From 96081587f3253fd4296d40123c798f847d1f6904 Mon Sep 17 00:00:00 2001
From: hamshkhawar <hamdahshafqat@hotmail.com>
Date: Tue, 28 May 2024 10:36:36 -0500
Subject: [PATCH 1/5] tabnet-training

---
 .../tabnet-classifier-tools/CHANGELOG.md      |   1 +
 clustering/tabnet-classifier-tools/Dockerfile |  20 +
 clustering/tabnet-classifier-tools/README.md  |  73 ++
 clustering/tabnet-classifier-tools/VERSION    |   1 +
 .../tabnet-classifier-tools/build-docker.sh   |   4 +
 .../examples/feature_importances.json         |  16 +
 .../examples/tabnet_adult.zip                 | Bin 0 -> 41356 bytes
 .../tabnet-classifier-tools/plugin.json       | 672 ++++++++++++++++++
 .../tabnet-classifier-tools/pyproject.toml    |  31 +
 .../tabnet-classifier-tools/run-plugin.sh     |  93 +++
 .../clustering/pytorch_tabnet/__init__.py     |   3 +
 .../clustering/pytorch_tabnet/__main__.py     | 348 +++++++++
 .../clustering/pytorch_tabnet/tabnet.py       | 241 +++++++
 .../clustering/pytorch_tabnet/utils.py        | 224 ++++++
 .../tabnet-classifier-tools/tests/__init__.py |   1 +
 .../tabnet-classifier-tools/tests/conftest.py | 208 ++++++
 .../tabnet-classifier-tools/tests/test_cli.py |  50 ++
 .../tests/test_tabnet.py                      | 151 ++++
 18 files changed, 2137 insertions(+)
 create mode 100644 clustering/tabnet-classifier-tools/CHANGELOG.md
 create mode 100644 clustering/tabnet-classifier-tools/Dockerfile
 create mode 100644 clustering/tabnet-classifier-tools/README.md
 create mode 100644 clustering/tabnet-classifier-tools/VERSION
 create mode 100644 clustering/tabnet-classifier-tools/build-docker.sh
 create mode 100644 clustering/tabnet-classifier-tools/examples/feature_importances.json
 create mode 100644 clustering/tabnet-classifier-tools/examples/tabnet_adult.zip
 create mode 100644 clustering/tabnet-classifier-tools/plugin.json
 create mode 100644 clustering/tabnet-classifier-tools/pyproject.toml
 create mode 100644 clustering/tabnet-classifier-tools/run-plugin.sh
 create mode 100644 clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__init__.py
 create mode 100644 clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__main__.py
 create mode 100644 clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/tabnet.py
 create mode 100644 clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/utils.py
 create mode 100644 clustering/tabnet-classifier-tools/tests/__init__.py
 create mode 100644 clustering/tabnet-classifier-tools/tests/conftest.py
 create mode 100644 clustering/tabnet-classifier-tools/tests/test_cli.py
 create mode 100644 clustering/tabnet-classifier-tools/tests/test_tabnet.py

diff --git a/clustering/tabnet-classifier-tools/CHANGELOG.md b/clustering/tabnet-classifier-tools/CHANGELOG.md
new file mode 100644
index 0000000..425b725
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/CHANGELOG.md
@@ -0,0 +1 @@
+# PyTorch TabNet tool(0.1.0-dev0)
diff --git a/clustering/tabnet-classifier-tools/Dockerfile b/clustering/tabnet-classifier-tools/Dockerfile
new file mode 100644
index 0000000..48dd58e
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/Dockerfile
@@ -0,0 +1,20 @@
+FROM polusai/bfio:2.3.6
+
+# environment variables defined in polusai/bfio
+ENV EXEC_DIR="/opt/executables"
+ENV POLUS_IMG_EXT=".ome.tif"
+ENV POLUS_TAB_EXT=".arrow"
+ENV POLUS_LOG="INFO"
+
+# Work directory defined in the base container
+WORKDIR ${EXEC_DIR}
+
+COPY pyproject.toml ${EXEC_DIR}
+COPY VERSION ${EXEC_DIR}
+COPY README.md ${EXEC_DIR}
+COPY src ${EXEC_DIR}/src
+
+RUN pip3 install ${EXEC_DIR} --no-cache-dir
+
+ENTRYPOINT ["python3", "-m", "polus.tabular.clustering.pytorch_tabnet"]
+CMD ["--help"]
diff --git a/clustering/tabnet-classifier-tools/README.md b/clustering/tabnet-classifier-tools/README.md
new file mode 100644
index 0000000..642cb8c
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/README.md
@@ -0,0 +1,73 @@
+# PyTorch TabNet tool(v0.1.0-dev0)
+
+This tool uses [tabnet](https://github.com/dreamquark-ai/tabnet/tree/develop), a deep learning model designed for tabular data structured in rows and columns. TabNet is suitable for classification, regression, and multi-task learning.
