
    pj#                    z   0 S S SS0SS0S._SSSSSS	.S
SSS.S._SSSS0SS0S._SSSS0SS0S._SSSSSSSSS.SSSSS.S._SSSSS SSS!.SS"SS#.S._S$S%SSSS&S'S(S)S*SS+S,.
SS-S.S/S0S1SS2.S3S4S5.S3S6S5.S3S7S5.S8.S9._S:S;SSSS S<S)SS=.SS"S>SS.S._S?S@SSSS&S'S(S)S*SS+S,.
SS-S.S/S0S1SS2.S3S4S5.S3S6S5.S3S7S5.S8.S9._SASBSSCSD.SS0SESFSGSH.SI._SJSKSLS0SS0S._SMSMSS0SS0S._SNSNSOSSCSP.SS0SQSRSS.SI._STSTSSUSVSW.SS0S._SXSXSYSZ0S[S\0S]S^S_S5.0S`._SaSbSS0SS0ScSdSe.SI._SfSfSOSgSD.SS0ShSiS5.ShSjS5.Sk.S9._0 SlSmSS0SS0ScSdSe.SI._SnSoSS0SS0ScSdSe.SI._SpSpSYSO0SqSrSs.S._StStSS0SS0S._SuSvSS0SS0ScSdSe.SI._SwSxSySzS{S|.SS0S}S~S.S}S~S.S}S~S.S|.S9._SSSS0SS0SSS~S.0S9._SSSS0SS0SSS~S.0S9._SSSS0SSS.S._SSSS0SSS.S._SSSS0SS0S._SSSS0SS0S._SSSS0SS0S._SSSS0SS0S._SSSSS.S._SSSS0SS0S._SSSS0SS0S._E0 SSSS0S._SSSS0SS0S._SSSSSD.SS0S._SSSS0SS0S._SSSSSSSSSSSS.	SSSSSSS.S._SSSSSSSS.SSSS.S._SSSOSSCSP.SS0SQSRSS.SI._SSSS0SS0S._SSSOSSVS.SS0S._SSSSSSUSVS.SCSSSSSS.SS0SI._SSSSS.SS0S._SSSSS.SSS.S._SSSS0SS0S._SSSSSSUS.S[S0SS0SI._SSSSCSSVS.SS0S._SSSSSSUS.S[S0SS0SI._SSSSS.SS0SS^S~S.0S9._E0 SSSS0SSS.S._SSSSCSD.SS0S._SSSS0SS0S._SSSSCSD.SS0S._SSSSCSD.SS0S._SSSSCSD.SS0S._SSSS0S._SSSSSS.SS0S._SSSYS0SS0S._SSSS0SS0S._SSSSS.SS0S._SSSSS.SSGS .S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSSUGS.SGSGS	.S._E0 GS
GS
SS0SS0S._GSGSSSCSD.SS0S._GSGSGSS0GSGSGS.S._GSGSSS0SS0S3GSS5.S3GSS5.GS.S9._GSGSSS0SS0S._GSGSSYSO0S\GSGS.GSGS0SI._GSGSGSGSGSGS.GS GS!GS".GS#GS$0SI._GS%GS%SSCGS&.SS0S._GS'GS'SS0SS0SS0SS^GS(S5.0GS)._GS*GS+GS*0_GS,GS,SS0SS0S._GS-GS-SOGS.GS/.SS\0S._GS0GS0SOGS.SVGS1.SS\0GS2S^S~S.0S`._GS3GS3SOGS.SVGS1.SS\0GS2S^S~S.0S`._GS4GS4SOGS.GS/.SS\0S._GS5GS5SOGS.GS6.SS\0S._GS7GS7GS8GS9GS:.SS\0S._E0 GS;GS;SS0SS0S._GS<GS<SS0SS0S._GS=GS>SS0SS0S^GS?GS@GSA.S^GSBGSCGSA.GSD.S`._GSEGSESSCSD.SS0SSc0SI._GSFGSGSSS.SSGS .S._GSHGSHSS0SS0ScSc0SI._GSIGSISS0SS0SS^S~S.0S9._GSJGSJSGSKGSL.SSC0S._GSMGS+GSM0_GSNGSNSS0SS0S._GSOGSOSOGSPGS.GSQGSR.SS\0S._GSSGSSSOGS.GS/.SS\0GSTSGGSUGSVGSW.SI._GSXGSXSOGS.SVGS1.SS\0GS2S^S~S.0S`._GSYGSYSYSO0S[S\0SUGSZGS[.SI._GS\GS\SSUGS.SS0S._GS]GS]SSUGS^GS_GS`GSa.SCGSbGScGSdGSe.S._GSfGSfSGSgGSh.SS0S._E0 GSiGSjSSO0SS0S._GSkGSkSS0SS0S._GSlGSlSSGSmSSGSnGSoGSp.SGSqSGSrGSs.S._GStGSuSS0SS0S._GSvGSvSYSO0SS0S._GSwGSwSSO0SS0S._GSxGSxSS0SS0S._GSyGSyGSzSQ0SS0S._GS{GS{SSCSD.SS0S._GS|GS|SS0SS0S3GS}S5.S3GS~S.GS.S9._GSGSSSCSD.SS0S._GSGSSSCSD.SS0S._GSGSSGSGS.SGSQGS.GSGSGSGSGSGS.GSGSS~00GS)._GSGS+GS0_GSGSGSGSGSGSGS.GSSGS.S._GSGSSS0GSGSGS.S._GSGSSS0GSGSGS.S._E0 GSGSSS0SS0S._GSGSSS0GSGSGS.GSGS0SI._GSGSSGS0SGSGSGS.S._GSGSSSCSD.SS0S._GSGSSS0SS0S._GSGSSSGS.SS0GSGS0SI._GSGSSS0GSSGS@GSGSA.0GS._GSGSSSCSD.SS0S._GSGSSSCSD.SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0GSGS0GSS^GS?GSGSA.0GS._GSGSSSCSD.SS0S._GSGSSS0SS0S._GSGSSS0SS0GSGS0SI._GSGSSS0GSS~0GSS~0GS.GS._E0 GSGSSGSGS.SS0S._GSGSSS0SGSbGS.S._GSGSSS0SGSbGS.S._GSGSSSUGS.SS0S._GSGSSS0SGSbGS.S._GSGSSS0SGSbGS.S._GSGSSGSGS_GS`GS.SGSbGSdGScGS.S._GSGSSGSGS_GS`GS.SGSbGSdGScGS.S._GSGSSGSGSGS.SGSbGSGS.S._GSGSGSGSGSGS.GSGSGS.S._GSGSSOSSVS.SS0GSTSGGSVGS.SI._GSGSSS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0GSS3S~S.0S9._E0 GSGSSS0SS0S._GSGSSSCSD.SS0S._GSGSSS~S.SS~S.GS.GS._GSGSSS~S.SS~S.GS.GS._GSGSSS~S.SS~S.GS.GS._GSGSSS0SGSGS.GSGSGS.SI._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSSCSD.SS0S._GSGSSSCSD.SS0S._GSGSSSCSD.SS0S._GSGSSS0SSC0S._GS GS SS0SS0S._GSGSSS0SS0ScSdSe.SS^S~S.0GS._GSGSSS0S._GSGSSS0SS0GSS3S~S.0S9._GSGSGSSS~S.0GS._E0 GSGS	GS
GSSGSGSGSGSGSGSS+GS.
GSGSGSGSGSGSGSGS.S._GSGSSGSGSGSGSGS GS!GS"GS#GS$GS%GS&.0 GS'GS(_GS)GS*_GS+GS,_GS-GS._GS/GS0_GS1GS2_GS3GS4_GS5GS6_GS7GS8_GS9GS:_GS;GS<_GS=GS>_GS?GS@_GSAGSB_GSCGSD_GSEGSF_GSGGSH_GSIGSJSCGSK.ES._GSLGSLSSSSUSVS.SCSSSSSS.S._GSMGSMSGSNSVGSGSGSO.GSIGSDGS(GS*SCGSP.S._GSQGSRSGSSSSSUSVGST.SCSSSSSS.S._GSUGSUSOGS.SVGSVGSW.S[S\0GSTSGGSUGSVGSW.SI._GSXGSXSOGS.SVGSVGSY.S\GSZGS[.S._GS\GS\SOGS.SVGSVGSW.S[S\0GSTSGGSUGSVGSW.SI._GS]GS]SSCSD.SS0S._GS^GS^SSCSD.SS0S._GS_GS_SSCSD.SS0S._GS`GS`SSCSD.SS0S._GSaGSaSGSbSVSUGSc.SS0S._GSdGSdSGSeGSfSVSSGSg.SSGShGSiGSjGSkGSlGSmGSn.S._GSoGSoSSSCGSpSUGSqGSr.SSSGSs.S._GStGStSGSuGSvGSwGSxGSyGSzGS{GS|GS$GS%GS}.SGS~GSGSGSGSFGSHGSGSGSGSGS.GSGSGSGSGS.SI._GSGSGSGSGSGSGSGSGSGS!GSGSGSGS#GS.GSGSGSGSGS8GSGS@GSSGS.	S._E0 GSGSSSCSVGS.SSGS.S._GSGSSSCSGSGS.GSGSGSGS.S._GSGS0 SS_GSGS_GSGS_GSGS_GSGS_GSGS _GSGS_GSGS!_GSGS"_GSGS#_GSGS_GSGS_GSGS_GSGS_GSGS_GSGS_GSGS_GSGSGSGS.EGSJSGS.S._GSGSSGSKGSL.SS0S._GSGSSS0SS0S._GSGSSSGSGSfSVGS.GSlGShGSiGSSGS.GSGSGSVGS.SI._GSGSSSGSGSfSVSGS.SSSGShGSiGSjGSkGSlGSmGSGS.
GSGSGSVGS.SI._GSGSSSSVGS.GSSGS.S._GSGSSGS.SVGS1.SGSGS.GSGSGSGS.SI._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSGSGS.SS0SS^GSS5.0S9._GSGSSGSGS.SS0S._GSGSSGSGS.SS0S._GSGSSS0GSSGS@GSGSA.0GS._GSGSSS0SS0S._GSGSSGSGSSqSrGS.GSGSGSGS.GSGS$GS.SI._E0 GSGSSS0SS0S._GSGSSS0SS0S._GSGSSSCSD.SS0S._GSGSGSGSSGSGS.GS GSGSGSGSGS.S._GSGSGSGSSGSGSGS.SGS	GSGS
.S._GSGSSSCSD.SS0S._GSGSSSCSD.SS0S._GSGSSGSGS.SS\0S._GSGSSSUSVSW.SCSSGS.SS0SI._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0GSGS0SI._GSGS SSCSD.SS0S._GS!GS!SGS"GS#.GS$S0S._E0 GS%GS%SSGS&.SS0S._GS'GS(SSSSVGS)GS*GS+.SGS,GS-GS..S._GS/GS/SSCS.SGSNGS0.S._GS1GS1SSCSD.SS0S._GS2GS2SS0SS0S._GS3GS3SS0SS0S._GS4GS4SGSGS5GS6.SS0S._GS7GS7SGS8GS9GS:GS;GS<GS=.SS0GS>GS?GS@GSA.SI._GSBGS+GSB0_GSCGSCSS0S._GSDGSDSGSGS.SS0S._GSEGSESGSGS.SS0SSc0SI._GSFGSFSSUSVSW.SCSSGSG.S._GSHGSHSSO0SS\0S._GSIGSISS0SS0S._GSJGSJSOSgSD.SS0ShSiS5.ShSjS5.Sk.S9._GSKGSLSS0SS0S._E0 GSMGSNSS0SS0S._GSOGSPSS0SS0S._GSQGSQSGSRGSS.SS0S._GSTGSTSSCSD.SS0S._GSUGSUSS0SSS.S._GSVGSVSS0SS0S._GSWGSWSGSXGSY.SS0S._GSZGS[SSSS&S'S)S*SS+GS\.	SS-S.S0S1SGS].S._GS^GS^SSUSVSW.SCSSGSs.S._GS_GS_SS0SS0GS`GSz0SI._SGSaSSSSUS.S[S0SS0SI._GSGSbSS0SS0GSGSc0GSS^GSdGSeGSA.0GS._GSGSfSGSGSgSqSrGSh.GSGSGSGS.GSGS$GS.SI._GSiGSjSSCGSGSkGSlGSm.SGSGSGSn.GSoGSpGSq.SI._GSrGSsSSSSUS.S[S0SS0SI._GStGSuSS0SGSGS.GSS^GS?GS@GSA.0S`._GSvGSvSSCSGSw.SS0S._E0 GSxGSxSSCSD.SS0S._GSyGSySSCSD.SS0S._GSzGSzSS0SS0S._GS{GS|SSSSUS.S[S0SS0SI._GS}GS+GS}0_GS~GS~SyGSGSGS.SS0S._GSGSSSCSD.SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSGSGS"S.SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSSCSD.SS0S._GSGSSS0SS0S._GSGSSSCSD.SS0S._E0 GSGSSSCSD.SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSSGS.SS0S._GSGSSS0SGSGS.S._GSGSSSCSD.GSGSGSGSGS.GSS3S~S.0S9._GSGSSGSSD.GSGSGSGS.S._GSGSSSS.SSGS .S._GSGSSGSQGS.SSC0S._GSGSSSCSGS.SGSGS.S._GSGSiSSCSD.SS0GSGSGSq.SI._GSGSSSCSD.SS0S._GSGSGSSGS.SGSGS!GS.S._GSGSSS0SS0S._GSGSSGSGS.SS0S._GSGSSS0SS0ScSdSe.SS^S~S.0GS._E0 GSGSSS0SGSQGS.GSGS0SI._GSGSSS0SGSQGS.GSGS0SI._GSGSSSCSD.SS0S._GSGSSS0SS0S._GSGSSS0SS0ScSdSe.SS^S~S.0GS._GSGSSS0SS0S._GSGSSS~S.SS~S.GS.GS._GSGSSS0SS0S._GSGSSSSS&S'S(S)S*SGS.	SS-S.S/S0S1SS2.S3S~S.S3S~S.S3S~S.S8.S9._GSGSSSGSmSSGS.SGSqSGS.S._GSGSSLS0SS0S._GSGSSS0SS0ScSdSe.SS^S~S.0GS._GSGSSSCSD.SS0S._GSGSSGSGS.SS0S._GSGSSS0SSS.S._GSGSSSGSmSSGS.SGSqSGS.S._GSGSSGS_GS`GS.SGSbGSdGScGS.S._E0 GSGSSS0SS0S._GSGSSGSGSfSVGSVGS.GSS0GSGSGS.SI._GSGSGSSGS.SS0S._GSGSGSSGS!GS.SGSGSGS.S._GSGSSOSSVGSGS.SS0GSGSGSGS.SI._GSGSSS0SS0GSS^S~S.0S9._GSGSSSGS.SS0S._GSGSSSCSD.SS0S._GSGSSGSGS.SS0S._GSGSSS0SGSGS.SGSS^00S`._GSGSrSSSSUS.S[S0SS0SI._GSGSSSSGS.SGSGS.S._GSGSSGS0SGS0GSGS0SI._GS GSGSGS0SS0S._GSGSSS0SGSGS.S._GSGSSSCSD.SS0S._GSGS	SSO0GSS0S._E0 GS
GSSS0SS0GSS^GSS5.0S9._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0GSS3S~S.0S9._GSGS+GS0_GSGSSS0SS0GSGS0GSS^GSS5.0GS._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0GSGS0GSS^S~S.0GS._GSGSSS0SS0GSGS0SI._GSGSSS0SS0GS GS!0GS S3GS"S5.0GS)._GS#GS#SSQGS$.SS0S._E0 GS%GS%GS&GS'0SS0S._GS(GS(SOGS)GS*.GSSrGS+.S._GS,GS-SS0SS0ScSdSe.SS^S~S.0GS._GS.GS.SGS/GS!GS0.SS0S._GS1GS1SOGSGS2GS3.SGS40GSGS5GS6GS7.SI._GS8GS+GS80_GS9GS9SS0GS:GS;GS<.S._GS=GS=SYSO0S[S\0SUGSZGS6GS>.SI._GS?GS?SSUGS^GS_GS`GSa.SCGSbGScGSdGSe.S._GS@GS@SS0GSSGS@GSGSA.0GS._GSAGSASS0S._GSBGSBSySzS{S|.SS0S}S~S.S}S~S.S}S~S.S|.S9._GSCGSCSS0SS0S._GSDGSDSS0SS0S._GSEGSESS0SS0S._GSFGSFSS0SS0S._GSGGSHSSCSD.SS0S._E0 GSIGSISS0SS0S._GSJGSJSS0SS0GSKScGSL.SI._GSMGSMSGSNGSO.SS0SSc0SI._GSPGSPSYSO0S[S\0GSTSGGSQGSRGSS.SI._GSTGStSS0SGSGS.GSS^GS?GS@GSA.0S`._GSUGSU0 SS_GSVGSW_GSXGSY_GSZGSq_GS[GS\_GS]GS^_GS_GS`_GSaGSb_GScGSd_GSeGSf_GSgGSh_GSiGSj_GSkGSl_GSmGSn_GSoGSp_GSqGSr_0 SSC_GSsGSt_GSuGSv_GSwGSx_GSyGSz_GS{GS|_GS}GS~_GSGS_GSGS_GSGS_GSGS_GSGS_GSGS_GSGS_GSGS_GSGS_GSGS_GSGSGSGSGSGSGS.ES._GSGSSGSGSGSGSGSGSSGSGSGSGSGSGS.SCGSGSGSGSGSGSGSGSGSGSGSGS.S._GSGSSS0SS0SS^S~S.0S`._GSGSSGSGSSSS SGS.SS"GSGSSGS.S._GSGSSOGSGSGSGS.SGSGSGSGS.S._GSGSSGS/GS!GS0.SS0S._GSGSSGS/GS!GS0.SGSGS.S._GSGSSS0SS0GSSGSS5.0S`._GSGSSS0SS0S._GSGSSGS.GS6.SS0S._GSGSSS0SS0S._GSGSSS0S._E0 GSGSSS0SS0S3GS}S5.S3GS~S.GS.S9._GSGSSSGSGS.SS0S._GSGSSGSGSGS.SS0S._GSGSGSGSGS.SS0S._GSGSSS0S._GSGSSGSGS.SGSGS.S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSGSGSGS.SGSGS.GSSGSS5.0S`._GSGSSSCGSGSGS.SGSGS.GSSGSS5.0S`._GSGS+GS0_GSGSSSCSD.SS0S._GSGSSS0GS SC0GSGS0GSS^GSS5.0GS)._GSGSSS0SS0S._GSGSSSSSGS.SSGS.S._GSGS+GS0_GSGSSYS0SS0S._E0 GSGSSS0SGSGS.S._GSGSSS0SS0S._GSGSSGSGS.SGSGSGS .S._GSGSSGSGS.SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GS	GS	SYSO0SS0S^GS
GSGSA.S^GSGSGSA.GS.S`._GSGSSSO0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0GSGS0SI._GSGSSS0SS0S._GSGSSSCSD.SS0S._E0 GSGSSS0S._GSGSGSSGS@GSGSA.0GS._GSGSS^GS
GSGSA.S^GSGSGSA.GS.GS._GSGSSS^GS(S5.0GS._GSGSSS0S._GSGSSGSGSGS.SS0S._GSGSSS0SS0SS^S~S.0GS S^S~S.0GS!._GS"GS"SS^S~GS(GS#.0GS._GS$GS$SS0SS0S._GS%GS%SS0SS0S._GS&GS'SS0SGSGS.SGS(0SS^S~S.0GS._GS)GS)SS0SSC0S._GS*GS*SS0SS0S._GS+GS+SS0SS0S._GS,GS,SSO0SS0S^GS
GSGSA.S^GSGSGSA.S^GS-GS.GSA.GS/.S`._GS0GS1SSCSD.SS0S._GS2SJSS0SS0ScSdGS3GS4.SS^S~S.0GS._E0 GS5GS5SS0GSGS6GS7GS8.S._GS9GS9SS0SS0S._GS:GS:SSUSVSSGS;.SCSSSSSS.SS0SI._GS<GS<SOGSGS=.SGS40SGS0SI._GS>GS>SS0SS0S._GS?GS?SS0SS0S._GS@GS@SS0SS0S._GSAGSASGSBGSCGSDGSE.GSFGSGGSH.S._GSIGSISGSJGSKGSL.SGS"GSQGSM.S._GSNGSNSS0SS0S._GSOGSOSS0SS0GSPS^GSQGSRGSA.0S`._GSSGSTSS0SSS.GSUS^GSVS5.0S9._GSWGSXSS0SSS.GSUS^GSVS5.0S9._GSYGSYSSO0SS0S._GSZGS[SS0SS0GS\S0SI._GS]GS]SSCSD.SS0S._GS^GS^SS0SS0S._E0 GS_GS`SSSSUS.S[S0SS0SI._GSaGSaSYS0SS0S._GSbGSbSS0S._GScGScSYS0SS0S._GSdGSdSS0SSC0S._GSeGSfSS0S3S~S.S3S~S.GS.GSSGS@GSGSA.0GSg._GShGSiSS0SS0S._GSjGSjSS0SSGSGSkGSl.S._GSmGSmSS0SGSGSkGSn.S._GSoGSoSSGSp.SS0GSqGS0GS2S^S~S.0GS._GSrGSrSSGSp.SS0S._GSsGStSS0SGSGS.SGS(0SS^GSuGSvGSA.0GS._GSwGSwSS0SSC0S._GSxGSySGSGSzGS{GS|GS}.SGS~GSGSGS.GSGSGSS5.0S9._GSGS+GS0_GSGSSOGSGSGS.SGSGS.S._GSGSSSGSGSGS.GSSGS.S._E0 GSGSGSSSCGS.SS0S._GSGS+GS0_GSGSSGSGS.GSSGS.S._GSGSSS0SS0S._GSGSGSGSGSGSGSGS.SGSGS.S._GSGSSGSGSGSGS.SGSGSGS.S._GSGSSOGSgGS.S[S\0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSSS0SS0S._GSGSGSSGSGS.GSGSGSGSGSGSGS.S._GSGS+GS0_GSGSGSGSSGSGS.GSGS0S._GSGSSSGSS.SSC0S._GSGSSSGSS.SCGSGSGS.S._GSGSSOGSGS.S\GSGS.S._E0 GSGSSOGSGSSGSGSGS.GSGSSGSGSGS.S._GSGSSSS SGS.SS"GS.S._GSGSSGSGS.SS0S._GSGSGSGSSSGS.GSGSGSGSrGSGS.S._GSGSGS GS!GS".GSGSGSGS.S._GSGSSSGS.0 GSGS_GSGS_GSGS_GSGS_GSS&_GSS'_GSS)_GSS*_GSGS_GSGS_GSGS_GSGS_GSGS_GSS_GSS_GSGS_GS GS_GSS{0ES._GSGSSSSGS.GSS0S._GSGSSS0SS0S._GSGSSS0SGSGS.S._GS	GS	SGS
GSSSGS.SGSGSGS.S._GSGSSYSO0SS0S._GSGSSSCSD.SGSGS.S._GSGSSSCSD.SGSGS.S._GSGSSSCGSVGS.SS0SESFSGGSUGSGSGSGSGSGSGS.
SI._GSGSSGSQGS.SS0S._GS GS SS0SS0S._GS!GS!GS"SGSQGS#GS$.GS%GS&GS'.S._E0 GS(GS(SLGS)0GS*GS+0S._GS,GS,SGSKGSL.SS0S._GS-GS-SSCSD.GS.GS/GS0SGS1.S._GS2GS2SS0SS0S._GS3GS3SOSSSVGS4.GS5GS6GS7SGS8.S._GS9GS9SOGS:SSVGS;.GS<GS=SGS>.S._GS?GS?SS0SS0S._GS@GSASSGSmSSGS.SGSqSGS.S._GS>GSBSSCSD.SS0GSCS^GSBS5.0S`._GSDGSDSS0S._GSEGSESS0SS0S._GSFGSFSyGSGSGSGGSH.SS0S._GSIGSISSGS.SS0S._GSJGSJSOSSSSVGSK.SSGS5GSLGSM.S._GSNGSNSS0SGSGSOGSP.GSQGSR0SI._GSSGSSSS0SS0S._GSTGSTSS0SS0S._E0 GSUGSUSS0SS0S._GSVGSVSOSSSVGSWGSXGSYGSZGS[.GS\GS]GS^GS_.S._GS`GS`SS0SS0S._GSaGSaGSbGSb0SS0S._GScGScSS0S._GSdGSdGSeGSfGSg.GShGSi0S._GSjGSjSSGSkGSlGSm.SGSnGSoGSp.S._GSqGSqGSrGSsGStGSu.SS0S._GSvGSvSGSwGSxGSy.GSzSGS{GS|.GS}GS~GS.SI._GSGSSGSGS.SS0S._GSGSSS0S._GSGSSGSGSGSGSGS.GSGSGSGS.S._GSGSSS0SGSGS.S._GSGSSS0S._GSGSSGSGS.GS.SS0GSGSGSGSGS.SI._GSGSSS0SGSGS.S._GSGSSSO0SS0SGSGS.SGSGS.SGSGS.GS.S`._EGSSSO0SS0SGSGS.SGSGS.SGSGS.GS.S`.GSSSGSS.SS0S.GSSSCSUSVGS.SS0S.GSGS:GSVGSGSPGSGSGSGS.SGSSGS.S.GSSSGSmGSGSSSGS.SGSqSGS.SS0SS^GSS5.0GS).GSSS0SS0S.GSSGSGS.SS0S.GSSS0SS0S.GSSS0SS0S.GSSS0SS0S.GSSS0SS0S.GSSYSO0SS0S.GSSGSGS.SS0S.GS.Er 0 S GSGSGS._SGSGSGS._GSGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._S%GSGSGS._S;GSGSGS._S@GSGSGS._SMGSGSGS._SNGSGSGS._STGSGSGS._SXGSGSGS._SbGSGSGS._GSGSGSGS._GSGSGSGS._0 GSGSGSGS._SfGSGSGS._SmGSGSGS._SoGSGSGS._StGSGSGS._SvGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._GSGSGSGS._E0 GSGSGSGS._SGSGSGS._SGSGSGS._GSGSGSGS._SGSGSGS._GSGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._GSGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._GSGSGSGS._E0 SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._SGSGSGS._GSGSGSGS._SGS GSGS._SGSGSGS._GSGSGSGS._SGSGSGS._SGSGSGS._GSGSGSGS._SGSGSGS._SGSGSGS._GSGSGSGS._SGSGSGS._GS	GS
GSGS._E0 SGSGSGS._GSGSGSGS._GSGSGSGS._SGSGSGS._GSGSGSGS._GSGSGSGS._SGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGS GS._GS!GS"GS#GS._GS
GS$GSGS._E0 GSGS%GSGS._GSGS&GS'GS._GSGS(GSGS._GSGS)GSGS._GSGS*GS+GS._GSGS,GS-GS._GS%GS.GSGS._GS'GS/GSGS._GS,GSGSGS._GS-GS0GSGS._GS0GS1GSGS._GS3GS2GSGS._GS4GS3GSGS._GS5GS4GSGS._GS5GS6GSGS._GS7GSGSGS._GS7GS8GSGS._E0 GS;GSGSGS._GS<GSGSGS._GSGS9GS:GS._GS>GS;GSGS._GSEGS<GSGS._GSGGS=GSGS._GS>GS?GS@GS._GSGSAGSBGS._GSGSCGSDGS._GSEGSFGSGS._GSGGSFGSGS._GSHGSHGSGS._GSIGSIGSGS._GSJGSJGSGS._GSKGSLGSGS._GSGSMGSNGS._GSNGSOGSGS._E0 GSOGSPGSGS._GSGSQGSGS._GSSGSRGSGS._GSSGSTGSGS._GSXGS1GSGS._GSUGSVGSGS._GS\GSWGSGS._GSfGSXGSGS._GSjGSGSGS._GSGSYGSZGS._GSkGS[GSGS._GSlGS\GS]GS._GSuGS^GSGS._GSvGS_GSGS._GSwGS`GSGS._GSxGSGSGS._GSyGSaGSGS._E0 GSbGScGSGS._GS{GSdGSGS._GSGSGSGS._GSGSeGSfGS._GSGSgGShGS._GSGSiGSjGS._GSGSGSkGS._GSGSlGSkGS._GSGSGSGS._GSGSmGSnGS._GSGSoGSGS._GSpGSqGSrGS._GSsGStGSGS._GSGSuGSGS._GSvGSwGSGS._GSxGSyGSGS._GSGSGSGS._E0 GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSzGSGS._GSGS{GSGS._GS|GS}GS~GS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSWGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._E0 GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._E0 GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGShGS._GSGSGShGS._GS GSGSGS._GSGSGSGS._GS	GSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSLGSGSGS._GSQGSGSGS._E0 GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._E0 GSGSGSGS._GS3GSGSGS._GS9GSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GS GSGSGS._GS!GSGSGS._GS%GSGSGS._GS(GSGSGS._GS/GSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GS?GSGSGS._E0 GS1GSGSGS._GS2GSGSGS._GS3GSGSGS._GS4GSGSGS._GS7GSGSGS._GSBGSGSGS._GSCGSGSGS._GSGSGSGS._GSDGSGSGS._GSEGSGSGS._GSGSGSGS._GSFGSGSGS._GSGSGSGS._GSHGSGSGS._GSIGSGSGS._GSJGSGSGS._GSLGSGSGS._E0 GSNGSGSGS._GSPGSGSGS._GSQGSGSGS._GSTGSGSGS._GSUGSGSGS._GSVGSGSGS._GSWGSGSGS._GS[GSGSGS._GS^GSGSGS._GS_GSGSGS._GSvGSGSGS._GSzGSGSGS._GSDGSGSGS._GS|GSGSGS._GS}GSGSGS._GS~GSGSGS._GSGSGSGS._E0 GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGS GSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSFGSGSGS._GSGSGSGS._GSIGSGSGS._GSGSGSGS._E0 GSJGS	GS
