpf words_file props_file sentenceslabelsone_segwordlabelrx a_kind_label verb_listlblcur_tag is_in_bracketlbl_seq verb_wordl data_path props_name words_names rreadercorpus_reader..readerPs \\) $ ^^J ' ^^J ' MM" % MM" %IFG":: zz|**, ,,.446u:?"3wqz?36='>g!g '> l346{a'$& !'A Cx ) 0 0 3"+'0qr &;FA&)G,1M&(G(*I%(#$8M$+NN3$7%&#X-$+NN4'>$B%&$Y$+NN4'>$B49M%&VVC[B%6166#;";L./AFF3K.@G$+NN4'>$B49M%&VVC[B%6166#;";L./AFF3K.@G$+NN4'>$B48M*69KA37O*P$P#&)&#,q\7"BB1'<4!#IF G$$T*NN5)] ; & &l      Y(? & % & %sPA L7L&%BL5L 5L# T"5GHupn[U5nURS5nS/[U5-nUS:aSXTS- 'XS- nOSnUS:aSXTS- 'XS- nOSnSXT'XnU[U5S- :aSXTS-'XS-n OSn U[U5S- :aSXTS-'XS-n OSn UV s/sHn TRU [5PM n n TRU[5/U-n TRU[5/U-nTRU[5/U-nTRU [5/U-nTRU [5/U-nTRU5/U-nUV s/sHn TRU 5PM nn U U UUUUUUU4 v GM gs sn fs sn f7f)NzB-Vrrbosr eos)r2rgetUNK_IDX)sentence predicater>sen_len verb_indexmarkctx_n1ctx_n2ctx_0ctx_p1ctx_p2wword_idx ctx_n2_idx ctx_n1_idx ctx_0_idx ctx_p1_idx ctx_p2_idxpred_idx label_idxrQ label_dictpredicate_dict word_dicts rrNreader_creator..readers+8? 'H(mGe,J3V$DA~'(!^$!q.1A~'(!^$!q.1 D (ECK!O+'(!^$!q.1CK!O+'(!^$!q.1;CD8a a18HD#--89GCJ#--89GCJ"ug67'AI#--89GCJ#--89GCJ&**956@H4:;Fq*FI;  W,;@EPlease use new dataset API which supports paddle.io.DataLoader)since update_tolevelreasoncr[[RRR [ S[ 55n[[RRR [S[55n[[RRR [S[55nXU4$)z> Get the word, verb and label dictionary of Wikipedia corpus. conll05st) r!paddledatasetcommondownload WORDDICT_URL WORDDICT_MD5 VERBDICT_URL VERBDICT_MD5r TRGDICT_URL TRGDICT_MD5)rn verb_dictrls rget_dictrs&&|[,OI&&|[,OI!&&{KMJ  ++rRch[RRR[S[ 5$)z8 Get the trained word vector based on Wikipedia corpus. rv)rwrxryrzEMB_URLEMB_MD5rPrRr get_embeddingrs# >> ) )'; HHrRc[5upn[[RRR [ S[5SSS9n[X0X5$)aP Conll05 test set creator. Because the training dataset is not free, the test dataset is used for training. It returns a reader creator, each sample in the reader is nine features, including sentence sequence, predicate, predicate context, predicate context flag and tagged sequence. :return: Training reader creator :rtype: callable rvz2conll05st-release/test.wsj/words/test.wsj.words.gzz2conll05st-release/test.wsj/props/test.wsj.props.gz)rMrL) rrQrwrxryrzDATA_URLDATA_MD5rp)rnrrlrNs rtestrsK$(0z$I* &&xhGGGF &Y CCrRc[RRR[S[ 5 [RRR[ S[5 [RRR[S[5 [RRR[S[5 [RRR[S[5 g)Nrv)rwrxryrzr{r|r}r~rrrrrrrPrRrfetchrs NN""<lK NN""<lK NN""; [I NN""7KA NN""8[(CrR)NNN)__doc__r-r+paddle.dataset.commonrw paddle.utilsr__all__rrr{r|r}r~rrrrrXrr!rQrprrrrrPrRrrs  #  O -K 1 K 1 L 0 = ,  $ HXDH;| . K  ,  ,  . K  I  I . K  D  D* . K  D  DrR