From 6d3cafc579e35d47bf8b2f941ace3d0c40196d08 Mon Sep 17 00:00:00 2001 From: mongar Date: Sat, 4 May 2024 21:34:43 -0500 Subject: [PATCH] Se corrigen algunos errores en los modelos de ollama --- .../streamlit_tools.cpython-311.pyc | Bin 2307 -> 2135 bytes chats/chat_tools.py | 46 +++++++++++------- chats/streamlit_tools.py | 8 +-- .../data_level0.bin | Bin 6284000 -> 6284000 bytes .../length.bin | Bin 4000 -> 4000 bytes .../__pycache__/lc_tools.cpython-311.pyc | Bin 9317 -> 8815 bytes .../__pycache__/pdf_tools.cpython-311.pyc | Bin 2962 -> 2708 bytes langchain_tools/lc_tools.py | 6 +-- langchain_tools/pdf_tools.py | 4 +- 9 files changed, 38 insertions(+), 26 deletions(-) diff --git a/chats/__pycache__/streamlit_tools.cpython-311.pyc b/chats/__pycache__/streamlit_tools.cpython-311.pyc index b2463fdd6be10bff9dbfd9eda3ad24d636817bf7..f77c1388133ffdef873b2b33071beaf2c83d5323 100644 GIT binary patch delta 400 zcmZn`x-P)GoR^o20SN9cGD{1Z$eYddj&Wj@Eo0uq9hTahAPERaVXa}B%`lfKg>4xV z1H)<{h5%EBTIL$&1+0@9*+f`s7#FaC7?US3Mo#|2#^ucp(X@;as0D~2fRQ1EqXfnQ z(pelZP8mayF?N|O&dGL6^4xHh3%H>6@J;Sw5|o6C$HSdm!xYS*$vJroli6e?W=}@$ z$vMnQ(k%G}sd=|p%ZoBgQg5*or6#7_VlOUDPEIW@p1h8E36})WC2T-kY%n>7#f#}H z&*bwgZv}WYoEVu8F|s(ZvLE7LbYk6nhIJDoqrl{Pc2i+-pz2%P$vLTsMe)gr$r-6d z>XVPNZ`76ni8B|M6cvHYEK&y&D;bJ3Kx~kfUmP~M`6;D2sdhy=lNWQuuz+L$)lgGI delta 573 zcmcaE&@9BeoR^o20SMfC&C+})@@6xcF-@$pWt=o|hov+lNCE<~SYWIa))H<8hRJJL zB^+im%w<{zRJ0ndkddK;7p9PbA%(4m38W^4eHjx_6%a#!DMKxD4f6u_$%;(EtPC}b z3phZm$?l9%p$Ka@AbQcwPvIl!&NTehT6uj#mG>@ zgk4iSI8Z=_*DwV$XmU-?VKOr}bIno6E6q(U%2!B9RY)vK&d4mwFIGs+Q^-kF$jK}& zNmM|UC str: - text_source: str = "" for index, document in enumerate(documents_source): quote: str = document.page_content - source: str = document.metadata['source'].replace( - 'documents/pdfs/', '') - page: str = document.metadata['page'] + 1 - fuente: str = f"**Fuente #{index + 1}:** \n '{quote}'\n(*{source}, P.{page})*" + source: str = document.metadata["source"].replace("documents/pdfs/", "") + page: str = document.metadata["page"] + 1 + fuente: str = ( + f"**Fuente #{index + 1}:** \n '{quote}'\n(*{source}, P.