GS._GSGSGSGS._GSNGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGS GSGS._GS!GS"GSGS._GSGS#GSGS._GSGS$GShGS._E0 GSGS%GShGS._GS&GS'GS(GS._GSGS)GSGS._GSGS#GSGS._GSGSGSGS._GS*GS+GS,GS._GSTGS-GSGS._GS.GS-GSGS._GSGS/GS0GS._GSGSGSGS._GSGS1GS2GS._GSGS3GS4GS._GSGS5GS6GS._GS7GS'GS(GS._GSGS8GSGS._GSGS9GSGS._GS:GS;GS<GS._E0 GS=GS;GS<GS._GS>GS?GS<GS._GS@GSAGSBGS._GSCGSDGSEGS._GSFGSAGSBGS._GSGGSHGSIGS._GSJGSKGSBGS._GSLGSMGSNGS._GSOGSPGSQGS._GSGSRGSSGS._GSUGSGSGS._GSGSTGSGS._GSGSUGSVGS._GSGSWGSGS._GSGSXGSGS._GSGSYGSGS._GSZGS[GS\GS._E0 GSGS+GS,GS._GS]GS^GSGS._GSrGSGSGS._GS_GSGSGS._GSGS`GSaGS._GSGSbGSGS._GSGScGSGS._GSGSdGSeGS._GSfGSgGSGS._GS	GSGShGS._GSGSiGSGS._GSjGSuGSGS._GSkGSlGSGS._GSGSmGSGS._GSGSnGSGS._GSGSoGSGS._GSGSpGSGS._E0 GSGSqGSGS._GSGSrGSGS._GSGSrGSGS._GSGSsGSGS._GSGStGSGS._GSGSGSGS._GSuGSGSGS._GSGSvGSGS._GSGSwGSGS._GSGSxGSGS._GS#GSyGSGS._GS(GSzGS{GS._GS-GS|GSGS._GSdGS}GS~GS._GS.GSGSGS._GS1GSGSGS._GSjGSGSGS._E0 GS9GSGSGS._GSGSGSGS._GS@GSGSGS._GSGSGSGS._GSqGSGSGS._GSAGSGSGS._GSvGSGSGS._GSGSGSGS._GSCGSGSGS._GSDGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSEGSGSGS._GSFGSGSGS._GSIGSGSGS._E0 GSJGSGSGS._GSMGSGSGS._GStGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._E0 GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGS<GS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._E0 GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GS	GSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._E0 GSGSGSGS._GS"GSGSGS._GS$GSGSGS._GS%GSGSGS._GSGSGSGS._GS'GSGSGS._GS)GSGSGS._GSGSGSGS._GS*GSGSGS._GSGSGSGS._GSGSGSGS._GS,GSGSGS._SJGS^GSGS._GS5GSGSGS._GSGSGSGS._GSGSGSGS._GS9GSGSGS._E0 GSGSGSGS._GSGSGSGS._GS<GSGSGS._GS>GSGSGS._GS?GSGSGS._GS@GSGSGS._GSAGSGSGS._GSIGSGSGS._GSGSGSGS._GSNGSGSGS._GSGSGSGS._GSTGSGSGS._GSYGS`GSGS._GSGSGSGS._GS[GSGSGS._GS]GSGSGS._GSGSGSGS._E0 GSGSGSGS._GS`GSGSGS._GSGSGSGS._GSGSGSGS._GSaGSGSGS._GSGSGSGS._GSbGSGSGS._GScGSGSGS._GSdGSGSGS._GSfGSGSGS._GSiGSGSGS._GSG	S GSGS._GSmG	SG	SGS._GSoG	SGSGS._GSrG	SGSGS._GStG	SGSGS._GSwG	SG	SGS._E0 GSyG	SG	S	GS._G	S
GSGSGS._G	SGSGSGS._G	SG	SGSGS._GSG	SGSGS._G	SG	SGSGS._GSG	SG	SGS._GSG	SG	SGS._G	SG	SG	SGS._G	SG	SGSGS._G	SG	SGSGS._G	SG	SG	SGS._G	SG	S G	S!GS._GSG	S"GSGS._GSG	S#G	S$GS._GSG	S%GSGS._GSG	S&G	S'GS._E0 GSG	S(G	S)GS._G	S*GSuGSGS._G	S+G	S,GSGS._G	S-G	S.GSGS._GSG	S/G	S0GS._G	S1G	S2G	S3GS._G	S4G	S5G	S6GS._G	S7G	S8G	S9GS._G	S:G	S;G	S<GS._G	S=G	S5G	S6GS._G	S>G	S?GSGS._GSaG	S@GSGS._G	SAG	SBG	S6GS._SBGSGSGS._SKGS5GSGS._SxG	SCGSGS._SGSGSGS._E0 SGSGSGS._SG	SDGSGS._GSG	SEG	SFGS._GS|G	SGGSGS._GSGSGSGS._GSG	SHG	SIGS._GSGSGSGS._GSG	SJGSGS._G	SKG	SLG	SMGS._GSGSGSGS._GSGSGSGS._GS	G	SNG	SOGS._GSG	SPG	SQGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSxGSGSGS._E0 GSyGSGSGS._GSiG	SRGSGS._GSGSGSGS._GSGS#GSGS._GSGSGSGS._GSGSGSGS._GSGSGSGS._GSBG	SCGSGS._GSHGSGSGS._GSG	SSGSGS._GSG	STGSGS._GS1GSGSGS._GSG	SUG	SVGS._G	SWG	SXGSGS._GSOG	SYGSGS._GSjG	SZG	S[GS._ErG	g\(]	  absxXoutOut)phi_nameinputsoutputsaccuracyIndicesLabel)r   indiceslabelAccuracyCorrectTotal)r
   correcttotalacosacoshadadelta	adadelta_ParamGradAvgSquaredGradAvgSquaredUpdateLearningRateMasterParam)paramgradavg_squared_gradavg_squared_updatelearning_ratemaster_paramParamOutAvgSquaredGradOutAvgSquaredUpdateOutMasterParamOut)	param_out
moment_outinf_norm_outmaster_param_outadagradadagrad_Moment)r   r   momentr"   r#   	MomentOut)r(   r)   r+   adamadam_Moment1Moment2
Moment2MaxBeta1PowBeta2Pow
SkipUpdate)
r   r   r"   moment1moment2moment2_max	beta1_pow	beta2_powr#   skip_update
Moment1Out
Moment2OutMoment2MaxOutBeta1PowOutBeta2PowOut)r(   moment1_outmoment2_outmoment2_max_outbeta1_pow_outbeta2_pow_outr+   floatBeta1Tensor)	data_typetensor_nameBeta2TensorEpsilonTensor)beta1beta2epsilon)r   r   r	   scalaradamaxadamax_InfNorm)r   r   r"   r/   inf_normr<   r#   
InfNormOutadamwadamw_elementwise_addaddY)r   yScale_xScale_y	Scale_out)scale_xscale_y	scale_out)r   r   r	   attrssumadd_nr   add_position_encodingaddmmInput)inputr   r]   AlphaBeta)alphabetaaffine_channelScaleBias)r   scalebiasaffine_gridrj   ThetaoutputOutputoutput_shapeintOutputShape)r   r   r	   	int_array
reduce_allalldimkeep_dim)axiskeepdimallcloseOtherzstd::stringRtolAtol)rtolatolreduce_amaxamaxreduce_aminaminanchor_generatorAnchors	Variances)anchorsvariances_outangle
reduce_anyanyrangearangeStartEndStep)startendstepdoubleTrue)rK   support_tensorarg_maxargmaxr   int64_targ_minargminargsort)r   r   tensor_array_to_tensorarray_to_tensorOutIndex)r   	out_index
as_complexas_realasinasinhassertCondData)conddata)r   r   assign
assign_posassign_value)r   r	   atanatan2X1X2atanhattention_lstmC0H0AttentionWeightAttentionBiasAttentionScalarAttentionScalarBias
LSTMWeightLSTMBias)	r   c0h0attention_weightattention_biasattention_scalarattention_scalar_biaslstm_weight	lstm_biasHiddenCellAttentionedXAttentionFCOutLSTMXLSTMOUT)hiddencellattentioned_xattention_fc_outlstm_xlstm_outaucPredictStatPosStatNegInsTagWeight)r   r   stat_posstat_negins_tag_weightAUC
StatPosOut
StatNegOut)r   stat_pos_outstat_neg_outbaddbmmbarrierbatch_fcW)rj   wrs   
batch_normMeanVariance)r   meanvariancerr   rs   MeanOutVarianceOut	SavedMeanSavedVarianceReserveSpace)r   mean_outvariance_out
saved_meansaved_variancereserve_spacedata_formatdata_layoutbce_loss)rj   r   beam_search_decodeIdsScores)idsscoresSentenceIdsSentenceScores)sentence_idssentence_scores	bernoullibicubic_interp_v2bicubic_interpOutSize
SizeTensor)r   out_sizesize_tensorscale_tensorbilinear_tensor_productbilinearWeight)r   r]   weightrs   bilinear_interp_v2bilinear_interpbincountWeights)r   weights	minlengthbipartite_matchdist_matDistMatColToRowMatchIndicesColToRowMatchDist)col_to_row_match_indicescol_to_row_match_distbitwise_andbitwise_not
bitwise_orbitwise_xorbmm
bn_act_xpu)r   rd   	box_coderPriorBoxPriorBoxVar	TargetBox)	prior_boxprior_box_var
target_box
output_box	OutputBoxbroadcast_tensorsc_concatc_embedding)r  r   c_softmax_with_cross_entropyLogits)logitsr   SoftmaxLoss)softmaxlossc_splitcastceilcelucheck_finite_and_unscalecheck_finite_and_unscale_)r   rr   FoundInfinite)r   found_infinitecholeskycholesky_solveclass_center_sampler   RemappedLabelSampledLocalClassCenter)remapped_labelsampled_local_class_centerclipMinMax)minmaxclip_by_normcoalesce_tensorFusedOutput)rv   fused_outputsize_of_dtypeuser_defined_size_of_dtypecollect_fpn_proposalsMultiLevelRoisMultiLevelScoresMultiLevelRoIsNum)multi_level_roismulti_level_scoresmulti_level_rois_numFpnRoisRoisNum)fpn_roisrois_numpost_nms_topnpost_nms_topNcomplex)realimagconcat
AxisTensor)r   r   r	   rd   rR   conditional_blockr   conjconv2dFilter)rj   filterconv2d_transpose)r   rf  rs   output_sizeconv2d_transpose_biasconv3dconv3d_transpose)r   rf  correlationInput1Input2)input1input2coscoshcrop_tensorcropShapeShapeTensor)rK   rL   tensors_nameOffsetsOffsetsTensor)shapeoffsetscrosssoftmax_with_cross_entropycross_entropy_with_softmaxcumprodcumsumcvmCVM)r   r  	data_normdecode_jpegdeformable_convOffsetMask)r   offsetrf  maskdepthwise_conv2dScale_inScale_in_eltwiseScale_weights)scale_inrc   scale_in_eltwisescale_weightsdepthwise_conv2d_transpose
dequantizeShift)rr   shiftdequantize_abs_maxdequantize_linear	ZeroPointInAccumInState)r   rr   
zero_pointin_accumin_stateOutScaleOutAccumOutState)r]   	out_scale	out_accum	out_statedequantize_logDict)r   dictdeterminantdetdgc_clip_by_normdgc_momentumVelocitycurrent_stepnranks)r   r   velocityr"   r#   current_step_tensornranks_tensorVelocityOutGrad_out)r(   velocity_outr+   grad_outdiag_v2diag
diag_embeddiagonaldigamma	dirichletrm   dist	div_scaleScaleTensorFalse)rr   rs   elementwise_divdividedotdropoutSeed)r   seed_tensor)r   r  dropout_probis_testdropout_implementationseedfix_seed)pr  moder  r  r  r   
dropout_ndedit_distanceHypsRefs
HypsLength
RefsLength)hypsrefs
hypslength
refslengthSequenceNum)sequencenumr   eigEigenvaluesEigenvectors)out_wout_veigheigvalseigvalsh)eigenvalueseigenvectorsuploUPLOeinsumOperands
InnerCacheXShape)r   inner_cachexshapeelementwise_powelulookup_table_v2	embedding)r   r  sparse	is_sparseemptyrz  ShapeTensorList)r   r	   r{   equal	equal_allerferfinvexp	expand_v2expandexpand_shapes_tensor)r   r   r	   rd   r{   expand_as_v2	expand_asexpm1exponentialexponential_lamlambdaeye)num_rowsnum_columns)r   r	   rR   $fake_channel_wise_dequantize_max_absScales)r   scales"fake_channel_wise_quantize_abs_max)r   r  -fake_channel_wise_quantize_dequantize_abs_maxfake_dequantize_max_absfake_quantize_abs_max fake_quantize_dequantize_abs_max/fake_quantize_dequantize_moving_average_abs_maxInScale)r   in_scaler  r  )r   r  r  r  $fake_quantize_moving_average_abs_maxfake_quantize_range_abs_maxIter)r   r  iter	OutScales)r   r  
out_scalesfaster_tokenizerVocabTextTextPair)vocabtext	text_pairInputIds
SegmentIds)	input_idssegment_idsfc)r  rc   r  feedfetch_barrierfft_c2cfft_c2rfft_r2cfill_anyfillvaluefill_diagonalfill_diagonal_tensorflash_attn_unpadded)max_seqlen_qmax_seqlen_k)r   rR   flash_attn_v3_varlenflash_attn_varlen_qkvpackedflatten_contiguous_rangeflatten)r   r  
start_axis	stop_axis)r1  r2  flipfloorelementwise_floordivfloor_divideelementwise_fmaxfmaxelementwise_fminfminfoldframefrobenius_normfill_constantfullfill_any_like	full_likefull_with_tensor)r   r{   
fused_adamfused_adam_ParamsGradsMoments1Moments2Moments2Max	Beta1Pows	Beta2PowsMasterParams)
paramsgradsr"   moments1moments2moments2_max
beta1_pows
beta2_powsmaster_paramsr>   	ParamsOutMoments1OutMoments2OutMoments2MaxOutBeta1PowsOutBeta2PowsOutMasterParamsOut)
params_outmoments1_outmoments2_outmoments2_max_outbeta1_pows_outbeta2_pows_outmaster_params_outfused_attentionLnScaleLnBiasQKVWQKVBiasCacheKVSrcMask
OutLinearWOutLinearBiasLn2ScaleLn2Bias)r   ln_scaleln_bias
qkv_weightqkv_biascache_kvsrc_maskout_linear_weightout_linear_bias
ln_scale_2	ln_bias_2ln_meanLnMeanln_var
LnVarianceln_outLnOutqkv_outQKVOutqkv_bias_out
QKVBiasOuttranspose_out_2TransposeOut2qk_outQKOutqktv_outQKTVOutsoftmax_out
SoftmaxOutattn_dropout_mask_outAttnDropoutMaskOutattn_dropout_outAttnDropoutOutsrc_mask_out
SrcMaskOutfmha_outFMHAOutout_linear_outOutLinearOutdropout_mask_outDropoutMaskOut	ln_mean_2Ln2Meanln_var_2Ln2VarianceBiasDropoutResidualOut
CacheKVOut)bias_dropout_residual_outcache_kv_outr   fused_batch_norm_act&fused_bias_dropout_residual_layer_normResidual)r   residualrs   rn  ro  )r  r  rx  ln_variancer]   fused_bn_add_activationfused_bn_add_activation_Z)r   zr   r   rr   rs   fused_conv2dResidualData)rj   rf  rs   residual_paramfused_conv2d_add_act)rj   rf  rs   residual_dataOutputs)rv   r	   fused_conv3dfused_elementwise_addfused_elementwise_divfused_elementwise_mulfused_elementwise_sub!fused_embedding_eltwise_layernormEmbs)r   embsrs   rr   fused_embedding_fc_lstm
EmbeddingsWeightH)r   
embeddingsweight_hrs   r   r   XXBatchedInputBatchedHiddenBatchedCellReorderedH0ReorderedC0)r   r   xxbatched_inputbatched_hiddenbatched_cellreordered_h0reordered_c0fused_fc_elementwise_layernormBias0Bias1)r   r   r]   bias0rr   bias1)r   r   r   fused_feedforwardDropout1SeedDropout2SeedLinear1WeightLinear1BiasLinear2WeightLinear2BiasLn1ScaleLn1Bias)r   dropout1_seeddropout2_seedlinear1_weightlinear1_biaslinear2_weightlinear2_bias	ln1_scaleln1_bias	ln2_scaleln2_biasDropout1MaskDropout2MaskLn1MeanLn1Variance
Linear1OutLn1OutDropout1OutDropout2Out)r   dropout1_maskdropout2_maskln1_meanln1_varianceln2_meanln2_variancelinear1_outln1_outdropout1_outdropout2_outr  r  dropout1_ratedropout2_rate)dropout1_seed_valdropout2_seed_valdropout1_probdropout2_probfused_gate_attentionQueryKeyQueryWeight	KeyWeightValueWeight	QKVWeightNonbatchedBias
GateWeightGateBiasOutLinearWeight)querykeyquery_weight
key_weightvalue_weightrp  nonbatched_biasrs  gate_weight	gate_biasrt  ru  QueryTransposeOutKeyTransposeOutValueTransposeOutQKVTransposeOut
SoftmaxLseGateOut)	query_transpose_outkey_transpose_outvalue_transpose_outqkv_transpose_outr  softmax_lser  gate_outr   fused_gemm_epilogue)r   r]   rs   )r   r   fused_gemm_epilogue_gradDOut)r   r]   r   out_gradDXDYDBias)x_grady_grad	bias_gradfused_multi_transformer_int8rn  ro  qkv_wrq  rr  	time_stepTimeSteprs  out_linear_wru  ffn_ln_scale
FFNLnScaleffn_ln_bias	FFNLnBiasffn1_weight
FFN1Weight	ffn1_biasFFN1Biasffn2_weight
FFN2Weight	ffn2_biasFFN2Biasqkv_out_scaleQKVOutScaleOutLinearOutScaleFFN1OutScaleFFN2OutScale)out_linear_out_scaleffn1_out_scaleffn2_out_scale)r  r   fused_seqpool_cvmfused_transpose
fusion_gruWeightX)r   r   weight_xr  rs   
BatchedOut)r  r  r  batched_outr   
Scale_data
Shift_data)
scale_data
shift_datar  fusion_lstm)r   r   r<  r  rs   r   CheckedCell)
r   r   r   r  r  r  r  r  r  checked_cellfusion_repeated_fc_relu)r   r   rs   ReluOut)relu_outr   fusion_seqconv_eltadd_reluColMat)r   col_matcontextLengthcontextStartcontextStride)context_lengthcontext_startcontext_stridefusion_seqpool_concatfusion_transpose_flatten_concatgatherIndex)r   indexAxis	gather_ndgather_treeParents)r   parentsgaussian_randomgaussiangelugenerate_proposals_v2generate_proposals
BboxDeltasImShape)r   bbox_deltasim_shaper   	variancesRpnRoisRpnRoiProbs
RpnRoisNum)rpn_roisrpn_roi_probsrpn_rois_numpre_nms_topN)pre_nms_top_npost_nms_top_nglobal_gatherglobal_scattergrad_addgraph_khop_samplerRowCol_PtrEids)rowcolptrr   eidsOut_SrcOut_DstSample_Index	Reindex_XOut_Eids)out_srcout_dstsample_index	reindex_xout_eidsgraph_sample_neighborsPerm_Buffer)rv  rw  r   rx  perm_buffer	Out_Count)r   	out_countr  greater_equalgreater_thangrid_samplergrid_sampleGrid)r   grid
group_norm)r]   r   r   gumbel_softmaxhard_shrink
hardshrinkhard_sigmoidhardsigmoid
hard_swish	hardswishbreluhardtanhhashruntime_shape&ALL_KERNELS_MUST_COMPUTE_RUNTIME_SHAPEelementwise_heaviside	heaviside
hinge_lossLabels)r1  labelsr5  	histogram)rj   r  hierarchical_sigmoidhsigmoid_loss	PathTablePathCode)r   r   r   rs   pathcodePreOutW_Out)r   pre_outw_out
huber_loss)r   r  im2sequencer_  	increment	index_addAddValue)r   rV  	add_valueindex_elementwise_getrV  
input_dimsinput_strides
index_dimsindex_stride)r   rV  r  r  r  r  slice_offset
accumulateis_combined)r  r  r  index_elementwise_put!index_elementwise_put_with_tensorindex_sampleindex_selectinstance_norm)r]   r   r   inverseis_emptyiscloseisfinite_v2isfiniteisinf_v2isinfisnan_v2isnan
kldiv_lossTarget)r   r   kronkthvaluel1_normlabel_smooth	PriorDist)r   
prior_distlamblamb_)	r   r   r"   r9   r:   r<   r=   r#   r>   )r(   rD   rE   rG   rH   master_param_outs
layer_norm
leaky_relunegative_slopelegacy_bilinear_interplegacy_expandexpand_timesExpandTimesexpand_times_tensorlegacy_generate_proposalsImInfo)r   rc  im_infor   re  matmullegacy_matmulDDXDDY)r   r]   r  x_grad_grady_grad_grad)r   r  r  transpose_Xtranspose_Y)transpose_xtranspose_ynearest_interplegacy_nearest_interpreshapelegacy_reshapelerp)r   r]   r  