+
+## Inputs:
+
+### Input data:
+The input tabular data that need to be trained. This plugin supports `.csv`, `.feather`and `.arrow` file formats
+
+### Details:
+
+PyTorch-TabNet can be employed for:.
+
+1. TabNetClassifier: For binary and multi-class classification problems
+2. TabNetRegressor: For simple and multi-task regression problems
+3. TabNetMultiTaskClassifier: multi-task multi-classification problems
+
+
+## Building
+
+To build the Docker image for the conversion plugin, run
+`./build-docker.sh`.
+
+## Install WIPP Plugin
+
+If WIPP is running, navigate to the plugins page and add a new plugin. Paste the contents of `plugin.json` into the pop-up window and submit.
+For more information on WIPP, visit the [official WIPP page](https://isg.nist.gov/deepzoomweb/software/wipp).
+
+## Options
+
+This plugin takes 38 input arguments and one output argument:
+
+| Name             | Description                                                                 | I/O    | Type          |
+| ---------------- | --------------------------------------------------------------------------- | ------ | ------------- |
+| `--inpdir`       | Input tabular data                                                          | Input  | genericData   |
+| `--filePattern`  | Pattern to parse tabular files                                              | Input  | string        |
+| `--testSize`     | Proportion of the dataset to include in the test set                        | Input  | number        |
+| `--nD`           | Width of the decision prediction layer                          | Input  | integer       |
+| `--nA`           | Width of the attention embedding for each mask                          | Input  | integer       |
+| `--nSteps`       | Number of steps in the architecture                         | Input  | integer       |
+| `--gamma`        | Coefficient for feature reuse in the masks                         | Input  | number       |
+| `--catEmbDim`    | List of embedding sizes for each categorical feature                          | Input  | integer       |
+| `--nIndependent`    | Number of independent Gated Linear Unit layers at each step          | Input  | integer       |
+| `--nShared`    | Number of shared Gated Linear Unit layers at each step          | Input  | integer       |
+| `--epsilon`     | Constant value                        | Input  | number        |
+| `--seed`    | Random seed for reproducibility          | Input  | integer       |
+| `--momentum`     | Momentum for batch normalization                        | Input  | number        |
+| `--clipValue`     | Clipping of the gradient value                        | Input  | number        |
+| `--lambdaSparse`     | Extra sparsity loss coefficient                        | Input  | number        |
+| `--optimizerFn`     | Pytorch optimizer function                        | Input  | enum        |
+| `--lr`     | learning rate for the optimizer                        | Input  | number         |
+| `--schedulerFn`     | Parameters used initialize the optimizer                        | Input  | enum         |