{page})*" + ) text_source += fuente + "\n\n\n" diff --git a/chats/streamlit_tools.py b/chats/streamlit_tools.py index be6d840..b95b0ee 100644 --- a/chats/streamlit_tools.py +++ b/chats/streamlit_tools.py @@ -14,16 +14,16 @@ def import_file() -> str: # Agregamos el nombre a la lista de archvios para luego podr verificarlos List_of_files.append(nombre_archivo) - st.success( - f"El numero de archivos en la lista de archivos es de: {len(List_of_files)}" - ) + # st.success( + # f"El numero de archivos en la lista de archivos es de: {len(List_of_files)}" + # ) # Abrir un archivo en modo escritura binaria ('wb') para guardar el archivo de audio with open(f"documents/pdfs/{nombre_archivo}", "wb") as new_file: # Leer los datos del archivo cargado y escribirlos en el nuevo archivo new_file.write(archivo.read()) - st.success(f"Se carga el archivo con nombre: {nombre_archivo}") + # st.success(f"Se carga el archivo con nombre: {nombre_archivo}") # Verificamos que en la lista solo haya un archivo, de lo contrario, limpiamos la session_state if ( diff --git a/embeddings/etapas_del_conflicto_armado/2c0b725c-3f3b-4401-b0b8-1795745afa11/data_level0.bin b/embeddings/etapas_del_conflicto_armado/2c0b725c-3f3b-4401-b0b8-1795745afa11/data_level0.bin index ea3192e8ec5112eb8b4f2a7bde5d13bbb95e6ac0..8734d37ee3ab6ebd86d03ad7226a18ddcbfc9098 100644 GIT binary patch delta 22221 zcmeHP32+qG5#9IyNJv7*N+3W;fCM@b0&z-!gv22bLV!RBAx?1$%wY}z!j2hZ{IMY> z@!Al4V4Fj@%BdV4$Iely93&y9s}d4Bah!5uSMH>oqmoME9FXo=5L&#mth75jsc0#- zpl1I3ncu(r_3^v=?JL*C+w^m##bT(cd|(a3w|8V&lc!*3()? z=QXRbc3#)*oIK#%`_KJDF54IXk|>?I*_AwCn1SQa0fC{pWumh2h-^Lj9b2hFeBtUB zpX}=P$CGq6Thy$5%<98wr%oAoHAP46_0en5wE4ONQse^{X_v0cjmLi=X11Kb zh~il;sl^)D2|Ye>07g$+DHq<=sonj=?`@`n50qd`#TpNu9|_2!obxO zJubm5YBxP)D}(uv%olB)6FcumW!$3O-e>+Im+twbD4pGgw2XWMOQ35paTr!mDW)&E zkHJhkc=6v>6t&~JyZfZF`QT?g=*-24(KEYW_xV44persh&A=NZo5PE$W!uR&eFg;B zi!hIKS~l~2In7+%t~0+i_jpR(K4j)i^~!4rbRrpYCoB}zEyv|*{c9Rwa+s-ci_iS* zKyKidbzZ}O%pwoYWy^-0FIbJ0s}6s~z&l<|THXARcHqLltj5}T-D!~?ACS~q`*VMj ztM0ojs+KF>a#za05eDL+6Le#6_IQ+3t`oHzpOl-9zZOYQEbq;h$oj44Jvf_-fmy}c z!Os6!