less_equal	less_thanlgammalinear_interp_v2linear_interp	linear_v2linspaceStopNum)r   stopnumber	lod_resetloglog10log1plog2log_loss	Predictedlog_softmaxlogcumsumexplogical_andlogical_not
logical_orlogical_xorlogit
logsigmoid	logsumexplookup_table)r   r   lrnMidOut)r   mid_outlstsqSolution	ResidualsRankSingularValues)solution	residualsranksingular_valuesrcond	lu_unpackPivotsPmatLU)pmatlumargin_cross_entropymasked_select)r   r  match_matrix_tensor)r   r]   r   Tmp)r   tmp	matmul_v2trans_xtrans_ymulmatmul_with_flatten
matrix_nmsBBoxes)bboxesr   )r   rV  roisnummatrix_powermatrix_rank	TolTensor)r   
tol_tensor
reduce_maxrI  max_pool2d_with_indexkernel_sizeksizemax_pool3d_with_indexelementwise_maxmaximummaxoutreduce_meanr   mean_allmemory_efficient_attentionmerge_selected_rowsmerged_adam_)	r   r   r"   r9   r:   r;   r<   r=   r#   merged_momentummerged_momentum_)r   r   r  r"   r#   )r(   r  r+   meshgrid
reduce_minrH  elementwise_minminimummish	threshold)r   r  r  momentum	momentum_moving_average_abs_max_scale)r   r  r  	multi_dot	multi_gru)r   r<  r  rs   r  r   )rA  rB  multiclass_nmsmulticlass_nms3)r)  r   rZ  
NmsRoisNum)r   rV  nms_rois_nummultihead_matmulBiasQK)rj   r   rs   bias_qktranspose_Qtranspose_Ktranspose_V)transpose_qtranspose_ktranspose_vmultinomialnum_samples	multiplex)r   rV  elementwise_mulmultiplymvVec)r   vec	nanmedianMedianIndex)r   mediansrK   nearest_interp_v2nll_loss)rj   r   r  Total_weight)r   total_weightnmsBoxesKeepBoxesIdxsiou_thresholdwhere_indexnonzero	condition	ConditionnormNorm)r   rm  	not_equalsizenumel
one_hot_v2one_hotdepthdepth_tensoroverlap_addp_normpad	pad_valuepad2dpad3dpaddingsPaddingspartial_allgatherpartial_concatpartial_recvpartial_sumpixel_shufflepixel_unshufflepoissonpool2dpool3dpowr]   factorFactorTensorprelu)r   rm   printinInr'  Image)rj   image)r   varreduce_prodprod
psroi_poolROIs)r   boxes	boxes_numput_along_axisValue)arrr   valuesResultReduceInclude_self)r   reduceinclude_selfpylayerqrQR)qrquantize)rr   r  r  quantize_linearrandintrandpermrange_v2r^  
reciprocalrelurelu6elementwise_mod	remainderrenormrepeat_interleaveRepeats)repeatsr   #repeat_interleave_with_tensor_indexRepeatTensor)r   r  
requantizeShift_in	Shift_out)r  rc   shift_in	shift_outreshape2resnet_basic_blockfilter1Filter1scale1Scale1r  mean1Mean1var1Var1filter2Filter2scale2Scale2bias2Bias2mean2Mean2var2Var2filter3Filter3scale3Scale3bias3Bias3mean3Mean3var3Var3conv1Conv1saved_mean1
SavedMean1saved_invstd1SavedInvstd1	mean1_outMean1Outvar1_outVar1Outconv2Conv2conv2_input
Conv2Inputsaved_mean2
SavedMean2saved_invstd2SavedInvstd2	mean2_outMean2Outvar2_outVar2Outconv3Conv3saved_mean3
SavedMean3saved_invstd3SavedInvstd3	mean3_outMean3Outvar3_outVar3Out	MaxInput1
MaxFilter1	MaxInput2
MaxFilter2	MaxInput3
MaxFilter3)
max_input1max_filter1
max_input2max_filter2
max_input3max_filter3resnet_unitFilterXScaleXBiasXMeanXVarXFilterZScaleZBiasZMeanZVarZ)r   filter_xra   bias_xmean_xvar_xr  filter_zscale_zbias_zmean_zvar_zBitMaskConvX
SavedMeanXSavedInvstdXRunningMeanXRunningVarXConvZ
SavedMeanZSavedInvstdZRunningMeanZRunningVarZ)r   bit_maskconv_xsaved_mean_xsaved_invstd_xrunning_mean_xrunning_var_xconv_zsaved_mean_zsaved_invstd_zrunning_mean_zrunning_var_zreversermsproprmsprop_
MeanSquareMeanGrad)r   mean_square	mean_gradr"   r   r/   r#   MeanSquareOutMeanGradOut)r(   r)   mean_square_outmean_grad_outr  rnnPreState
WeightListSequenceLength)r   	pre_stateweight_listsequence_lengthDropoutStateStateReserve)r   dropout_state_outstatereserve	roi_alignroi_poolArgmax)r   r   rollshiftsShiftsTensorroundrow_convrsqrtsave_combinerr   scatterUpdates)r   rV  updatesscatter_nd_addsearchsortedSortedSequenceValues)sorted_sequencer  segment_pool)r   r  	SummedIds)r   
summed_idsself_dp_attentionselugraph_send_recvsend_u_recv	Src_index	Dst_index)r   	src_index	dst_index	Dst_count)r   	dst_countr	  Out_sizegraph_send_ue_recvsend_ue_recv)r   r]   rb  rc  graph_send_uvsend_uvsequence_expandsequence_maskmax_lenmaxlenMaxLenTensorsequence_softmaxsgdsgd_)r   r"   r   r#   )r(   r+   shard_indexshare_bufferXOut)r   xout
share_datashare_data_shuffle_batch)r   r  
ShuffleIdxSeedOut)r   shuffle_idxseed_outshuffle_channelgroup)r   r  sigmoidsignsilusinsinhsliceStartsTensorStartsTensorList
EndsTensorEndsTensorList)startsendsslogdeterminantslogdet	soft_relur4  softplus
softshrinksoftsignsolvesparse_batch_normsparse_reshapesparse_slice
sparse_sumsparse_sync_batch_normspectral_normV)r  r  vsplitsections)r   r   r	   rR   r{   split_with_num)rK   r   rL   sqrtsquaresqueeze2squeezeaxesstackstanhstraight_through_estimator_gradstrided_sliceStridesTensorStridesTensorList)r  r  strideselementwise_subsubtract
reduce_sum	out_dtype)r   r   dtypesvdSVH)r  svhswishsync_batch_norm)r   rr   rs   r   r   take_along_axis)r  r   tantanhtanh_shrink	tdm_childTreeInfo
child_numsr  )r   	tree_infor  r  ChildLeafMask)child	leaf_masktdm_samplerTravelLayer)r   travellayer)r   r  r  thresholded_relutilerepeat_timesRepeatTimesrepeat_times_tensortop_k_v2topkkKtop_ktopk_v1trace
transpose2	transposepermtriangular_solve	tril_triutrilinear_interp_v2trilinear_interptrunctruncated_gaussian_randomunbindunfolduniform_randomuniform)r   r	   rR   r{   uniform_random_inplaceuniform_inplaceuniqueCounts)r   r   r  countsunique_consecutive)r   rV  r  unpool)r   r   paddingunpool3d
unsqueeze2	unsqueeze
AxesTensorAxesTensorListunstackupdate_loss_scalingupdate_loss_scaling_PrevLossScalingInGoodSteps
InBadSteps)r   r=  prev_loss_scalingin_good_stepsin_bad_stepsLossScalingOutGoodStepsOutBadSteps)r   loss_scalingout_good_stepsout_bad_stepsstop_updatebool
StopUpdate
view_shapeviterbi_decode
TransitionLength)
potentialstransition_paramslengthsPath)r   r  warpctcLogitsLengthLabelLength)r1  r   logits_lengthlabels_lengthWarpCTCGrad)warpctcgradr5  where)rk  r   r]   whileyolo_boxImgSize)r   img_size)r  r   yolo_box_headyolo_box_postBoxes0Boxes1Boxes2
ImageShape
ImageScale)boxes0boxes1boxes2image_shapeimage_scale)r   rL  yolov3_loss	yolo_lossGTBoxGTLabelGTScore)r   gt_boxgt_labelgt_scoreObjectnessMaskGTMatchMask)r5  objectness_maskgt_match_maskbox_clip)rj   r  c_allreduce_sum
c_identity	c_scatterchannel_shuffle
chunk_eval	Inference	SeqLength)	inferencer   
seq_length	PrecisionRecallzF1-ScoreNumInferChunksNumLabelChunksNumCorrectChunks)	precisionrecallf1_scorenum_infer_chunksnum_label_chunksnum_correct_chunkscomm_init_allcrf_decodingEmission)emission
transitionr   lengthviterbi_pathViterbiPathcross_entropycross_entropy2MatchX)r   x_shapematch_x	ctc_alignInputLength)rj   input_lengthOutputLength)rv   output_length
cudnn_lstmInitHInitC)r   init_hinit_cr   r?  r@  StateOutLastHLastC)rF  	state_outr   last_hlast_cdecayed_adagrad)r   r   r/   r"   )r(   r)   dependDep)r   depdgc)r  r  r   r   U_outV_out
EncodeGrad
GatherBuff)u_outv_outencode_gradr  gather_buffdistribute_fpn_proposalsMultiFpnRoisRestoreIndex)multi_fpn_roisrV  restore_indexdistributed_fused_lamb_init)r   r   fp32_fused_paramFP32FusedParamfp32_fused_gradFP32FusedGradfp16_fused_paramFP16FusedParamfp16_fused_gradFP16FusedGradr9   r:   r<   r=   fused_param_offsetsFusedParamOffsetsfp32_shard_fused_param_offsetsFP32ShardFusedParamOffsetsfp16_shard_fused_param_offsetsFP16ShardFusedParamOffsets
param_info	ParamInfoparam_order
ParamOrderr(   r+   r  GradOutglobal_scaleGlobalScaler   dpsgd)r   r   r"   fetch_v2fetchflatten2)r   rK  ftrlSquaredAccumulatorLinearAccumulator)r   squared_accumulatorlinear_accumulatorr   r"   SquaredAccumOutLinearAccumOut)r(   squared_accum_outlinear_accum_outfill_constant_batch_size_likefull_batch_size_likefused_elemwise_activationIntermediateOut)r   intermediate_outfused_elemwise_add_activationfused_matmul)r   r]   r  fused_reshape_Xfused_transpose_Xfused_reshape_Yfused_transpose_Yfused_reshape_Outfused_transpose_Out)
ra   rb   rc   r  fused_reshape_xfused_transpose_xfused_reshape_yfused_transpose_yfused_reshape_outfused_transpose_outfused_softmax_maskfused_softplusfused_token_pruneAttnNewMask)attnr   r  new_maskSlimmedXCLSInds)	slimmed_xcls_indsfusion_groupInputsoutsOutsfusion_seqpool_cvm_concatfusion_squared_mat_subSquaredXSquaredY	SquaredXY)	squared_x	squared_y
squared_xyr   get_tensor_from_selected_rowsgru)rj   r   r  rs   	BatchGateBatchResetHiddenPrevBatchHidden)
batch_gatebatch_reset_hidden_prevbatch_hiddenr   gru_unit
HiddenPrev)rj   hidden_prevr  rs   GateResetHiddenPrev)gatereset_hidden_prevr   identity_losslars_momentumlars_momentum_legacy_cropr{  limit_by_capacitylod_array_lengthlogspaceBase)r   r  numbaselookup_table_dequantlstm)rj   r   r   r  rs   BatchCellPreAct)r   r   r  batch_cell_pre_actluInfos)r   pivotsinfospivotr  memcpy
memcpy_d2hmp_allreduce_sumnceSampleWeightCustomDistProbsCustomDistAliasCustomDistAliasProbs)rj   r   r  rs   sample_weightcustom_dist_probscustom_dist_aliascustom_dist_alias_probsCostSampleLogitsSampleLabels)costsample_logitssample_labelsnopnumber_countnumberspartial_sendprune_gate_by_capacityGateIdxExpertCount)gate_idxexpert_countout_gate_idx
NewGateIdxpyramid_hash	WhiteList	BlackList)r   r   
white_list
black_listDropPos
X_Temp_Out)r   drop_pos
x_temp_outrandom_routingProb
TopK_ValueTopK_Idx)prob
topk_valuetopk_idxrank_attention
RankOffset	RankParam)r   rank_offset
rank_param	InputHelpInsRank)
input_helpr   ins_rankMaxRankMaxSize)max_rankmax_sizeread_from_arrayI)arrayirecv_v2graph_reindexreindex_graph	NeighborsCountHashTable_ValueHashTable_Index)r   	neighborscounthashtable_valuehashtable_indexReindex_SrcReindex_Dst	Out_Nodes)reindex_srcreindex_dst	out_nodesrreluNoise)r   noisesend_v2sequence_convPaddingData)r   padding_datarf  paddingTrainable)padding_trainablerO  rP  rQ  sequence_poolMaxIndex)r   	max_index	set_value)rK   rw  StepsTensorList)r  r  stepsset_value_with_tensor!sigmoid_cross_entropy_with_logitsskip_layernorm)r   r]   rr   rs   sparse_attentionColumnsKeyPaddingMaskAttnMask)r  r  r  r  columnskey_padding_mask	attn_maskSparseDotSdd)r   sparse_dot_sddr4  sparse_momentum)r   r   r  rV  r   r"   r#   squared_l2_normstftWindow)r   windowsync_calc_streamsync_comm_streamtemporal_shifttransfer_layoutuniform_random_batch_size_likewrite_to_array)r   r  )r<  r=  r>  r?  rH  rI  rJ  c_sync_calc_streamc_sync_comm_streamrO  rP  rQ  rR  zTensor xzTensor(out))argsrv   z&Tensor x, Tensor indices, Tensor labelz0Tensor(accuracy), Tensor(correct), Tensor(total)accuracy_checkzZTensor x, Tensor y, str fn_name, double rtol=1e-5, double atol=1e-8,  bool equal_nan=falsezTensor param, Tensor grad, Tensor avg_squared_grad, Tensor avg_squared_update, Tensor learning_rate, Tensor master_param, float rho = 0.95f, float epsilon = 1.0e-6f, bool multi_precision = falsezUTensor(param_out), Tensor(moment_out), Tensor(inf_norm_out), Tensor(master_param_out)zTensor param, Tensor grad, Tensor moment, Tensor learning_rate, Tensor master_param, float epsilon = 1.0e-6f, bool multi_precision = falsez?Tensor(param_out), Tensor(moment_out), Tensor(master_param_out)a  Tensor param, Tensor grad, Tensor learning_rate, Tensor moment1, Tensor moment2, Tensor moment2_max, Tensor beta1_pow, Tensor beta2_pow, Tensor master_param, Tensor skip_update, Scalar beta1 = 0.9f, Scalar beta2 = 0.999f, Scalar epsilon = 1.0e-8f, bool lazy_mode = false, int64_t min_row_size_to_use_multithread = 1000, bool multi_precision = false, bool use_global_beta_pow = false, bool amsgrad = falsezTensor(param_out), Tensor(moment1_out), Tensor(moment2_out), Tensor(moment2_max_out), Tensor(beta1_pow_out), Tensor(beta2_pow_out), Tensor(master_param_out)zTensor param, Tensor grad, Tensor learning_rate, Tensor moment, Tensor inf_norm, Tensor beta1_pow, Tensor master_param, float beta1 = 0.9f, float beta2 = 0.999f, float epsilon = 1.0e-8f, bool multi_precision = falsea  Tensor param, Tensor grad, Tensor learning_rate, Tensor moment1, Tensor moment2, Tensor moment2_max, Tensor beta1_pow, Tensor beta2_pow, Tensor master_param, Tensor skip_update, Scalar beta1 = 0.9f, Scalar beta2 = 0.999f, Scalar epsilon = 1.0e-8f, float lr_ratio = 1.0f, float coeff = 0.01f, bool with_decay = false, bool lazy_mode = false, int64_t min_row_size_to_use_multithread = 1000, bool multi_precision = false, bool use_global_beta_pow = false, bool amsgrad = falsez/Tensor x, float alpha = 1.0f, float beta = 1.0fzTensor (out)zATensor input, Tensor x, Tensor y, float beta=1.0, float alpha=1.0zBTensor x, Tensor scale, Tensor bias, str data_layout = "AnyLayout"z?Tensor input, IntArray output_shape={}, bool align_corners=truezTensor(output)z/Tensor x, int64_t[] axis={}, bool keepdim=false
all_gatherz'Tensor x, int ring_id = 0, int nranks=0
all_reducez.Tensor x, int ring_id = 0, int reduce_type = 0
all_to_allzTensor x, int ring_id = 0z\Tensor x, Tensor y, Scalar(double) rtol=1e-5, Scalar(double) atol=1e-8, bool equal_nan=falseTensor	ap_facadezTensor[] xs, int64_t num_outputs, str custom_op_name, str infer_meta_func_name, str infer_symbolic_func_name, str serialized_attributeszTensor[](out){num_outputs}ap_trivial_fusion_beginzTensor[] xsap_trivial_fusion_endap_variadiczTensor[] xs, int num_outputs, str code_module_lambda, str infer_symbolic_lambda, str infer_meta_lambda, str rnel_dispatch_lambda, str kernel_dispatch_const_data_lambdaapply_per_channel_scalezTensor x, Tensor scaleszmTensor x, Scalar(int64_t) axis, bool keepdims = false, bool flatten = false, DataType dtype = DataType::INT64z?Tensor x, int axis=-1, bool descending=false, bool stable=falsezTensor(out), Tensor(indices)