+| `--stepSize`     | Parameter to apply to the scheduler_fn                        | Input  | integer         |
+| `--deviceName`     | Platform used for training                        | Input  | enum         |
+| `--maskType`     | A masking function for feature selection                        | Input  | enum         |
+| `--groupedFeatures`     | Allow the model to share attention across features within the same group | Input  | integer         |
+| `--nSharedDecoder`     | Number of shared GLU block in decoder | Input  | integer         |
+| `--nIndepDecoder`     | Number of independent GLU block in decoder | Input  | integer         |
+| `--evalMetric`     | Metrics utilized for early stopping evaluation                        | Input  | enum         |
+| `--maxEpochs`     | Maximum number of epochs for training | Input  | integer         |
+| `--patience`     | Consecutive epochs without improvement before early stopping | Input  | integer         |
+| `--weights`     | Sampling parameter only for TabNetClassifier | Input  | integer         |
+| `--lossFn`     | Loss function                        | Input  | enum         |
+| `--batchSize`     | Batch size                       | Input  | integer         |
+| `--virtualBatchSize`     | Size of mini-batches for Ghost Batch Normalization    | Input  | integer         |
+| `--numWorkers`     | Number or workers used in torch.utils.data.Dataloader    | Input  | integer         |
+| `--dropLast`     | Option to drop incomplete last batch during training    | Input  | boolean         |
+| `--warmStart`     | For scikit-learn compatibility, enabling fitting the same model twice    | Input  | boolean         |
+| `--targetVar`     | Target feature containing classification labels   | Input  | string         |
+| `--computeImportance`     | Compute feature importance    | Input  | boolean         |
+| `--classifier`     | Pytorch tabnet Classifier for training   | Input  | enum         |
+| `--preview`      | Generate JSON file of sample outputs                       | Input | boolean         |
+| `--outdir`       | Output collection                            | Output | genericData   |
diff --git a/clustering/tabnet-classifier-tools/VERSION b/clustering/tabnet-classifier-tools/VERSION
new file mode 100644
index 0000000..206c085
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/VERSION
@@ -0,0 +1 @@
+0.1.0-dev0
diff --git a/clustering/tabnet-classifier-tools/build-docker.sh b/clustering/tabnet-classifier-tools/build-docker.sh
new file mode 100644
index 0000000..eff76ca
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/build-docker.sh
@@ -0,0 +1,4 @@
+#!/bin/bash
+
+version=$(<VERSION)
+docker build . -t polusai/pytorch-tabnet-tool:${version}
diff --git a/clustering/tabnet-classifier-tools/examples/feature_importances.json b/clustering/tabnet-classifier-tools/examples/feature_importances.json
new file mode 100644
index 0000000..8e56739
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/examples/feature_importances.json
@@ -0,0 +1,16 @@
+{
+    "capital-gain": 0.2208,
+    "marital-status": 0.1949,
+    "educational-num": 0.1724,
+    "hours-per-week": 0.1041,
+    "age": 0.0789,
+    "gender": 0.0756,