&B3OHf!`1$3?5M`TOWMYYM{;6orZ*2ejve*IGR-0^YousqSz)iRRY6BY+RBD zZ|O;CS!~jZ@mO8_@sdzoM+;) zL&hpzjIG)rW;7j@3%6aAYxFDH?kC^1o8S>8u3UHY+b$tvy1N~TDi?VZ7YACWn795T z2Hs0yDKzctKIbrcdCUeZWea%Y!`~-w^d(WZAy(3Oy! z?jrwGcu>O@vMVVylcOo@0ime8Sxl(eE~YO%Eat5B@D9nAd%j@$A~qN#5#?AvY?%`f z4w!--oGeI49;9tQ^$S1GdmkrrlRf|QaO-TA%5@KX%W`Z$gz15P^FNObT;=~o7If1? zUV+8LLC&j_+ShhJ^#_eT1qN&WQD}^frmo-S5B|t|kNj&M%fI$>e-Fo&>EW0pggG{c zM!q}kH5(i`W9NCG3P{z-mV3V#=vzh2y2lM9kWhfla&5<{Uj`a(+dchG%d+{t(@fFV zL1Ckv`@8$;Ke7er!Qs&q^sU?SY(O^Nb>??23N9K*b?dQliK43cSh%1k;Cua0Aj*~Y zVx=FF`)rZZsMPB7HFx^jp4>M@V$>85_7{^#cu$1_>u(fKUu7VRGzEE2vSHUpBN@8s zMu)zRluq-c)0niz9!Mag6geexJ=j?s6PrMxnoo4&Ilr&uaE^gO5(tX#utRsJ2W+|@ z6^OvVWfR%Ec;f$2AUdzHU(KN-4_Zr+KCC2|gD5l2p*yx@_+FoZaNN zm_I5NeOoF*2_pNa94Coz=*WYniKHULw1#_w@(bA18kjlSKm`c}opj5UANaBb({^Es zP4}Zh;eTswBKP~Y(`i*e+5HwCQ_WbjMK};&)F7qt`fz&q=SO1j39+Xr;NV15^O;4LN!fpKbKBVZnPiMe< z=wo0R3(1%{oT3!HVJ&yqafs$D1p^)2&nb0Vi&1J<(i2Y)7z++t;@(Dz^TZiEv%Eeds5_ zGEoTQ@~lCJ5^gw!j6-Dx4)KX$l9)Vyk9OB%?}YF|>-YI1M7L{EN$EA|!m?fe6E-46 zn7+iOl$rWR*<)Pd6jZ3H8zc3S8tFdd=lnyC15>?~+f2MHn7ov8GXo`3_wvO)@AWrF zR5I>JB7bBH+qH;G2Du8H-Ndn1&VTj9q;fv1S&0>NS7jQHE)`<7G zbt;B1<#GGtKM(Z0UH775;H((5g^sb73!pQ)?{Ogo6%r(Rkk>|8OypJXBMVPz*u zG2}252+fJRg@G5yv8MPI6{^a{y)Q%WcRVrz~W3C zh6yaGn2;ZGi2t09LiP1N9YW}hU&8RPl?i5TU=zI8n~Rw1t@0M3~HC4IS?`E z>vBG~ci7LRi`?PD?0#W3EP9h6n-j&w)W!F3wQj_V`y(EIyG=ZtW~Cxa0W}LvDJQ@z z!gdcHMWyewTNKY=E1iT<)^EEYw>|nZJG8sioK92AIdJ#NP67$fn98-Tl=N&WI!ZH( zj@XUjt`251kjS=)a*iBc2HDa(9J1mr^E#DOb9JzowdxUWJ>ac*tW2YGni%GMCcmg& z@i42Tpi??;llg$}w;1a*=hs-`gx}ThAv7y(HqN=z2vLqh{suOu>HXb9xTc7Exp2;r z+4jmTu#QasMdv8V4WKriNPpd=47I4}*4-!3!09xz>04Sx0Y;9i<&+s8FOZTA^5Bj6#XRScP#4r3&K}CMZl)n4~aSVTwYTLb<|J zg=q@Y6)F@e6=o<@DO4-WRG6hOTVakujlx`oc?z`(^A+k87AVv!EL2#euvno%VTr<0 fg+_&C3QY>sr>t3Fg+hzMN`+Mlt6#yIWBvaN4#JA^ delta 6149 zcmeIwF%1A93;@9d1@*t8fS`dL2dxu82f5)5e6H4M^_n{)dZj#IfB^;=V1NMz7+`<_ W1{h#~0R|XgfB^;=VBiA-BX+y*C$8QA diff --git a/embeddings/etapas_del_conflicto_armado/2c0b725c-3f3b-4401-b0b8-1795745afa11/length.bin b/embeddings/etapas_del_conflicto_armado/2c0b725c-3f3b-4401-b0b8-1795745afa11/length.bin