as_stridedzLTensor input, int64_t[] dims = {}, int64_t[] stride = {}, int64_t offset = 0asgd_z~Tensor param, Tensor grad, Tensor learning_rate, Tensor d, Tensor y, Tensor n, Tensor master_param, bool multi_precision=falsezITensor(param_out), Tensor(d_out), Tensor(y_out), Tensor(master_param_out)assign_out_zTensor x, Tensor outputz.Tensor x, Tensor cum_count, Tensor eff_num_lenassign_value_zMTensor output, int[] shape, DataType dtype, Scalar[] values, Place place = {}zTensor x, Tensor ya  Tensor x, Tensor c0, Tensor h0, Tensor attention_weight, Tensor attention_bias, Tensor attention_scalar, Tensor attention_scalar_bias, Tensor lstm_weight, Tensor lstm_bias, str gate_activation = "sigmoid", str cell_activation = "tanh", str candidate_activation = "tanh"zuTensor (hidden), Tensor (cell), Tensor (attentioned_x), Tensor (attention_fc_out), Tensor (lstm_x), Tensor (lstm_out)zTensor x, Tensor label, Tensor stat_pos, Tensor stat_neg, Tensor ins_tag_weight, str curve = "ROC", int num_thresholds = (2 << 12) - 1, int slide_steps = 1z7Tensor(auc), Tensor(stat_pos_out), Tensor(stat_neg_out)average_accumulates_zTensor param, Tensor in_sum_1, Tensor in_sum_2, Tensor in_sum_3, Tensor in_num_accumulates, Tensor in_old_num_accumulates, Tensor in_num_updates, float average_window = 0, int64_t max_average_window = INT64_MAX, int64_t min_average_window = 10000LzTensor(out_sum_1), Tensor(out_sum_2), Tensor(out_sum_3), Tensor(out_num_accumulates), Tensor(out_old_num_accumulates), Tensor(out_num_updates)zTensor x, int ring_id=0z#Tensor input, Tensor w, Tensor biaszTensor input, Tensor labelbeam_searchz~Tensor pre_ids, Tensor pre_scores, Tensor ids, Tensor scores, int level, int beam_size, int end_id, bool is_accumulated = truezDTensor (selected_ids), Tensor (selected_scores), Tensor (parent_idx)zTensor x, Tensor out_size, Tensor[] size_tensor, Tensor scale_tensor, str data_format="NCHW", int out_d=0, int out_h=0, int out_w=0, double[] scale={}, str interp_method="bilinear", bool align_corners=true, int align_mode=1z.Tensor x, Tensor y, Tensor weight, Tensor biasz3Tensor x, Tensor weights, Scalar(int) minlength = 0binomialzTensor count, Tensor probzITensor dist_mat, str match_type = "bipartite", float dist_threshold = 0.5zATensor (col_to_row_match_indices), Tensor (col_to_row_match_dist)bitwise_left_shiftz-Tensor x, Tensor y, bool is_arithmetic = truebitwise_right_shiftzTensor input, Tensor im_infozTensor (output)zTensor prior_box, Tensor prior_box_var, Tensor target_box, str code_type = "encode_center_size", bool box_normalized = true, int axis = 0, float[] variance = {}zTensor(output_box)	broadcastz'Tensor x, int ring_id = 0, int root = 0zTensor[] inputzTensor[]{input.size()}"build_src_rank_and_local_expert_idzWTensor expert_num_global_tensor, int64_t[] expert_num_global, int64_t num_local_expertsz'Tensor(vector), Tensor(local_expert_id)zDTensor x, int ring_id, bool use_calc_stream, bool use_model_parallelzZTensor x, int rank, int nranks, int ring_id, bool use_calc_stream, bool use_model_parallelzUTensor x, int ring_id = 0, int root = 0, int nranks = 0, bool use_calc_stream = falsez`Tensor logits, Tensor label,  int64_t ignore_index=-100, int ring_id=0, int rank=0, int nranks=0zTensor(softmax), Tensor(loss)zWTensor x, int rank = 0, int nranks = 1, int ring_id = 0, bool use_model_parallel = truecal_aux_losszTensor gate_prob, Tensor dispatch_mask, Tensor tokens_mask, Tensor dispatch_tokens_mask, int64_t num_experts, bool use_group, int64_t moe_k, float clip_minz4Tensor(l_aux_loss), Tensor(seqlen_float), Tensor(ce)calc_reduced_attn_scoresz&Tensor q, Tensor k, Tensor softmax_lsezTensor(reduced_scores)zTensor x, DataType dtypezTensor x, float alpha = 1.0z,Tensor x, int groups, str data_format="NCHW"zTensor[] x, Tensor scalez/Tensor[](out){x.size()}, Tensor(found_infinite)check_numericszTensor tensor, str op_type = "", str var_name = "", int check_nan_inf_level = 0, int stack_height_limit = -1, str output_dir = ""zTensor(stats), Tensor(values)zTensor x, bool upper=falsez$Tensor x, Tensor y, bool upper=falsezTensor label, int num_classes, int num_samples, int ring_id = 0, int rank = 0, int nranks = 1, bool fix_seed = false, int seed = 0z:Tensor(remapped_label), Tensor(sampled_local_class_center)z.Tensor x, Scalar(float) min, Scalar(float) maxzTensor x, float max_norma  Tensor[] input, DataType dtype, bool copy_data = false, bool set_constant = false, bool persist_output = false, float constant = 0.0, bool use_align = true, int align_size = -1, int size_of_dtype = -1, int64_t[] concated_shapes = {}, int64_t[] concated_ranks = {}z4Tensor[](output){input.size()}, Tensor(fused_output)zhTensor[] multi_level_rois, Tensor[] multi_level_scores, Tensor[] multi_level_rois_num, int post_nms_topnz$Tensor (fpn_rois), Tensor (rois_num)zTensor real, Tensor imagzTensor[] x, Scalar axis=0zTensor input, Tensor filter, int[] strides={1, 1}, int[] paddings={0, 0}, str padding_algorithm="EXPLICIT", int[] dilations={1, 1}, int groups=1, str data_format="NCHW"zTensor x, Tensor filter, int[] strides={1, 1}, int[] paddings={0, 0}, int[] output_padding={}, IntArray output_size={}, str padding_algorithm="EXPLICIT", int groups=1, int[] dilations={1, 1}, str data_format="NCHW"zTensor x, Tensor filter, Tensor bias, int[] strides={1, 1}, int[] paddings={0, 0}, int[] output_padding={}, IntArray output_size={}, str padding_algorithm="EXPLICIT", int groups=1, int[] dilations={1, 1}, str data_format="NCHW"zTensor input, Tensor filter, int[] strides={1, 1, 1}, int[] paddings={0, 0, 0}, str padding_algorithm="EXPLICIT", int groups=1, int[] dilations={1, 1, 1}, str data_format="NCDHW"zTensor x, Tensor filter, int[] strides={1, 1, 1}, int[] paddings={0, 0, 0}, int[] output_padding={}, int[] output_size={}, str padding_algorithm="EXPLICIT", int groups=1, int[] dilations={1, 1, 1}, str data_format="NCHW"copy_toz$Tensor x, Place place, bool blockingcopysignzTensor input1, Tensor input2, int pad_size, int kernel_size, int max_displacement, int stride1, int stride2, int corr_type_multiply=1z?Tensor emission, Tensor transition, Tensor label, Tensor lengthzTensor (viterbi_path)z4Tensor x, IntArray shape = {}, IntArray offsets = {}z Tensor x, Tensor y, int axis = 9zTensor input, Tensor label, bool soft_label=false, bool use_softmax=true, bool numeric_stable_mode=true, int ignore_index=-100, int axis=-1)cross_entropy_with_softmax_bwd_w_downcastz.Tensor label, Tensor softmax, Tensor loss_gradzTensor(input_grad)zcTensor input, Tensor input_length, int blank = 0, bool merge_repeated = true, int padding_value = 0z'Tensor (output), Tensor (output_length)zTensor x, Tensor init_h, Tensor init_c, Tensor w, Tensor[] weight_list, Tensor sequence_length, float dropout_prob = 0.0, bool is_bidirec = false, int hidden_size = 100, int num_layers = 1, bool is_test = false, int seed = 0zTTensor (out), Tensor (last_h), Tensor (last_c), Tensor (reserve), Tensor (state_out)cummaxz7Tensor x, int axis=-1, DataType dtype = DataType::INT64cumminz<Tensor x,  int dim, bool exclusive=false, bool reverse=falsezVTensor x, Scalar axis=-1, bool flatten=false, bool exclusive=false, bool reverse=falsez)Tensor x, Tensor cvm, bool use_cvm = truer   z5str name, IntArray shape, DataType dtype, Place placezlTensor param, Tensor grad, Tensor moment, Tensor learning_rate, float decay = 0.95f, float epsilon = 1.0e-6fz%Tensor(param_out), Tensor(moment_out)zTensor x, str mode, Place placezTensor x, Tensor offset, Tensor filter, Tensor mask, int[] strides, int[] paddings, int[] dilations, int deformable_groups, int groups, int im2col_stepzTensor x, Tensor[] depzTensor input, Tensor filter, int[] strides={1, 1}, int[] paddings={0, 0}, str padding_algorithm="EXPLICIT", int groups=1, int[] dilations={1, 1}, str data_format="NCHW"depthwise_conv2d_biaszTensor input, Tensor filter, Tensor bias, int[] strides={1, 1}, int[] paddings={0, 0}, str padding_algorithm="EXPLICIT", int groups=1, int[] dilations={1, 1}, str data_format="NCHW"depthwise_conv3d_biaszTensor input, Tensor filter, Tensor bias, int[] strides={1, 1, 1}, int[] paddings={0, 0, 0}, str padding_algorithm="EXPLICIT", int groups=1, int[] dilations={1, 1, 1}, str data_format="NCDHW"z'Tensor x, Tensor scale, float max_rangezTensor x, Tensor dictzTensor u, Tensor v, Tensor grad, Tensor param, Tensor current_step, Tensor nranks, float m=0.9, bool use_nesterov=true, float[] sparsity={}, float rampup_begin_step=0.0, float rampup_step=0.0, float regular_coeff=0.0, int regular_type=0zcTensor(u_out), Tensor(v_out), Tensor(encode_grad), Tensor(grad_out), Tensor(k), Tensor(gather_buff)zMTensor x, Tensor current_step, float max_norm, float rampup_begin_step = -1.0aH  Tensor param, Tensor grad, Tensor velocity, Tensor learning_rate, Tensor master_param, Tensor current_step_tensor, Tensor nranks_tensor, float mu, bool use_nesterov = false, str regularization_method = "", float regularization_coeff = 0.0f, bool multi_precision = false, float rescale_grad = 1.0f, float rampup_begin_step = -1.0zWTensor (param_out), Tensor (velocity_out), Tensor (master_param_out), Tensor (grad_out)z3Tensor x, int offset = 0, float padding_value = 0.0z:Tensor input, int offset = 0, int dim1 = -2, int dim2 = -1z6Tensor x, int offset = 0, int axis1 = 0, int axis2 = 1zTensor alphadisable_check_model_nan_infzTensor x, int flag = 0z!Tensor x, Tensor y, float p = 2.0zTensor param, Tensor grad, Tensor learning_rate, float clip = 10.0f, float batch_size = 16.0f, float sigma = 1.0f, int seed = 0zTensor(param_out)zTensor x, Tensor seed_tensor, Scalar p = 0.5f, bool is_test = false, str mode = "downgrade_in_infer", int seed = 0, bool fix_seed = falsezTensor(out), Tensor(mask)zWTensor hyps, Tensor refs, Tensor hypslength, Tensor refslength, bool normalized = falsez Tensor(sequencenum), Tensor(out)zTensor(out_w), Tensor(out_v)zTensor x, str UPLO = "L"z.Tensor x, str uplo = "L", bool is_test = falsez)Tensor(eigenvalues), Tensor(eigenvectors)zTensor x, float alpha = 1.0fembedding_grad_add_toz8Tensor token_indices, Tensor main_grad_, Tensor out_gradzTensor(main_grad_out)embedding_with_scaled_gradientz/Tensor x, Tensor weight, int64_t padding_idx=-1zHIntArray shape, DataType dtype=DataType::FLOAT32, Place place=CPUPlace()
empty_likez@Tensor x, DataType dtype = DataType::UNDEFINED, Place place = {}enable_check_model_nan_infzTensor x, int flag = 1zTensor x, IntArray shape = {}z/Tensor x, Tensor y, int64_t[] target_shape = {}expand_modality_expert_idztTensor expert_id, int64_t num_expert_per_modality, int64_t group_size, int64_t modality_offset, bool is_group_expertzTensor(expert_id_out)zTensor x, float lamzUScalar num_rows, Scalar num_columns, DataType dtype=DataType::FLOAT32, Place place={}z]Tensor x, Tensor[] scales, int[] quant_bits = {8}, int quant_axis = 0, int x_num_col_dims = 1zZTensor x, int bit_length = 8, int round_type = 1, int quant_axis = 0, bool is_test = falsezTensor(out), Tensor(out_scale)zDTensor x, int bit_length = 8, int round_type = 1, int quant_axis = 0z0Tensor x, int bit_length = 8, int round_type = 1zTensor x, Tensor in_scale, Tensor in_accum, Tensor in_state, float moving_rate = 0.9, int bit_length = 8, bool is_test = false, int round_type = 1zDTensor(out), Tensor(out_scale), Tensor(out_state), Tensor(out_accum)z~Tensor x, Tensor in_scale, Tensor iter, int window_size = 10000,  int bit_length = 8, bool is_test = false, int round_type = 1z2Tensor(out), Tensor(out_scale), Tensor(out_scales)z9Tensor x, int64_t[] axes, str normalization, bool forwardzSTensor x, int64_t[] axes, str normalization, bool forward, int64_t last_dim_size=0LzHTensor x, int64_t[] axes, str normalization, bool forward, bool onesidedz Tensor x, Scalar(double) value=0z6Tensor x, float value=0, int offset=0, bool wrap=falsezBTensor x, Tensor y, int64_t offset = 0, int dim1 = 0, int dim2 = 1
flash_attnzTensor q, Tensor k, Tensor v, Tensor fixed_seed_offset, Tensor attn_mask, float dropout = 0.0, bool causal = false, bool return_softmax = false, bool is_test = false, str rng_name = ""zFTensor(out), Tensor(softmax), Tensor(softmax_lse), Tensor(seed_offset)flash_attn_qkvpackedzTensor qkv, Tensor fixed_seed_offset, Tensor attn_mask, float dropout = 0.0, bool causal = false, bool return_softmax = false, bool is_test = false, str rng_name = ""a  Tensor q, Tensor k, Tensor v, Tensor cu_seqlens_q,  Tensor cu_seqlens_k, Tensor fixed_seed_offset, Tensor attn_mask, Scalar max_seqlen_q, Scalar max_seqlen_k, float scale, float dropout = 0.0, bool causal = false, bool return_softmax = false, bool is_test = false, str rng_name = ""flash_attn_v3a  Tensor q, Tensor k, Tensor v, Tensor q_v_, Tensor q_descale_, Tensor k_descale_, Tensor v_descale_, float softmax_scale, bool is_causal, int window_size_left, int window_size_right, float softcap, int num_splits, bool manual_set_pack_gqa, bool pack_gqa_, int sm_marginz Tensor(out), Tensor(softmax_lse)a{  Tensor q, Tensor k, Tensor v, Tensor cu_seqlens_q, Tensor cu_seqlens_k, Tensor seqused_q, Tensor seqused_k, Tensor qv, Tensor q_descale, Tensor k_descale, Tensor v_descale, Scalar max_seqlen_q, Scalar max_seqlen_k, float softmax_scale, bool causal, int window_size_left, int window_size_right, float softcap, int num_splits, bool manual_set_pack_gqa, bool pack_gqa, int sm_margina#  Tensor qkv, Tensor cu_seqlens_q,  Tensor cu_seqlens_k, Tensor fixed_seed_offset, Tensor attn_mask, Scalar max_seqlen_q, Scalar max_seqlen_k, float scale, float dropout = 0.0, bool causal = false, bool return_softmax = false, bool is_test = false, str rng_name = "", bool varlen_padded = trueflashmask_attentionzTensor q, Tensor k, Tensor v, Tensor startend_row_indices,  Tensor fixed_seed_offset, float dropout = 0.0, bool causal = false, bool return_softmax = false, bool is_test = false, str rng_name = ""flashmask_attention_v2zqTensor q, Tensor k, Tensor v, Tensor startend_row_indices, Tensor block_mask, float softmax_scale, bool is_causalz/Tensor x, int start_axis = 1, int stop_axis = 1zTensor x, int[] axiszaTensor x, int[] output_sizes, int[] kernel_sizes,  int[] strides, int[] paddings, int[] dilationsfractional_max_pool2dzfTensor x, int[] output_size, int[] kernel_size = {0, 0}, float random_u = 0.0, bool return_mask = truefractional_max_pool3dziTensor x, int[] output_size, int[] kernel_size = {0, 0, 0}, float random_u = 0.0, bool return_mask = truez7Tensor x, int frame_length, int hop_length, int axis=-1z9Tensor x, IntArray axis,  bool keep_dim,  bool reduce_allzTensor param, Tensor squared_accumulator, Tensor linear_accumulator, Tensor grad, Tensor learning_rate, float l1=0.0f, float l2=0.0f, float lr_power=-0.5fzFTensor(param_out), Tensor(squared_accum_out), Tensor(linear_accum_out)z^IntArray shape, Scalar(double) value, DataType dtype=DataType::FLOAT32, Place place=CPUPlace()full_zmTensor output, IntArray shape, Scalar(double) value, DataType dtype=DataType::FLOAT32, Place place=CPUPlace()z~Tensor input, int[] shape, DataType dtype, Scalar(double) value, int input_dim_idx, int output_dim_idx, Place place=CPUPlace()full_int_arrayzIint64_t[] value, DataType dtype=DataType::FLOAT32, Place place=CPUPlace()zNTensor x, Scalar value, DataType dtype = DataType::UNDEFINED, Place place = {}z>Tensor value, IntArray shape, DataType dtype=DataType::FLOAT32znTensor x, Tensor scale, Tensor bias, Tensor mean, Tensor variance, float momentum, float epsilon, str act_typezvTensor(out), Tensor(mean_out), Tensor(variance_out), Tensor(saved_mean), Tensor(saved_variance), Tensor(reserve_space)zxTensor x, Tensor z, Tensor scale, Tensor bias, Tensor mean, Tensor variance, float momentum, float epsilon, str act_typefused_rms_norm_quantzTensor x, Tensor bias, Tensor residual, Tensor norm_weight, Tensor norm_bias, float epsilon, int begin_norm_axis, float quant_scale, int quant_round_type, float quant_max_bound, float quant_min_boundz2Tensor(out), Tensor(residual_out), Tensor(inv_var)zTensor x, Tensor mask!fused_softmax_mask_upper_trianglezTensor XzTensor(Out)	gammainccgammalnz%Tensor x, Tensor index, Scalar axis=0zTensor x, Tensor indexzTensor ids, Tensor parentszOIntArray shape, float mean, float std, int seed, DataType dtype, Place place={}gaussian_inplacez1Tensor x, float mean=0, float std=1.0, int seed=0z#Tensor x,  bool approximate = falsezTensor scores, Tensor bbox_deltas, Tensor im_shape, Tensor anchors, Tensor variances, int pre_nms_top_n, int post_nms_top_n, float nms_thresh, float min_size, float eta, bool pixel_offset=truez=Tensor(rpn_rois), Tensor(rpn_roi_probs), Tensor(rpn_rois_num)zBTensor x, Tensor local_count, Tensor global_count, int ring_id = 0zVTensor row, Tensor colptr, Tensor x, Tensor eids, int[] sample_sizes, bool return_eidsz[Tensor(out_src), Tensor(out_dst), Tensor(sample_index), Tensor(reindex_x), Tensor(out_eids)z~Tensor row, Tensor colptr, Tensor x, Tensor eids, Tensor perm_buffer, int sample_size, bool return_eids, bool flag_perm_bufferz0Tensor(out), Tensor(out_count), Tensor(out_eids)zcTensor x, Tensor grid, str mode = "bilinear", str padding_mode = "zeros", bool align_corners = truezeTensor x, Tensor scale, Tensor bias, double epsilon = 1e-5, int groups = -1, str data_format = "NCHW"z)Tensor(y), Tensor(mean), Tensor(variance)zTensor input, Tensor h0, Tensor weight, Tensor bias, str activation = "tanh", str gate_activation = "sigmoid", bool is_reverse = false, bool origin_mode = false, bool is_test=falsez]Tensor (batch_gate), Tensor (batch_reset_hidden_prev), Tensor (batch_hidden), Tensor (hidden)zTensor input, Tensor hidden_prev, Tensor weight, Tensor bias, int activation = 2, int gate_activation = 1, bool origin_mode = falsez:Tensor (gate), Tensor (reset_hidden_prev), Tensor (hidden)zCTensor x, float temperature = 1.0, bool hard = false, int axis = -1zTensor x, float threshold = 0.5z/Tensor x, float slope = 0.2, float offset = 0.5z'Tensor x, float t_min=0, float t_max=24zTensor logits, Tensor labelszTensor (loss)zgTensor input, Tensor weight, int64_t bins = 100, float min = 0.0, float max = 0.0, bool density = falsezhTensor x, Tensor label, Tensor w, Tensor bias, Tensor path, Tensor code, int num_classes, bool is_sparsez+Tensor(out), Tensor(pre_out), Tensor(w_out)z'Tensor input, Tensor label, float deltazTensor(out), Tensor(residual)i0i0ei1i1ezTensor x, int reduction = 1zsTensor x, Tensor y, int[] kernels, int[] strides = {1, 1}, int[] paddings = {0, 0, 0, 0}, int[] out_stride = {1, 1}zTensor x, float value = 1.0z7Tensor x, Tensor index,  Tensor add_value, int axis = 0zTensor x, Tensor[] index, int64_t[] input_dims, int64_t[] input_strides, int64_t[] index_dims, int64_t[] index_stride, int64_t slice_offset = 0, bool accumulate = true, bool is_combined = falsezTensor x, Tensor[] index, Scalar value, int64_t[] input_dims, int64_t[] input_strides, int64_t[] index_dims, int64_t[] index_strides, int64_t slice_offsetzTensor x, Tensor[] index, Tensor value, int64_t[] input_dims, int64_t[] input_strides, int64_t[] index_dims, int64_t[] index_strides, int64_t slice_offset	index_putz?Tensor x, Tensor[] indices, Tensor value, bool accumulate=falsez$Tensor x, Tensor index, int axis = 0index_select_stridedz%Tensor x, int64_t index, int axis = 0z7Tensor x, Tensor scale, Tensor bias, float epsilon=1e-5z5Tensor(y), Tensor(saved_mean), Tensor(saved_variance)interp_antialiasz]Tensor x, Tensor y, Scalar(double) rtol=1e-5, Scalar(double) atol=1e-8,  bool equal_nan=falsezGTensor x, Tensor label, str reduction = "mean", bool log_target = falsez<Tensor x, int64_t k = 1, int axis = -1, bool keepdim = falsez5Tensor label, Tensor prior_dist, float epsilon = 0.0fa   Tensor param, Tensor grad, Tensor learning_rate, Tensor moment1, Tensor moment2, Tensor beta1_pow, Tensor beta2_pow, Tensor master_param, Tensor skip_update, float weight_decay, float beta1=0.9, float beta2=0.999, float epsilon=1.0e-6f, bool always_adapt=false, bool multi_precision=falsezTensor(param_out), Tensor(moment1_out), Tensor(moment2_out), Tensor(beta1_pow_out), Tensor(beta2_pow_out), Tensor(master_param_outs)zRTensor x, Tensor scale, Tensor bias, float epsilon = 1e-5, int begin_norm_axis = 1z+Tensor(out), Tensor(mean), Tensor(variance)z&Tensor x, double negative_slope = 0.02z!Tensor x, Tensor y, Tensor weightz2Tensor expert_count, Tensor capacity, int n_workerzETensor input, Tensor weight, Tensor bias, bool transpose_weight=falsezETensor start, Tensor stop, Tensor number, DataType dtype, Place placellm_int8_linearzNTensor x, Tensor weight, Tensor bias, Tensor weight_scale, float threshold=6.0z)Tensor input, Tensor label, float epsilonzTensor x, int axis = -1zSTensor x, int axis=-1, bool flatten=false, bool exclusive=false, bool reverse=falsezTensor x, double eps = 1e-6zRTensor start, Tensor stop, Tensor num, Tensor base, DataType dtype, Place place={}zDTensor x, int[] axis={},  bool keepdim=false,  bool reduce_all=falsez.Tensor w, Tensor ids, int64_t padding_idx = -1	lp_pool2da%  Tensor x, IntArray kernel_size, int64_t[] strides = {1,1}, int64_t[] paddings = {0,0}, bool ceil_mode = false, bool exclusive = true, str data_format = "NCHW", str pooling_type = "", bool global_pooling = false, bool adaptive = false, str padding_algorithm = "EXPLICIT", float norm_type = 0.0fzTensor input, Tensor h0, Tensor c0, Tensor weight, Tensor bias, bool use_peepholes = true, bool is_reverse = false, bool is_test = false, str gate_activation = "sigmoid", str cell_activation = "tanh", str candidate_activation = "tanh"zPTensor (hidden), Tensor (cell), Tensor (batch_gate), Tensor (batch_cell_pre_act)z8Tensor x, Tensor y, Scalar rcond=0.0f, str driver="gels"zJTensor(solution), Tensor(residuals), Tensor(rank), Tensor(singular_values)zTensor x, bool pivot = truez*Tensor(out), Tensor(pivots), Tensor(infos)lu_solvez-Tensor b, Tensor lu, Tensor pivots, str transzHTensor x, Tensor y, bool unpack_ludata = true, bool unpack_pivots = truez"Tensor(pmat), Tensor(l), Tensor(u)zTensor logits, Tensor label, bool return_softmax = false, int ring_id = 0, int rank = 0, int nranks = 1, float margin1 = 1.0f, float margin2 = 0.5f, float margin3 = 0.0f, float scale = 64.0fmasked_fillz#Tensor x, Tensor mask, Tensor valuemasked_multihead_attention_a  Tensor x, Tensor cache_kv, Tensor bias, Tensor src_mask, Tensor cum_offsets, Tensor sequence_lengths, Tensor rotary_tensor, Tensor beam_cache_offset, Tensor qkv_out_scale, Tensor out_shift, Tensor out_smooth, int seq_len, int rotary_emb_dims, bool use_neox_rotary_style=false, str compute_dtype = "default", float out_scale=-1, int quant_round_type=1, float quant_max_bound=127.0, float quant_min_bound=-127.0z@Tensor(out), Tensor(cache_kv_out), Tensor(beam_cache_offset_out)z+Tensor x, Tensor y, Tensor w, int dim_t = 1zTensor (out), Tensor (tmp)zTensor bboxes, Tensor scores, float score_threshold, int nms_top_k, int keep_top_k, float post_threshold=0., bool use_gaussian = false, float gaussian_sigma = 2., int background_label = 0, bool normalized = truez+Tensor(out), Tensor(index), Tensor(roisnum)zTensor x, int nzDTensor x, float tol, bool use_default_tol=true, bool hermitian=falsematrix_rank_atol_rtolz8Tensor x, Tensor atol, Tensor rtol, bool hermitian=falsematrix_rank_tolzMTensor x, Tensor atol_tensor, bool use_default_tol=true, bool hermitian=falsez.Tensor x, IntArray axis={}, bool keepdim=falsezTensor x, int[] kernel_size, int[] strides= {1, 1}, int[] paddings = {0, 0}, bool global_pooling = false, bool adaptive = false, bool ceil_mode = falsezTensor x, int[] kernel_size, int[] strides = {1, 1, 1}, int[] paddings = {0, 0, 0}, bool global_pooling = false, bool adaptive = false, bool ceil_mode = falsemax_with_indexzITensor x, Scalar(int64_t) dim, bool keepdim = false, bool flatten = falsezTensor(values), Tensor(indices)z"Tensor x, int groups, int axis = 1medianzATensor x, IntArray axis = {}, bool keepdim = true, str mode="avg"zTensor(out), Tensor(medians)zTensor x, int dst_place_type
memcpy_h2dzTensor query, Tensor key, Tensor value, Tensor bias, Tensor cu_seqlens_q, Tensor cu_seqlens_k, Tensor causal_diagonal, Tensor seqlen_k, Scalar max_seqlen_q, Scalar max_seqlen_k, bool causal, double dropout_p, float scale, bool is_testz:Tensor(output), Tensor(logsumexp), Tensor(seed_and_offset)aJ  Tensor[] param, Tensor[] grad, Tensor[] learning_rate, Tensor[] moment1, Tensor[] moment2, Tensor[] moment2_max, Tensor[] beta1_pow, Tensor[] beta2_pow, Tensor[] master_param, Scalar beta1 = 0.9f, Scalar beta2 = 0.999f, Scalar epsilon = 1.0e-8f, bool multi_precision = false, bool use_global_beta_pow = false, bool amsgrad = falsea  Tensor[](param_out){param.size()}, Tensor[](moment1_out){param.size()}, Tensor[](moment2_out){param.size()}, Tensor[](moment2_max_out){param.size()}, Tensor[](beta1_pow_out){param.size()}, Tensor[](beta2_pow_out){param.size()}, Tensor[](master_param_out){param.size()}a  Tensor[] param, Tensor[] grad, Tensor[] velocity, Tensor[] learning_rate, Tensor[] master_param, float mu, bool use_nesterov = false, str[] regularization_method = {}, float[] regularization_coeff = {}, bool multi_precision = false, float rescale_grad = 1.0fzqTensor[](param_out){param.size()}, Tensor[](velocity_out){param.size()}, Tensor[](master_param_out){param.size()}zTensor[] inputszTensor[](out){inputs.size()}min_with_indexzTensor x, float lambdaz/Tensor x,  int axis = -1,  bool keepdim = falsemoe_combinez6Tensor x, Tensor combine_weights, Tensor scatter_indexz	Tensor(y)moe_combine_automoe_combine_no_weightzNTensor x, Tensor combine_weight, Tensor scatter_index, float epsilon = 1.0e-15moe_gate_dispatchzYTensor x, Tensor gate_logits, Tensor corr_bias, int64_t k, int64_t capacity, bool use_padzcTensor(y), Tensor(combine_weights), Tensor(scatter_index), Tensor(expert_offset), Tensor(expert_id)moe_gate_dispatch_and_quantznTensor x, Tensor gate_logits, Tensor corr_bias, int64_t k, int64_t capacity, bool use_pad, bool use_pow2_scalezxTensor(out_fp8), Tensor(scale), Tensor(combine_weights), Tensor(scatter_index), Tensor(expert_offset), Tensor(expert_id)moe_gate_dispatch_auto'moe_gate_dispatch_partial_nosoftmaxtopkzTensor x, Tensor combine_weights, Tensor expert_id, int64_t k, int64_t capacity, int64_t num_experts, bool use_pad, int64_t expert_start_index, int64_t expert_end_index, bool reverse_token_dropzTensor(y), Tensor(combine_weights_out), Tensor(scatter_index), Tensor(scatter_index_rev), Tensor(expert_offset), Tensor(expert_nums_local)moe_gate_dispatch_permutez_Tensor x, Tensor gate_logits, Tensor corr_bias, int64_t k, int64_t capacity, int64_t world_sizemoe_permutea
  Tensor hidden_states, Tensor scale, Tensor expert_routemap_topk, Tensor expert_prob_topk, int num_experts, int[] tokens_per_expert, int padding_alignment, bool do_gather, bool using_ue8m0_scale = false, bool return_expert_indices=false, int override_buffer_size = -1zTensor(hidden_states_unzipped), Tensor(zipped_expertwise_rowmap), Tensor(token_prob_unzipped), Tensor(scale_unzipped), Tensor(expert_indices)moe_unpermutezTensor hidden_states_unzipped, Tensor zipped_expertwise_rowmap, Tensor expert_routemap_topk, Tensor token_prob_unzipped, int total_zipped_tokens_num, int num_experts, bool use_mix_precision, bool using_weighted_combine=falsez/Tensor(hidden_states), Tensor(expert_prob_topk)zTensor param, Tensor grad, Tensor velocity, Tensor learning_rate, Tensor master_param, float mu, bool use_nesterov = false, str regularization_method = "", float regularization_coeff = 0.0f, bool multi_precision = false, float rescale_grad = 1.0fzATensor(param_out), Tensor(velocity_out), Tensor(master_param_out)z
Tensor[] xzTensor bboxes, Tensor scores, Tensor rois_num, float score_threshold, int nms_top_k, int keep_top_k, float nms_threshold=0.3, bool normalized=true, float nms_eta=1.0, int background_label=0z0Tensor(out), Tensor(index), Tensor(nms_rois_num)z?Tensor x, Scalar(int) num_samples = 1, bool replacement = falsezTensor[] inputs, Tensor indexzTensor x, Tensor vecnadam_a$  Tensor param, Tensor grad, Tensor learning_rate, Tensor momentum_decay_pow, Tensor beta2_pow, Tensor mu_product, Tensor moment1, Tensor moment2, Tensor master_param, float beta1 = 0.9f, float beta2 = 0.999f, float epsilon = 1.0e-8f, float momentum_decay = 0.004f, bool multi_precision = falsezTensor(param_out), Tensor(momentum_decay_pow_out), Tensor(beta2_pow_out), Tensor(mu_product_out), Tensor(moment1_out), Tensor(moment2_out), Tensor(master_param_out)nansumzRTensor x, IntArray axis={}, DataType dtype=DataType::UNDEFINED, bool keepdim=false	nextafterz^Tensor input, Tensor label, Tensor weight, int64_t ignore_index = -100, str reduction = "mean"z!Tensor(out), Tensor(total_weight)z Tensor x, float threshold = 1.0fzTensor conditionz/Tensor x, int axis, float epsilon, bool is_testzTensor(out), Tensor(norm)npu_identityzTensor x, int format = -1zTensor(size)z!Tensor x, Scalar(int) num_classesones	ones_likez<Tensor x, DataType dtype=DataType::UNDEFINED, Place place={}z%Tensor x, int hop_length, int axis=-1zkTensor x,  float porder=2,  int axis=-1,  float epsilon=1.0e-12f,  bool keepdim=false,  bool asvector=falsez*Tensor x, int[] paddings, Scalar pad_valuezeTensor x, IntArray paddings, str mode = "constant", double pad_value = 0.0, str data_format = "NCDHW"z/Tensor x, int nranks, int rank, int ring_id = 0z0Tensor[] x, int start_index = 0, int length = -1z6Tensor x, int upscale_factor=1, str data_format="NCHW"z8Tensor x, int downscale_factor=1, str data_format="NCHW"	polygammazTensor x, IntArray kernel_size, int64_t[] strides, int64_t[] paddings, bool ceil_mode, bool exclusive, str data_format, str pooling_type, bool global_pooling, bool adaptive, str padding_algorithmzTensor x, int64_t[] kernel_size, int64_t[] strides, int64_t[] paddings, bool ceil_mode, bool exclusive, str data_format, str pooling_type, bool global_pooling, bool adaptive, str padding_algorithmzTensor x, Scalar y=1.0fz>Tensor x, Tensor alpha, str data_format="NCHW", str mode="all"zTensor input, Tensor image, float[] min_sizes, float[] max_sizes = {}, float[] aspect_ratios = {}, float[] variances = {}, bool flip=true, bool clip=true, float step_w=0.0, float step_h=0.0, float offset=0.5, bool min_max_aspect_ratios_order=falsezTensor(out), Tensor(var)z6Tensor x, IntArray axis, bool keepdim, bool reduce_allzLTensor gate_idx, Tensor expert_count, int64_t n_expert=0, int64_t n_worker=0zTensor(out_gate_idx)zTensor x, Tensor boxes, Tensor boxes_num, int pooled_height=1, int pooled_width=1, int output_channels=1, float spatial_scale=1.0zdTensor