index 1dc89f8e47e591af4c36a35046077f0ba1d1ef9d..eb1879c2efa3e77948c59433ede343ba61caf41a 100644 GIT binary patch literal 4000 zcmcIndpOkD8=s5ZG7NIt(6~iJ-wBcHh(ftcnAorprD(61F(}+4yXrIaIk z9vx=`ncQ3;C*?p2<(iP1Istn`xPCtmc>vCT2?sT(y&e}}@1OD;jSr3(7>LFb8|LGZA;_Tn zbPnRt{XoxaIo^*A&WYFosDIyKA)7L~NwmsOwecUWMLXomh6i$*6LO7<6k{`={D8Hv zPkt+T{xqT3+JviApZj*3^6@$||Nak6+luk&Lx!O*X18WXnT&SZY0J8Jy$ za9+6QL%(a+TApmB~VEMV>;8X~AW$7xiPevJT-3 zuVn3e_QbqSJv~P_Bt~Q*Pv*PYZ)5e^@fVj!fGdlN?_TN@NlB+p9 zRoLA-BRTcfcU?vfJz{f*fo9lNwW=F~bS>f3%W)2Avg--L&B3v|=1YS%CeL`5^?Gih zPx@&3zxp4d=wf%878ciaVp!?wSc3f#X%|QPQJs^xl_$5apOYDIoggO$$90xl|7Wu$ zPQFy}N>G)z_4-u3pv}oXEr}=Z1wJay39X(;@NO77XzwtTcx6!k`(05kx9M*}?Scn2 zr;cf!Q|xT_{B*IvxN(#i-E8Ub$ui1~MgqU|zonKv6yWalP>@|K3I$_Y~CUe~Cl&|0abc_Wv?LjsDU<1lAXHKMPtw zzEk+K2|hs+KsbVj2EU>A7+Sx8^z(1eE--$%J`awk#_KQcH@dH%M*>b_1qSgYuSCRz z@iZY}S=@FiSqsq+bi`$q-M8AhI+HxY|E?I3wGk}=3@@wdzAqp$oX(0}c4xvj6{4y@ zAa6o8Eq;PqN`YW#ijc{se3790Z<|lSsUrkoG2L{B^jQKHi(P@ir{JV8ND59Imx9A$ zp7ew}&LU%C2^T5JUUwVJ9{uPqb%Gj7w=nK{=^c@~I><@w`5xoe%~pEiey>_X^wK{v z&)=;2b=S$co|?MwK~fd#PP5$QouATJ=URxLRwT{2ow!1=j3idE^P4IXquQRVc2nw? zUb9ljXl8hZZV~-1aGG!_ZHy~M6;jAQ+;sCun{)WQa_jAP!p?JN17{P<_oo{(Z%QXQ z{luhrsdkh^yq~wCKIU#%YnkYy(HP4Z`@rRt-IuUo#?MY=D`$2;KRUzNK+E5ju-n|r z{NsNgGFcGx6(ibdK{X}|22_E!`4@4A?| zAls&&wc!_k?x~L6#`wqaMnNmrT$SifP*rm8t*F~BN7}WDVX2q0UaCk=#XiZN7gHcf zt_icsd67<6jaSTERo^wAVrWpfL;oYgoOwwvey#Squeu(7RA@do)mPF`efMNy+(6~E z=cfA72AQ3M)|X|>w^ZfeLJ!^I?D6k7*JgImVQMW+UEX&Ofv0VFRuRuS;_XJV|8}dh zdUQ5T`*n?Kc2BChN7I=B*OC7D5d-CXojTlIe7lyLveAB5jmqihlzbJACn74nV!@L$ z(!E&dO1B^sM1>#EDw0C-LaPE-CCqz;qi=$~w)vX>Rw9xDwGIxpf4-v zuR!p8!Ydf$@y-m)^9nf&JZV@r*#_WUg63zLMz{zm#cy13$m23y>v-|Vs@nxEZppW+ zgO8Z+-h49{YulwBf_uI97ncBA_NoWd0k5Z~nJ-;3o-i*?_jo=v*rOcVs&Og9&U3Qe z*x9Cahfe9~`qd*V6-1uSP8H2W4Rv0>n(W29$qS_KjlCQyx#qiK zJ@nXaAvv_`J$IV>dW}SPrj^XJH7IyZiTlL02mEZdftT;7ba*OgSg6i_8sojasgP-+ zlos2WT-z-x8O5|ROA@XYmajP0*mb>+M=A3;R@P(oB0Ra*=d{HgBYd>z!s~Tk`>Kg{ z`*mJm#y5|srgVta>Zo~`#T0LTHSw!! z!vlpy`}zj-w8?nYO^+?3A|-fR*S0*`*{0FGXLvY+NRc5}nwmVbvruHaiv{NJ?wu)5 zU>HQu)F#GxsVyp28&&U@A6oxr|D=yfMQQ&xS{wDAoT}0l>2J`9Nw?51FLJ)aLLLR~ zlp9S=-8odVU|8h5p9JqkwB9=gf_jbC;UxI(M&Ez2Q9#dV{CsZ+-kWG+%K#3=E5ZLC zwea0Zg#VLJJmV|Cv7n%5#%}Td@BG!5v*vZ1CW_x*-=^beIN)Wg8?f${b6rE+`4)m{ znXiNHLSKD^Z)v7zw*-!iVOQ!)hP+TbKk7cIHA)lZQg4mh`OBwNI@Z=FKaa?WX>h9F ayISUhyfb&1y%9QCPh$(U4*$nf$3Fq1)AG9j literal 4000 ocmeIuF#!Mo0K%a4Pi+hzh(KY$fB^#r3>YwAz<>b*1`NCh1`nVB0RR91 diff --git a/langchain_tools/__pycache__/lc_tools.cpython-311.pyc b/langchain_tools/__pycache__/lc_tools.cpython-311.pyc index 3dcd76d87998db6dcc6b1e007bdcb1ba828c2bc4..83006cd6ea2882975e714e71befc2e7271b0302e 100644 GIT binary patch delta 707 zcmaFr@!o}ZIWI340}$+5WR}*rk=Kb`p@fNnVLC$!V+lwY3bGhstOZPy1v!*B${35h zYnT==Pd>=5K6x6u=;Xib!joSB>HorfR@imMv%<_sVZ*KhV)o>I4q!m+Q9HZih)OTvnkJHCPvrIC;5&u zGP+EjC?F#13N-Q-Uv6SWd|qmKd`W(GYF=>>$R16WBKysU1m-asdjWYxE+8T1TdcYH zDXBR{U|sBarMdCRB^5>dV7VfDkh(w+;kS9RkRJ=shDTyaY?dGqtI1yCUW^W#r;5um zG96)pz7t532Z%5S>EugJOwLG+FG?-WFD*(=E%FD+1cQip5RtX{ iwSq2_R1t_R3nHX}#4ipTV2G6Fq}mnLZZ=XjW&{9mPmWOl delta 1197 zcmZ`%OK8+U7|z6YGudsnb+>iZcGsyOx=_U;sBBeQZN&%LQeWUJ#?83Rn#Vdxw564* z7ddr?6UB;ZjH1uq^H)=NO}B6v~{dhsC6q-M2Hf5JE4|Ihb7W+q<;KkZT; zDoRpdcH_pzGm}r1?b6n}q9EK7Jiw!KR&DSp@d4wH8$J+p;KeewR_QZo1$~}P!}Eab zof2o7=nJWXK9Mr?Gh5$FSsGf5c+b&75V6lnx2bbIKgAIF`M}~N4QP^xb zIyOm^CE32Emc4P(8tsvOt{der9E@i6H?a=0OHLrvMpgl**d$j}YP(vU-pieYQ!x@2;f z;d;!~Og4at%43>WN>1`c9gB7*E+$779)iN*#k!nyZ9!7sYV zVSvL)4iyGrOWCo{W8!LjE@HBf3k7;Re+)dL-}9H&bO}RXAGp?1hpX$bw+=Vg;lMXh z=<9F(6I)k@n?AO7((%In#139|H+@rB);7ZFK@O)lIKhv?T5*njJr NcpJGAZ0X(regnhsC!YWS diff --git a/langchain_tools/__pycache__/pdf_tools.cpython-311.pyc b/langchain_tools/__pycache__/pdf_tools.cpython-311.pyc index 6d7c67aebde950d44d03b0ac4eee84361ed5f39c..90093927c6e57df18fe343301c9744630bf960d7 100644 GIT binary patch delta 275 zcmbOvK1GywIWI340}zxgGD~~9k++|Piy6oR;m=uE;QD?FNo2vK+HUHopIoudM zCW~-7vVP!T5D=MM&FQRoi>bKe7GH8=az<)=QEG91X;E@&ku*>}&{dhmB}K}cA8_(A ks%U~tU;`04AOhr~B4Hr$i^C>2KczG$)vm~3vLJUO0Hn}Bg#Z8m delta 520 zcmbOtI!T;&IWI340}$+VHA{QAk++|Piv`GI2I9|)CNE=&_bA~6Nkc#uD~y%GSi_LT z2;Q~U~fy3$szsO`pHd!u^E5Sb4H(7%%IiZ9d>MgTMk4hfC$COAKBd)GbdYcII4lYH6dk6-h#j@V(KecRtT?Pxgw@}L0JDHhrtyN