arr, Tensor indices, Tensor values, int axis, str reduce = "assign", bool include_self = truea=  Tensor x, Tensor w, Tensor white_list, Tensor black_list, int num_emb = 0, int space_len = 0, int pyramid_layer = 2, int rand_len = 0, float drop_out_percent = 0, int is_training = 0, bool use_filter = true, int white_list_len = 0, int black_list_len = 0, int seed = 0, float lr = 0.0, str distribute_update_vars = ""z4Tensor (out), Tensor (drop_pos), Tensor (x_temp_out)zTensor x, str mode = "reduced"zTensor(q), Tensor(r)radam_zTensor param, Tensor grad, Tensor learning_rate, Tensor beta1_pow, Tensor beta2_pow, Tensor rho, Tensor moment1, Tensor moment2, Tensor master_param, float beta1 = 0.9f, float beta2 = 0.999f, float epsilon = 1.0e-8f, bool multi_precision = falsezTensor(param_out), Tensor(beta1_pow_out), Tensor(beta2_pow_out), Tensor(rho_out), Tensor(moment1_out), Tensor(moment2_out), Tensor(master_param_out)zQint low, int high, IntArray shape, DataType dtype=DataType::INT64, Place place={}randomz"Tensor x, int64_t from, int64_t toz/Tensor prob, Tensor topk_value, Tensor topk_idxz%int n, DataType dtype, Place place={}zSTensor x, Tensor rank_offset, Tensor rank_param, int max_rank = 3, int max_size = 0z1Tensor(input_help), Tensor(out), Tensor(ins_rank)	read_filezIstr filename = "", DataType dtype=DataType::UINT8, Place place=CPUPlace()r  z?Tensor x, int ring_id = 0, int root_id = 0, int reduce_type = 0	reduce_aszTensor x, Tensor targetreduce_scatterz)Tensor x, int ring_id = 0, int nranks = 1zXTensor x, Tensor neighbors, Tensor count, Tensor hashtable_value, Tensor hashtable_indexz;Tensor(reindex_src), Tensor(reindex_dst), Tensor(out_nodes)z+Tensor x, float p, int axis, float max_normz9Tensor x, int repeats, int axis, int64_t output_size = -1z<Tensor x, Tensor repeats, int axis, int64_t output_size = -1zTensor x, IntArray shaperestrict_nonzeroz(Tensor condition, int64_t total_true_numzTensor x, IntArray axisrintzTensor param, Tensor mean_square, Tensor grad, Tensor moment, Tensor learning_rate, Tensor mean_grad, Tensor master_param, float epsilon = 1.0e-10f, float decay = 0.9f, float momentum = 0.0f, bool centered = false, bool multi_precision = falsezpTensor(param_out), Tensor(moment_out), Tensor(mean_square_out), Tensor(mean_grad_out), Tensor(master_param_outs)a  Tensor x, Tensor[] pre_state, Tensor[] weight_list, Tensor sequence_length, Tensor dropout_state_in, float dropout_prob=0.0, bool is_bidirec=false, int input_size=10, int hidden_size=100, int num_layers=1, str mode="RNN_TANH", int seed=0, bool is_test=falsezZTensor(out), Tensor(dropout_state_out), Tensor[](state){pre_state.size()}, Tensor(reserve)zTensor x, Tensor boxes, Tensor boxes_num, int pooled_height=1, int pooled_width=1, float spatial_scale=1.0, int sampling_ratio=-1, bool aligned=falsezjTensor x, Tensor boxes, Tensor boxes_num, int pooled_height=1, int pooled_width=1, float spatial_scale=1.0zTensor(out), Tensor(arg_max)z/Tensor x, IntArray shifts={}, int64_t[] axis={}zTensor x, int decimals = 0 rprop_zTensor param, Tensor grad, Tensor prev, Tensor learning_rate, Tensor master_param, Tensor learning_rate_range, Tensor etas, bool multi_precision=falsezXTensor(param_out), Tensor(prev_out), Tensor(learning_rate_out), Tensor(master_param_out)zDTensor x, float lower=1.0f/8, float upper=1.0f/3, bool is_test=falsezTensor(out), Tensor(noise)zGTensor x, Scalar scale=1.0, Scalar bias=0.0, bool bias_after_scale=truez;Tensor x, Tensor index, Tensor updates, bool overwrite=truez&Tensor x, Tensor index, Tensor updateszQTensor sorted_sequence, Tensor values, bool out_int32 = false, bool right = falsez0Tensor x, Tensor segment_ids, str pooltype="SUM"zTensor(out), Tensor(summed_ids)zfTensor x, float scale=1.0507009873554804934193349852946, float alpha=1.6732632423543772848170429916717z\Tensor x, Tensor src_index, Tensor dst_index, str reduce_op = "SUM", IntArray out_size = {0}zTensor(out), Tensor(dst_count)zxTensor x, Tensor y, Tensor src_index, Tensor dst_index, str message_op="ADD", str reduce_op="SUM", IntArray out_size={0}zNTensor x, Tensor y, Tensor src_index, Tensor dst_index, str message_op = "ADD"zTensor x, Tensor padding_data, Tensor filter, int context_length, bool padding_trainable = false, int context_start = 0, int context_stride = 1z1Tensor x, Scalar(int) max_len, DataType out_dtypezMTensor x, bool is_test=false, str pooltype = "AVERAGE", float pad_value = 0.0z Tensor (out), Tensor (max_index)setzWTensor x, Tensor source, int64_t[] dims = {}, int64_t[] stride = {}, int64_t offset = 0zTensor x, Tensor values, IntArray starts, IntArray ends, IntArray steps, int64_t[] axes, int64_t[] decrease_axes, int64_t[] none_axesz`Tensor param, Tensor learning_rate, Tensor grad, Tensor master_param, bool multi_precision=falsez+Tensor(param_out), Tensor(master_param_out)zTensor inputshape64zKTensor input, int index_num, int nshards, int shard_id, int ignore_value=-1z)Tensor x, Tensor seed, int startup_seed=0z2Tensor(out), Tensor(shuffle_idx), Tensor(seed_out)zTensor x, int group = 1zVTensor x, Tensor label, Tensor pos_weight, bool normalize=false, int ignore_index=-100zlTensor input, int64_t[] axes, IntArray starts, IntArray ends, int64_t[] infer_flags, int64_t[] decrease_axis
slogdet_v2zTensor(sign), Tensor(logdet)z4Tensor x, double beta = 1.0, double threshold = 20.0zfTensor q, Tensor k, Tensor v, Tensor offset, Tensor columns, Tensor key_padding_mask, Tensor attn_maskz7Tensor (out), Tensor (sparse_dot_sdd), Tensor (softmax)zWTensor weight, Tensor u, Tensor v, int dim = 0, int power_iters = 1, float eps = 1e-12fz-Tensor x, IntArray sections, Scalar(int) axiszTensor[]{sections.size()}z#Tensor x, int num, Scalar(int) axiszTensor[]{num}zTensor x, IntArray axis={}zTensor[] x, int axis = 0standard_gammaz4Tensor x, float scale_a=0.67f, float scale_b=1.7159fstdzXTensor x, int64_t[] axis={}, bool keepdim=false, bool unbiased=true, double correction=1zRTensor x, Tensor window, int n_fft, int hop_length, bool normalized, bool onesidedzFTensor x, int[] axes, IntArray starts, IntArray ends, IntArray stridesz$Tensor x, bool full_matrices = falsez Tensor(u), Tensor(s), Tensor(vh)svdvalsz	Tensor(s)swiglusync_batch_norm_zTensor x, Tensor mean, Tensor variance, Tensor scale, Tensor bias, bool is_test, float momentum, float epsilon, str data_format, bool use_global_stats, bool trainable_statisticsz$Tensor arr, Tensor indices, int axiszLTensor x, Tensor tree_info, int child_nums, DataType dtype = DataType::INT32z"Tensor (child), Tensor (leaf_mask)zTensor x, Tensor travel, Tensor layer, bool output_positive=true, int[] neg_samples_num_list={}, int[] layer_offset={}, int seed = 0, int dtype=2z)Tensor(out), Tensor(labels), Tensor(mask)zJTensor x, int seg_num, float shift_ratio = 0.25f, str data_format = "NCHW"z2Tensor x, float threshold = 1.0, float value = 0.0top_p_samplingzfTensor x, Tensor ps, Tensor threshold, Tensor topp_seed, int64_t seed=-1, int k=0, str mode="truncate"z@Tensor (out), Tensor(ids), Tensor(topk_scores), Tensor(topk_ids)zSTensor x, Scalar(int) k = 1, int axis = -1, bool largest = true, bool sorted = truetrans_layoutzTensor x, int[] permzSTensor x, Tensor y, bool upper=true, bool transpose=false, bool unitriangular=falsetrilzTensor x, int diagonal=0tril_indicesz>int rows, int cols, int offset, DataType dtype, Place place={}triutriu_indicesz<int row, int col, int offset, DataType dtype, Place place={}trunc_dividezpint[] shape, float mean, float std, int seed, float a, float b, DataType dtype=DataType::FLOAT32, Place place={}zTensor input, int axis = 0zMTensor[] {axis<0 ? input.dims()[input.dims().size()+axis]:input.dims()[axis]}zLTensor x, int[] kernel_sizes, int[] strides, int[] paddings, int[] dilationszTIntArray shape,  DataType dtype,  Scalar min,  Scalar max,  int seed, Place place={}ztTensor x, float min = -1.0, float max = 1.0, int seed = 0, int diag_num = 0, int diag_step = 0, float diag_val = 1.0zTensor input, int[] shape, int input_dim_idx = 0, int output_dim_idx = 0, float min=-1.0f, float max=1.0f, int seed=0, int diag_num=0, int diag_step=0, float diag_val=1.0f, DataType dtype=DataType::FLOAT32zvTensor x, bool return_inverse = false, bool return_counts = false, int[] axis = {}, DataType dtype = DataType::FLOAT32z*Tensor(out), Tensor(index), Tensor(counts)zjTensor x, Tensor indices, int[] ksize, int[] strides, int[] padding, IntArray output_size, str data_formatzTensor x, Tensor indices, int[] ksize, int[] strides={1,1,1}, int[] paddings={0,0,0}, int[] output_size={0,0,0}, str data_format="NCDHW"zTensor x, IntArray axis = {}zTensor x, int axis=0, int num=0zTensor[](out){num}zTensor[] x, Tensor found_infinite, Tensor prev_loss_scaling, Tensor in_good_steps, Tensor in_bad_steps, int incr_every_n_steps, int decr_every_n_nan_or_inf, float incr_ratio, float decr_ratio, Scalar stop_update=falsez\Tensor[](out){x.size()}, Tensor(loss_scaling), Tensor(out_good_steps), Tensor(out_bad_steps)r  r   
view_dtypezTensor input, DataType dtypez!Tensor input, int64_t[] dims = {}
view_slicez0Tensor input, int64_t begin_idx, int64_t end_idxz\Tensor potentials, Tensor transition_params, Tensor lengths, bool include_bos_eos_tag = truezTensor(scores), Tensor(path)zrTensor logits, Tensor label, Tensor logits_length, Tensor labels_length, int blank = 0, bool norm_by_times = falsez!Tensor(loss), Tensor(warpctcgrad)warprnntzrTensor input, Tensor label, Tensor input_lengths, Tensor label_lengths, int blank = 0, float fastemit_lambda = 0.0z"Tensor(loss), Tensor(warprnntgrad)weight_dequantizezJTensor x, Tensor scale, str algo = "weight_only_int8", int group_size = -1weight_only_linearzoTensor x, Tensor weight, Tensor bias, Tensor weight_scale, str weight_dtype, int arch = 80, int group_size = -1weight_quantizezKTensor x, str algo = "weight_only_int8", int arch = 80, int group_size = -1zTensor(out), Tensor(scale)weighted_sample_neighborszqTensor row, Tensor colptr, Tensor edge_weight, Tensor input_nodes, Tensor eids, int sample_size, bool return_eidsz:Tensor(out_neighbors), Tensor(out_count), Tensor(out_eids)z$Tensor condition, Tensor x, Tensor yzTensor x, Tensor img_size, int[] anchors={}, int class_num = 1, float conf_thresh = 0.01, int downsample_ratio = 32, bool clip_bbox = true, float scale_x_y=1.0, bool iou_aware=false, float iou_aware_factor=0.5zTensor(boxes), Tensor(scores)z&Tensor x, int[] anchors, int class_numa   Tensor boxes0, Tensor boxes1, Tensor boxes2, Tensor image_shape, Tensor image_scale, int[] anchors0, int[] anchors1, int[] anchors2, int class_num, float conf_thresh, int downsample_ratio0, int downsample_ratio1, int downsample_ratio2, bool clip_bbox, float scale_x_y, float nms_thresholdz!Tensor(out), Tensor(nms_rois_num)zTensor x, Tensor gt_box, Tensor gt_label, Tensor gt_score, int[] anchors={}, int[] anchor_mask={}, int class_num =1 , float ignore_thresh=0.7, int downsample_ratio=32, bool use_label_smooth=true, float scale_x_y=1.0z<Tensor(loss), Tensor(objectness_mask), Tensor(gt_match_mask)zeros
zeros_likez>Tensor x, DataType dtype=DataType::UNDEFINED, Place place = {}batched_gemmzMTensor lhs, Tensor rhs, int64_t[] batch_sizes, bool trans_lhs, bool trans_rhszTensor inference, Tensor label, Tensor seq_length, int num_chunk_types, str chunk_scheme = "IOB", int[] excluded_chunk_types = {}zTensor (precision), Tensor (recall), Tensor (f1_score), Tensor (num_infer_chunks), Tensor (num_label_chunks), Tensor (num_correct_chunks)fast_lnz2Tensor x, Tensor scale, Tensor bias, float epsilonz'Tensor(y), Tensor(mean), Tensor(invvar)fast_rms_normz%Tensor x, Tensor scale, float epsilonzTensor(y), Tensor(invvar)fp8_gemm_blockwise_a  Tensor A, Tensor A_scale, Tensor B, Tensor B_scale, Tensor input_result, Tensor bias, Tensor pre_gelu, Tensor workspace, bool transa, bool transb, bool grad, bool accumulate, bool use_split_accumulator, int math_sm_count, bool is_A_1d_scaled, bool is_B_1d_scaledz>Tensor (output), Tensor (pre_gelu_out), Tensor (workspace_out)fp8_quant_blockwisezTensor x, float epsilon, bool using_1x128_vec_quant, bool input_transpose, bool output_scale_transpose, bool return_transpose_only, bool using_e5m2, bool using_pow2_scale, bool using_ue8m0_scalezLTensor(out), Tensor(scale), Tensor(out_transposed), Tensor(scale_transposed)fused_rms_norm_extint_bincountz2Tensor x, int64_t low, int64_t high, int64_t dtypezTensor numbers, int upper_rangerms_normzKTensor x, Tensor scale, int64_t[] normalized_shape={}, double epsilon= 1e-5zETensor start, Tensor end, Tensor step, DataType dtype, Place place={}zETensor weight, Tensor x, int64_t start_index=0, int64_t vocab_size=-1zsTensor fpn_rois, Tensor rois_num, int min_level, int max_level, int refer_level, int refer_scale, bool pixel_offsetzTensor[](multi_fpn_rois){max_level - min_level + 1}, Tensor[](multi_level_rois_num){max_level - min_level + 1}, Tensor(restore_index)zTensor x, Scalar scale=1.0zTensor[] x, str equationzHTensor(out), Tensor[](inner_cache){x.size()}, Tensor[](xshape){x.size()}zBTensor x, Tensor weight, int64_t padding_idx=-1, bool sparse=falseembedding_grad_densezSTensor x, Tensor weight, Tensor out_grad, int64_t padding_idx=-1, bool sparse=falsezTensor(weight_grad)an  Tensor[] params, Tensor[] grads, Tensor learning_rate, Tensor[] moments1, Tensor[] moments2, Tensor[] moments2_max, Tensor[] beta1_pows, Tensor[] beta2_pows, Tensor[] master_params, Tensor skip_update, Scalar beta1, Scalar beta2, Scalar epsilon, int chunk_size, float weight_decay, bool use_adamw, bool multi_precision, bool use_global_beta_pow, bool amsgrad = falsea  Tensor[](params_out){params.size()}, Tensor[](moments1_out){params.size()}, Tensor[](moments2_out){params.size()}, Tensor[](moments2_max_out){params.size()}, Tensor[](beta1_pows_out){params.size()}, Tensor[](beta2_pows_out){params.size()}, Tensor[](master_params_out){params.size()}zKTensor x, Tensor y, Tensor bias, bool trans_x, bool trans_y, str activationz"Tensor(out), Tensor(reserve_space)zFTensor x, Tensor y, bool transpose_x = false, bool transpose_y = falsezTensor x, IntArray starts, IntArray ends, IntArray steps, int64_t[] axes, int64_t[] decrease_axes, int64_t[] none_axes, int64_t[] shape, Scalar[] valueszTensor x, int axiszTensor[] x, int ring_id = 0zTensor[](out){x.size()}tensor_unfoldz6Tensor input, int64_t axis, int64_t size, int64_t stepz$Tensor x, IntArray repeat_times = {}zpTensor x, bool return_index, bool return_inverse, bool return_counts, int[] axis, DataType dtype=DataType::INT64z=Tensor(out), Tensor(indices), Tensor(inverse), Tensor(counts)N)op_mapop_info     `/var/www/html/pdf-tiff/venv/lib/python3.13/site-packages/paddle/incubate/autograd/phi_ops_map.py<module>r     s*  Wf
	
 5
Wf
  
 # 
Wf
. 
 5
/Wf
@ 
 5
AWf
R  0"4+)
 $-1 0	
SWf
t +)
 $% 0
uWf
R +  '##)'
 $''.** 0
 %,
 %,
 %.
/%SWf
^ +!#)
 $%( 0	
_Wf
B +  '##)'
 $''.** 0
 %,
 %,
 %.
/%CWf
N 

 5
 ! $
OWf
l 
c
 5
mWf
~ +
 5
Wf
P 
 5
 
QWf
n $
 5

oWf
D !W
 h
 ",
EWf
b 
 5
 !
cWf
| 

 5

 +%
 +%	
}Wf
d 
 5
 !
eWf
~ 
 5
 !
Wf
X	 &W
 !(
	Y	Wf
l	 
 5
m	Wf
~	 
 5
 !
	Wf
X
 
 5

 &"(
 &"(
 &"(
Y
Wf
J 
 5
 &"(
KWf
h 
 5
 &"(
iWf
F 
  
	GWf
Z %
 #
	[Wf
n  
 5
oWf
@ 
 5
AWf
R 
 5
SWf
d 
 5
eWf
v 
wWf
D 
 5
EWf
V  
 5
WWf
h "5
iWf
t 
 5
uWf
F 

 5
	GWf
Z 
 5
[Wf
l $ 1- 1%:'#

 + 0!
mWf
X 
!!,
 ((
YWf
v 
 5
 
wWf
T 
 5
UWf
f 
 5

gWf
|  "
 !)%-+
 =
#}Wf
f 

 5
	gWf
z (

 */

{Wf
P 
 5
QWf
b $!'#	
 e
 =
cWf
@ 	
 5
 AWf
X %!'#	
 e
 =
YWf
v  

 5
 ""(
wWf
V %	
 )?%8
	WWf
j !

 5
	kWf
~ !
 5
Wf
P  

 5
	QWf
d !

 5
	eWf
x 


 5
	yWf
L  =
MWf
X #*%
 +

YWf
n 'S
 5
oWf
@ 
 5
AWf
R !

 5
	SWf
f #2

 !

%gWf
| 
 5
}Wf
N 
 5
OWf
` 
 5
aWf
r 
 5
sWf
D /

 -

!EWf
Z 
 5
[Wf
l $

 5
	mWf
@ )W
 .*C
	AWf
T 
 5

 %$
 %$	
UWf
z "
 5
{Wf
L %W
 )

 9
MWf
f + 0"4$7
 "!

 _
gWf
D 

 5
	EWf
X 
 5
 F
 "+
YWf
| '}Wf
B 
 5
CWf
T 

 8
	UWf
h &
 8
 ""(
iWf
J +
 8
 ""(
KWf
l 

 8
	mWf
@  &

 8
	A Wf
T  !

 8
	U Wf
h  

 5
i Wf
z  
 5
{ Wf
L! 
 5

 #& - #( /
M!Wf
v! 

 5
 E
w!Wf
P" !0

 !

#Q"Wf
f" 
 5
 5
g"Wf
~" 
 5
 ""(
"Wf
\# 


 3
	]#Wf
p# Kq#Wf
v# !
 5
w#Wf
H$ %	
 8
I$Wf
`$ &

 8
 #$ 2,	
a$Wf
@% !0
 8
 ""(
#A%Wf
b%  W
 h
 
c%Wf
|% (

 5
	}%Wf
P& '%!!
 ###	
Q&Wf
p& $

 5
	q&Wf
D' 
 5
E'Wf
V' &
 5
W'Wf
h' ""+)#1%
 $) 0"	
i'Wf
L( 
 5
M(Wf
^(  W
 5
_(Wf
p( 
 5
q(Wf
B) 
 5
C)Wf
T) W
 5
U)Wf
f) 

 5
	g)Wf
z) 
 5

 %,
 %")	
{)Wf
`* 

 5
	a*Wf
t* 


 5
	u*Wf
H+ !

 

   ,"
  &
#I+Wf
v+ Lw+Wf
|+ #&&	
 )
}+Wf
V, 

 ##
	W,Wf
j, 
 ##
	k,Wf
~, 
 5
,Wf
P- 
 )*

 F
Q-Wf
j- 
 '

k-Wf
@. %

 5
	A.Wf
T. 

 5
U.Wf
f. 

 5
 k
g.Wf
@/ 5
 &, 1
A/Wf
Z/ 

 5
	[/Wf
n/ 

 5
	o/Wf
B0 

 5
C0Wf
T0 
 5
U0Wf
f0 

 5
g0Wf
x0 
 5
 W
 "& 6
y0Wf
^1 

 5
	_1Wf
r1 
 5
s1Wf
D2 "
 5
 8
E2Wf
\2 
5

 !& !&	
]2Wf
x2 +:

 5
	-y2Wf
L3 )8
 #
	+M3Wf
`3 4C
 #
	6a3Wf
t3 -

 5
	 u3Wf
H4 +
 #
	I4Wf
\4 '6
 #
	)]4Wf
p4 6E!!!	
 ###	
8q4Wf
N5 +:!!!	
 ###	
-O5Wf
l5 "1!
 #%
$m5Wf
F6 &#
 $'
G6Wf
^6 	
 5
 #$,
_6Wf
~6 5
6Wf
J7 #
 5
K7Wf
\7 
 5
]7Wf
n7 
 5
o7Wf
@8 
 5
A8Wf
R8 
 5
 $"(
S8Wf
p8 #
 5
q8Wf
B9 *

 5
	C9Wf
V9 ) '"(
 '"(	
W9Wf
p9 * '"(
 '"(	
q9Wf
J: "1 '"(
 '"(	
$K:Wf
d: 
 

 '$
!e:Wf
@; 
 5
A;Wf
R; 
 5
S;Wf
d; "

 5
	e;Wf
x; 

 5
	y;Wf
L< 

 5
	M<Wf
`< 
 3
a<Wf
r< 
 5
s<Wf
D= $
 5
 !

 ""(
E=Wf
j= 5
k=Wf
v= 
 5
 $"(
w=Wf
T> &&"(
U>Wf
f> !+"")%%+'
 &)) 0,,!2
g>Wf
V? %! !!!!-.$"

x
l
 g
 x	

 L
 
 g
 	
 <
 $%9
  0
 L
 	
 n
  0
  !
" #
$ *B()
%W?Wf
b@ *"
 !)%-+
c@Wf
FA -<"!
 *B 0'
/GAWf
hA ."
 !)%-+
 iAWf
NB ",	
 h
 #$ 2,	
OBWf
rB *+	
  
sBWf
LC ",	
 h
 #$ 2,	
MCWf
pC +

 5
	qCWf
DD +

 5
	EDWf
XD +

 5
	YDWf
lD +

 5
	mDWf
@E (7	
 5
*AEWf
XE -&!
 +-)))	
 YEWf
BF %4
 "
'CFWf
bF '++-)-)#!#!
 ++!)!)'))
 "1!0,,	
9"cFWf
hG *)%)%/!'#!2.
 $7!2#6!2''!!

!iGWf
`H )
 +
aHWf
xH .+	
  
!yHWf
TI #2

	
 x
 V	

 	
 	
 
 	
 L
 
 L
 ;
 <
 
 <
  !
" ]#
$ %8,,)
. )
1%UIWf
NJ '

 5
	OJWf
bJ %
 5
cJWf
tJ  !!
 *+'
 '&,
!uJWf
`K !!!
 +-))))
 '&,
-aKWf
XL -
 "
 YLWf
pL !0
 

 .+-
#qLWf
RM +
 5
SMWf
dM &5
 5
(eMWf
vM 

 5
 "%
wMWf
VN 

 5
	WNWf
jN ! 

 5
	kNWf
~N 5
 &, 1
NWf
XO 
 5
YOWf
jO ('! $
 "*(
 ,-
kOWf
PP #
 5
QPWf
bP $
 5
cPWf
tP 

 5
	uPWf
HQ (	
 ! *$"
IQWf
hQ ,(
 $"
iQWf
FR #

 5
	GRWf
ZR "

 5
	[RWf
nR !

 8
	oRWf
BS  
 "
 =
CSWf
bS $
 5
cSWf
tS  
 5
uSWf
FT !
 5
GTWf
XT 
 5
YTWf
jT 
 5
kTWf
|T 
 5
 E
}TWf
TU 

 5
	UUWf
hU  

 F
	iUWf
|U 

 5
	}UWf
PV #
 
QVWf
pV  

 "

qVWf
FW !

 5
	GWWf
ZW 
 5
[WWf
lW 
 5
mWWf
~W #
 5

WWf
TX +&,&*
 5
 +&(
UXWf
zX +{XWf
@Y (75
*AYWf
LY "

 5
	MYWf
`Y "

 5
 E
aYWf
zY #
 %-
{YWf
TZ 
 8
UZWf
fZ 
 5
gZWf
xZ 

 5

 +%
 +%	
yZWf
`[ 
 5
a[Wf
r[ 
 5
s[Wf
D\ 
 5
E\Wf
V\  

 6
	W\Wf
j\ 

 5
	k\Wf
~\ 
  
	\Wf
R] 
 5
S]Wf
d] "%

 5
	e]Wf
x] +  ##)'

 $''**!1
y]Wf
d^  
 "
e^Wf
~^  
 5
 g
^Wf
V_ ,!'#	
 e
 =
W_Wf
t_ #
 5
 ^
 ", 5
u_Wf
Z` /' $
 "*(
 ,-
[`Wf
@a #  
 
 )(
AaWf
fa +!'#	
 e
 =
gaWf
Db $
 

 "& -
EbWf
fb 
 5

gbWf
|b  

 5
	}bWf
Pc 

 5
	QcWf
dc 
 5
ecWf
vc #!'#	
 e
 =
wcWf
Td KUdWf
Zd 
 5

[dWf
pd 

 5
	qdWf
De 

 5
EeWf
Ve 
 5
WeWf
he 
 5
ieWf
ze 
 5
{eWf
Lf  

 6
	MfWf
`f !
 5
afWf
rf "
 5
sfWf
Dg !

 5
	EgWf
Xg !
 5
YgWf
jg  

 5
	kgWf
~g !

 5
	gWf
Rh 
 5
ShWf
dh  
 5
ehWf
vh 
 5
whWf
Hi "

 5
	IiWf
\i 

 
	]iWf
pi 

 #$/	
 $"(
qiWf
Vj 

 
WjWf
nj *

 !

ojWf
Dk #

 3
	EkWf
Xk )
 
YkWf
pk 

 5
 %$
qkWf
Ll 
)

 5
	MlWf
`l  

  
alWf
xl "
 5
ylWf
Jm !%

 5
	KmWf
^m 
 5
 !

 ""(
_mWf
Dn +
 

 7
EnWf
^n +
 

 7
_nWf
xn 

 5
	ynWf
Lo 
 5
MoWf
^o 
 5
 !