zg9{u65BP;A*Ks&Q%)P~zoS2-E8ef!JoL^d$oLVFcG!YnlnZ+eV>YE>N@G`1^t!D#S dqz@uQL4+`n_{Cw9o1apelWJFFJXx5l5dcnOZB+mO diff --git a/langchain_tools/lc_tools.py b/langchain_tools/lc_tools.py index a75ab81..8047cbd 100644 --- a/langchain_tools/lc_tools.py +++ b/langchain_tools/lc_tools.py @@ -68,7 +68,7 @@ class LangChainTools: """ # db_name = _file_name.replace(".pdf", "").replace(" ", "_").lower() - st.success(_file_name) + # st.success(_file_name) if "db_name" not in st.session_state.keys(): st.session_state.db_name = ( @@ -265,7 +265,7 @@ class LangChainTools: # Pregunta: {question} # Respuesta en español:""" - conversation.combine_docs_chain.llm_chain.prompt.template = template - conversation.question_generator.prompt.template = "Dado el siguiente diálogo y una pregunta de seguimiento, reformula la pregunta de seguimiento para que sea una pregunta independiente, en su idioma original.\n\nHistorial del chat:\n{chat_history}\nPregunta de seguimiento: {question}\nPregunta independiente:" + # conversation.combine_docs_chain.llm_chain.prompt.template = template + # conversation.question_generator.prompt.template = "Dado el siguiente diálogo y una pregunta de seguimiento, reformula la pregunta de seguimiento para que sea una pregunta independiente, en su idioma original.\n\nHistorial del chat:\n{chat_history}\nPregunta de seguimiento: {question}\nPregunta independiente:" return conversation diff --git a/langchain_tools/pdf_tools.py b/langchain_tools/pdf_tools.py index 2d3a1aa..fae3474 100644 --- a/langchain_tools/pdf_tools.py +++ b/langchain_tools/pdf_tools.py @@ -39,7 +39,7 @@ class PdfLangChain: loader = PyPDFLoader(_self.file_path) _self.docs = loader.load() - st.success(f"Se carga el pdf : {_self.file_path}") + # st.success(f"Se carga el pdf : {_self.file_path}") return _self.docs def split_docs(self, data: list) -> list: @@ -60,6 +60,6 @@ class PdfLangChain: ) self.docs_split = splitter.split_documents(data) - st.success(f"{self.file_path[3][:200]}") + # st.success(f"{self.file_path[3][:200]}") return self.docs_split