 ""(
_oWf
Dp 
 5
EpWf
Vp !0 '"(
 '"(	
#WpWf
pp )
 5
qpWf
Bq "+  '##)

 $''.** 0
 %"(
 %"(
 %"(
-$CqWf
Lr &"+)
 $) 0
MrWf
jr c
 5
krWf
|r 
 5
 !

 ""(
}rWf
bs 

 5
	csWf
vs !

 5
	wsWf
Jt 
  
	KtWf
^t "+)
 $) 0
_tWf
|t #2!!
 ###	
%}tWf
Xu 
 5
YuWf
ju !!,
 h
 '&
kuWf
Lv $

 5
	MvWf
`v %!
 (
avWf
zv &	
 5
 )((
{vWf
\w !
 5
 ""(
]wWf
zw 

 5
	{wWf
Nx 

 5
	OxWf
bx 	

 5
	cxWf
vx 
 $

 U
wxWf
Ty $!'#	
 e
 =
UyWf
ry 
 *
syWf
Jz 

 ?
 
KzWf
bz 
 5
czWf
tz 
 
	uzWf
H{ 

 5
	I{Wf
\{ 
 E
]{Wf
n{ 
 5
 "-
o{Wf
L| !
 5
M|Wf
^| 
 5
_|Wf
p| 

 5
 $"(
q|Wf
N} GO}Wf
T} 
 5
 
 ")
U}Wf
x} '
 5
y}Wf
J~ $
 5
K~Wf
\~ "5
]~Wf
h~ !
 5
i~Wf
z~ #
 5
{~Wf
L %
 5
MWf
^ 
 5
_Wf
p 
 5
 7
 ""(
qWf
T@ 
 5
 7
U@Wf
l@ 

 5
 
 $-
m@Wf
PA 

 5
	QAWf
dA $
 5
eAWf
vA 

 

wAWf
LB 
 5
 !

 ""(
MBWf
rB  "
 5

sBWf
HC $
 8
 *
ICWf
hC IiCWf
nC 	
 
	oCWf
BD W
 h
 *
CDWf
^D %%!!
 ###	
_DWf
~D 5
 &, 1
DWf
XE 5
YEWf
dE 
 5

 &"(
 &"(
 &"(
eEWf
VF 
 5
WFWf
hF  
 5
iFWf
zF 
 5
{FWf
LG 
 5
MGWf
^G 

 5
	_GWf
rG 
 5
sGWf
DH '
 5
 !
EHWf
^H *9%

 5
 E
,_HWf
xH  W
 h
 #$"$	
yHWf
VI 
 

 "& -
WIWf
xI (

y
 h
 W	

 W
 F
 y
 h
 W
 W
 F
 y
 h
 W
 W
  F!
$
3
W
 <
 ^	

 
 	
 W
 <
 <
 ^
 
 	
 W
 <
 ^
  !
" 	#
$ &'%'%'/
)-yIWf
TK !!!
 !(,,*(,,*
!UKWf
RL 
 5
 ""(
SLWf
pL '#+)
 $%.*!1
qLWf
VM 
#'/	
 !/ 	
WMWf
tM "
 5

uMWf
JN "
 
KNWf
bN 
 5
 &-
cNWf
@O 
 5
AOWf
RO 

 5
	SOWf
fO 
 5
gOWf
xO "
yOWf
DP 
 5

 %,
 %")	
EPWf
jP  
 5

kPWf
@Q $ 
 5

AQWf
VQ "/

 5
	WQWf
jQ 5
kQWf
vQ "'

 %

wQWf
LR '
 5
MRWf
^R 
 5
_RWf
pR !$$
 $

 &)
qRWf
TS "$$	
 $

 &)
USWf
zS I{SWf
@T %

 5
	ATWf
TT #
 
 x
 "-
UTWf
xT &
 5
yTWf
JU 
+)	
 $ 0
KUWf
dU GeUWf
jU !S
 5
kUWf
|U "
 
	}UWf
PV !
 5
QVWf
bV #

 '!
cVWf
zV %

 5
	{VWf
NW 
 5
OWWf
`W 
 5
aWWf
rW 
 5
sWWf
DX 

 5
EXWf
VX 
 5
WXWf
hX W
 5

 #- 2 #+ 0
iXWf
RY 
 5
SYWf
dY 
 5
eYWf
vY 
 5
wYWf
HZ 
 5
IZWf
ZZ  
 5
 
[ZWf
rZ 
 5
sZWf
D[ 

 5
	E[Wf
X[ '=
Y[Wf
d[ $&, 1
	e[Wf
x[ " #- 2 #+ 0
y[Wf
V\  "+
W\Wf
h\ ,=
i\Wf
t\ #
 5

u\Wf
J] 
 5
 ""(
 ""(
K]Wf
t] $""(+
	u]Wf
H^ 
 5
I^Wf
Z^ 
 5
[^Wf
l^ 
 

 F
 ""(
m^Wf
R_ 
 3
S_Wf
d_ 
 5
e_Wf
v_ &5
 5
(w_Wf
H` #
 5

 #- 2 #+ 0 #. 3
I`Wf
|` 

 5
	}`Wf
Pa 
 5
 ! 
 ""(
QaWf
xa 

 

yaWf
Nb 
 5
ObWf
`b %"
 !)%-+
 =
#abWf
Jc %

 8
 F
KcWf
dc 

 5
ecWf
vc 
 5
wcWf
Hd !
 5
IdWf
Zd #&	
 #
[dWf
td !
 
udWf
Ne &
 5
OeWf
`e 
 5
 ", 5
aeWf
@f 
  

 ""
AfWf
`f 
  

 ""
afWf
@g 
 5
AgWf
Rg 
 5
 F
SgWf
jg &

 5
	kgWf
~g 
 5
gWf
Ph &!'#	
 e
 =
QhWf
nh S
 5
ohWf
@i  /5
"AiWf
Li S
 5
MiWf
^i 
 3
_iWf
pi 5

 %"(
 %"(	
 &, 1
qiWf
^j %
 5
_jWf
pj 
  	
qjWf
Hk (
 

IkWf
^k  

 5
 z
 ""(
_kWf
Dl  

 5
	ElWf
Xl 
 

 F
 "+ 0
YlWf
@m 
 3
AmWf
Rm *-!2*(
 ),*	
 #+
SmWf
~m LmWf
Dn $!!-
 
EnWf
\n +*	
 )
]nWf
vn $
 5

wnWf
Lo GMoWf
Ro !

 

SoWf
ho #
 5
ioWf
zo #''
 (
{oWf
Vp !!	
 /*
WpWf
rp 

 h
	spWf
Fq %
 5
GqWf
Xq  
 5
YqWf
jq 
 5
kqWf
|q %
 5
}qWf
Nr  $%
 %" 0 0"4
OrWf
nr OorWf
tr ""&	
 M
urWf
Ls #

 3
	MsWf
`s $

 
asWf
xs )

 +

ysWf
Nt  '/
 !#
OtWf
rt %+	
 $%
stWf
Lu 

 5
	MuWf
`u 
	
 '"'
auWf
@v .!!

 -$7+
!AvWf
Xv "1

 0

  0
 	

 y
 y
 
 
 "#6
 -.J
 -.J
 +
 <
 
  0
  	!
" M#
$ F%
$YvWf
Nw +
 

OwWf
dw 
 5
ewWf
vw 
 
	wwWf
Jx #7"5+
 $!2 0
KxWf
hx $*W
 5
&ixWf
zx  /

  1

"{xWf
Py $3

  1

&QyWf
fy "+
 5
 ! $ 20!40!4!4#8
gyWf
Tz (

 5
	UzWf
hz $
 5
izWf
zz '!	
 $!
{zWf
T{ "h
 F
U{Wf
f{  /

 5
	"g{Wf
z{ ,

 $#%	
{{Wf
T| $3
 5
&U|Wf
f| 
	
 &'=)	
g|Wf
D} '	
 !2
E}Wf
`} #
 5
a}Wf
r} $"+)
 $) 0
s}Wf
P~ !

 5
 "(
Q~Wf
p~ '5
q~Wf
|~ &
 5
}~Wf
N 	
 5
OWf
f *

 5
	gWf
z 
 %"3	
{Wf
Z@ 	
 
 X
[@Wf
v@ 
 5
w@Wf
HA  
 5
IAWf
ZA &
 5
[AWf
lA 
+!2!2'=	
 ++
mAWf
PB 

 5
QBWf
bB "y
 5
cBWf
tB "
uBWf
@C ,!)

 L
	ACWf
TC "%%	
 !&
UCWf
pC $&"
 5

qCWf
FD $'%
 &!
 "!
GDWf
hD %

 5
	iDWf
|D 5
}DWf
HE #$00
 )($
IEWf
fE 
 
	gEWf
zE 
{EWf
FF #)
 5
 "4-+-	
GFWf
hF #
 #
	iFWf
|F 
 5

 ' 2
 ' 0
 ' 1
}FWf
lG ,
 5

 ' 2
 ' 0
 ' 1
0 8

 5
	* %	
 5
 '  0#
 , 
$ &"+)
 $) 0
 F
 "%
'6 &
 5
 

 5
	 '
 5
 '
 5
 %
 5
 &
 5
 5W
 5
' %

 5
	[LWf
pLA#	A#
 8DA# lA# A#" #A#*  Ui+A#2  ]S3A#:  g q;A#B  jiCA#J  l qKA#R A SA#Z S[A#b V cA#j Q"kA#r 
AsA#z 9{A#B @CA#J +KA#R nSA#Z A[A#b AcA#j kA#r 
AsA#z  Z.{A#B  CA#J KA#R  z.SA#Z ) [A#b cA#j kA#r Q0sA#z {A#B CA#J ^KA#R  Q]SA#Z [A#b cA#j )kA#r @sA#z _{A#B CA#J $KA#R SA#Z  f J[A#b 
 pKcA#j  J ckA#r SsA#z ){A#B 5CA#J ,KA#R  QXSA#Z [A#b  v"cA#j @kA#r  v"sA#z E{A#B +CA#J ]UKA#R $SA#Z ?[A#b cA#j $kA#r ?sA#z ${A#B 
$CA#J .#KA#R  u&SA#Z 9[A#b  *cA#j )i;+kA#r VsA#z l{A#B	 VC	A#J	 gK	A#R	 #r1%S	A#Z	 i[	A#b	  nHc	A#j	 8*!k	A#r	 *s	A#z	 {	A#B
 -C
A#J
 @K
A#R
  *C"S
A#Z
  Z1[
A#b
 ,c
A#j
 6k
A#r
  UNs
A#z
 @{
A#B *CA#J  ZHKA#R z8SA#Z *[A#b +cA#j  kA#r  sA#z  m{A#B  zCA#J  IKA#R  sSA#Z 6[A#b $cA#j  XkA#r 
sA#z {A#B Q)CA#J FKA#R 2SA#Z ! ^1#[A#b 0@&2cA#j u;kA#r  shsA#z I0{A#B I0CA#J NKA#R hSA#Z 
; [A#b GcA#j ~9kA#r 1sA#z  j{A#B ( CA#J  KA#R  LSA#Z ! m#[A#b  VcA#j 9kA#r 'sA#z 
{A#B 
 wCA#J _KA#R  ]kSA#Z E[A#b LcA#j HkA#r sA#z {A#B "($CA#J 3KA#R 
$SA#Z  R%[A#b  ^-cA#j i4kA#r 
0sA#z ,0{A#B CA#J B=KA#R 
.SA#Z J)[A#b %A'cA#j ZkA#r RsA#z !(#{A#B $CA#J 
KA#R SA#Z 
[A#b /cA#j AkA#r   G)"sA#z {A#B %CA#J 
gKA#R +o-SA#Z )l2+[A#b 4V26cA#j 9 kA#r B2sA#z 'B2){A#B 6 eX8CA#J + eX-KA#R " QF$SA#Z K[A#b ecA#j ZkA#r 2sA#z H{A#B TCA#J  MZKA#R  {ZSA#Z  oZ[A#b  _4cA#j  N4kA#r " xZ$sA#z  YZ{A#B  D4CA#J AKA#R & SA#Z [A#b $cA#j $kA#r ssA#z x-{A#B {-CA#J IKA#R KSA#Z  mZ[A#b pcA#j kA#r  QsA#z [{A#B `CA#J PKA#R  A KSA#Z  K K [A#b  ZFcA#j 'kA#r (*sA#z ${A#B CA#J 7KA#R (SA#Z ,[A#b acA#j CkA#r 5sA#z  SQ{A#B TCA#J TKA#R hoSA#Z  QD[A#b ycA#j y=kA#r 
 KqsA#z  VN{A#B UCA#J 1 KA#R A SA#Z 9[A#b $cA#j .!kA#r ysA#z z?{A#B 91CA#J 	KA#R 
SA#Z 	[A#b 
cA#j -kA#r  F sA#z  {A#B -CA#J IKA#R  T SA#Z  m [A#b ( m *cA#j QkA#r (sA#z 6{A#B 7CA#J IIKA#R  v"SA#Z [A#b cA#j okA#r sA#z {A#B  C A#J  [K A#R  $S A#Z  N0[ A#b  c A#j  G k A#r   s Ys A#z  d?{ A#B! 8C!A#J! 3K!A#R! S!A#Z! D[!A#b!  v"c!A#j! Wk!A#r! Ws!A#z! `{!A#B" 
C"A#J" K"A#R" S"A#Z" ["A#b" ;c"A#j" )k"A#r" es"A#z" ${"A#B# C#A#J# $K#A#R# $S#A#Z# -[#A#b# c#A#j# dk#A#r# Vs#A#z# @ {#A#B$  ~C$A#J$  CdK$A#R$ L^S$A#Z$ 	->[$A#b$ ?c$A#j$ Z6k$A#r$  Q1s$A#z$ 5 {$A#B% " nT$C%A#J% ' K%A#R% =.S%A#Z%  f?[%A#b% !c%A#j% Vk%A#r% Js%A#z% _{%A#B& 
@C&A#J&  j-K&A#R&  q-S&A#Z& [3[&A#b& 4c&A#j& @k&A#r& s&A#z& U0{&A#B' .C'A#J' .K'A#R' ! }N#S'A#Z' ['A#b'  ] ac'A#j'  U Fk'A#r' !0s'A#z' [3{'A#B( (C(A#J( A0K(A#R( HS(A#Z( H[(A#b( `c(A#j( kwk(A#r( " A M$s(A#z( kw{(A#B) . T _0C)A#J)  qw"K)A#R)  ] bS)A#Z)  sC[)A#b)  KUc)A#j) +k)A#r) s)A#z)  PD{)A#B* QC*A#J* /K*A#R* 	&S*A#Z*  w y[*A#b* U0c*A#j* dk*A#r*  v"s*A#z* ${*A#B+ r5C+A#J+ 
2K+A#R+ "S+A#Z+ A-[+A#b+ +c+A#j+  k+A#r+ 3s+A#z+ Z{+A#B, NC,A#J, 7K,A#R, }S,A#Z, 
<[,A#b, {c,A#j, Ak,A#r, Bs,A#z, B{,A#B- JC-A#J- LK-A#R- S-A#Z- ![-A#b-  Vc-A#j-  Wk-A#r- 
)s-A#z- T{-A#B.  J,C.A#J. HK.A#R. ^(S.A#Z.  T[.A#b. xc.A#j.  RHk.A#r. 	2(s.A#z.  H i{.A#B/ cC/A#J/ 4K/A#R/ AS/A#Z/ 7[/A#b/ eEc/A#j/ ]k/A#r/  s/A#z/ {/A#B0 QC0A#J0 )K0A#R0 ;S0A#Z0 jO[0A#b0 c0A#j0 k0A#r0 =s0A#z0 K{0A#B1 *N,C1A#J1 *K1A#R1 : S1A#Z1 )[1A#b1 c1A#j1  F Ek1A#r1 
 Vns1A#z1  h{1A#B2 |0C2A#J2 AK2A#R2 -S2A#Z2  il[2A#b2 V.c2A#j2 k2A#r2 Ys2A#z2 M{2A#B3 8C3A#J3 cK3A#R3 D3S3A#Z3 x[3A#b3 p2c3A#j3  O2k3A#r3 bs3A#z3  b {3A#B4 CC4A#J4 a4K4A#R4 
i S4A#Z4  X[4A#b4 r?c4A#j4 k4A#r4 s4A#z4 ]{4A#B5  C5A#J5 ;FK5A#R5 )S5A#Z5 [5A#b5 (h*c5A#j5 k5A#r5 s5A#z5 
{5A#B6 C6A#J6 ~K6A#R6 S6A#Z6 0[6A#b6 Fc6A#j6 1k6A#r6 s6A#z6 ${6A#B7 xKC7A#J7 iK7A#R7 ?-S7A#Z7 5![7A#b7 c7A#j7 k7A#r7 s7A#z7 ,{7A#B8 * C8A#J8 K8A#R8 FS8A#Z8 
j[8A#b8 d c8A#j8 Xk8A#r8 
ds8A#z8 
64{8A#B9 C9A#J9 $K9A#R9 S9A#Z9  D K[9A#b9 c9A#j9 6k9A#r9 
s9A#z9 {9A#B: C:A#J: ^6K:A#R:  d=S:A#Z: ^[:A#b: Dc:A#j: zTk:A#r: e0s:A#z: H{:A#B; &C;A#J; &K;A#R; eS;A#Z; *[;A#b; Pc;A#j;  v"k;A#r; *s;A#z; N{;A#B< C<A#J< $K<A#R<   C"S<A#Z< ,a[<A#b< ^c<A#j< fk<A#r<  Gs<A#z< % ` '{<A#B=  I>C=A#J= |K=A#R=  ]S=A#Z= .[=A#b= 1&c=A#j=  lpk=A#r= 
js=A#z= A{=A#B> .C>A#J> 3K>A#R> BS>A#Z> n0[>A#b>  E5c>A#j>  E6k>A#r> ^s>A#z>  B{>A#B? _.C?A#J?   DN"K?A#R? 6S?A#Z?  d1[?A#b? 8c?A#j?  s5k?A#r?  jPs?A#z? Z{?A#B@ PC@A#J@ _"K@A#R@  V ^S@A#Z@ D;[@A#b@ 7-c@A#j@  YRk@A#r@  U`s@A#z@ 7-{@A#BA DCAA#JA 1KAA#RA ]-SAA#ZA 
$[AA#bA !cAA#jA WkAA#rA sAA#zA  D K{AA#BB WCBA#JB  F Z!KBA#RB ,SBA#ZB $[BA#bB *\cBA#jB $kBA#rB TsBA#zB e'{BA#BC $CCA#JC $KCA#RC  A oSCA#ZC ]6[CA#bC $cCA#jC $kCA#rC sCA#zC ${CA#BD $CDA#JD XKDA#RD $SDA#ZD 
@[DA#bD $cDA#jD $kDA#rD $sDA#zD W{DA#BE $ CEA#JE  kKEA#RE $SEA#ZE $[EA#bE -+cEA#jE HkEA#rE 6sEA#zE  CQ{EA#r  