+    "occupation": 0.0635,
+    "relationship": 0.0264,
+    "capital-loss": 0.0225,
+    "native-country": 0.0223,
+    "race": 0.0102,
+    "fnlwgt": 0.0061,
+    "education": 0.0016,
+    "workclass": 0.001
+}
\ No newline at end of file
diff --git a/clustering/tabnet-classifier-tools/examples/tabnet_adult.zip b/clustering/tabnet-classifier-tools/examples/tabnet_adult.zip
new file mode 100644
index 0000000000000000000000000000000000000000..dfff7a25638627153e1482a55e8318ced2d4b081
GIT binary patch
literal 41356
zcmV)!K#;#sO9KQH000080O4-9Shv=A22Y;=0BTVI015yA0B&V;cW-iQE^u_bod;aa
z-y6r%&_E^1hzh9`X%BVJa|=<CWF<sXk#xVVvK66-C{fa|MfNCk&&l4KjQWx6J+kNj
z-cx;VxuNy?_j{q&uk-yp@AI7VdCv2_-Qn&kBP%7Ppdcmjr7EQ+6&k>gn8=Lp?(f6*
zF%BCa;^e3-<yGY?QrgdNVn|4UA3u0v=p>~9jDG+V;NL6QkFUg&>f*s`WZ>COC>Aim
zH^5)`Bjd7ff<wnnGBz=e2nZe<#OF!-@h38VLB^ve^MgYsDUD_Vd?yEo_>blXgie~s
zj1D(b;>j2oc*+ZPCJ9ISj1Az)mR+Lc6f)6=Kd9_~D)Hnz8wtml@EUu{2><m9o6M7U
zb>Z_A2D(bQy0|u)&r@`5B>G1wHb}(QgwJd0!E5$MK+_69<sf1FAlaZYeN&!Fki4rR
zpQkF)RTK#}FVj^k(`~`$we;Yr|54YhLRX`Tt{JbDP*;`DYc0}M6$!Q}(`{R(tI6lJ
z^We4rqpo>{u2vOYbDnmYt`4i)p-i`9nQkXOPuGLj`H#956}oy=bS-%LLS1=2&p@Or
zFB0rhrfXQHYsBYu_26~;qpoFzu5lGzOI~-Ot_q)LBGOe637VGanw9CA^LZ8?Jj*}o
zT2<&;RnfKLSqpU=^LaKR-Nqt8+cMoAWx95Jp1lW;^G99l3SCr1*O~`HT{%9FE7Fw{
z2|ASNI+p47<nwxY@OuAI*QP?(sfw-*ua8hyn$PPi(v=noI+y9Xl<B(idHp<i{r{+I
zTcPV#Mc0-$K&U&N&vO^)P8SIdEYtNU(;dX;4ffy-@ofJmU$D*{$_h)1jA6nXCRFz0
z^E^e$ej>@?Wy)S<$|Lx^ksiEJe^zEK=Ur9Vls8(aJciF3BT^nClJqH4_AOHu2KD#g
z1^ii=HC<p;Wi#Gbp|T;L7bH?P6iEh`DUT~t9?$26c<?58wib@&`}hh=qH%C&FhAHQ
zWVCN+g=lD1QFGozp{O#S7bX%_p2y?4Hk!v{$^<9zd3+DvWKWH11$~2kCh@|n%31It
z%H*b$%S{!@O)HZVF60ak-b_!eYUP;8p`pSAjGhqS6Uv)aRo0R>yG%B+TsBH1JEu%`
zE}s|e!J9`D`e)g2ABH!-s;U)lL78eyxoWIPbzzz6B0ew9gSXhTbG53WlP8S!_2K&k
z1xy;vXMFs|2l(@rl&P23`+gHc$Fk)<zAT_MZ>ca~BR+4LFkqK1Wq}g-yhIP)a!(y`
zpny=niK1O0xq3-eXJEryQ8t5><ugbY%^;;rcom<Q>cLxGIo;I=(^X8XDr?J2FO$tE
zm(3K(t|^mU%jd20;H~%6`>Sk~)!a};-pquzu}pqbx%_64d{&wK7CvvQ2QRyF-v1&m
zTGee;)lGTZ%hY$2tM3%4?<!N@&FAg$;O+J7_E+`lR(D@{l(MBXGvn<qTgseDOPR~(
z9q`~C^yK_meNtFRFrP2{ACrQ7gzZQ9Kcho@rUo#T8jV9$=VZ>y3zF}jFAOITO-jdA
zc|H%jDvP{N%0lP!d51lCN9rD$t!GE8hPL1x6NZ)&7Wd<2p;hPePL!R^C(A;g;`2^>
z@XpjG^w^Ne;wcCERfIlU5jv=1y64#Go-dp3g|e6z`MgUWyvv^4I>r=jeAZM~s?OMw
zcU8EwI(*(W8hjq_x@ZFxlm))Q=iT(+-SR~B3QSFRyJ}b~USV0-qVll+iNf9~3wxK(
zEB4^s^K`0L*ea$hsT$szcfTzBgYxhXMd2Tng@4TFJ@Mc@_3Ygs;YFr>RyDc}?|E7D
z7v<4kilV<Ni~gF=d*i`->)F3v(W^7{yQ-$P<&~D1`h6u+f8g^zdhkBgaeJFq+TNep
z?OkE1FRZ1$mRagsS<LTz-VYDn&pK{z)2iG1SCv$on=n$s-il_MX(arvEa8l_@LOTd
z85!Zf6oQcze#d9zgnxT5jfH>LanGAp+VhOOD0CGZQ$~TsQ554SmE$yFahg(`W)w$R
zgi|*!m{#KjMnwY4j8SE=nv1d2%CTCoSS=}*I>piuVb#qezhenK!nBg0GG|(|RBgmm
zZOf@NS*muFsy(IB5>eI7JEnj34x=rmE3=Qe1*20&*ND$_D5EPk5z~?KbfP@EBAy!U
z?63g8;J{$NNq;ymn9dRjvSjoqg2-Krenp0a24f6FITCq{>B45okWv{@s;(lco^_|H
zY&oWz1eg_L%&uH_@yeN$ube5%WJa0HDU*eW$)UbX)OeN>MAnQIOJpr3vMDFBWr=!F
zB0EZCFCucTFHsfmaU{rW7{ro+n2cLa=D?CUQnH?ute1$aZ$puZ?AKcY%$9Lt!TN~7
z`j&$^vtTY1%$0)m6M?zc7pyu5_W#|17AA}v>%ak39q3Mx22v!CdO5Jtk;V-AJx3O%
z%wVwthg5dpP}YIN*hG0!s^OH%t6mPQdgw7DB*4s=k?i7)5-*;2`QnXcnZ{5iAIjug
zF9%jR`urq_%o%@{C_qdUSWYySB?_WM!IWrRy&PEW={H`2%z_DF$tH-&Ld(e}vSeYD
zj7Q0sdO5JV18|ZAm?gt!!6u8r!pp%TSg<J+Y$^qtRxbzs?F^hQ;Xo^92J66?RUJ5s
zBF&~qk$-StV1N%_q#imca3V9ItN{G=PLYYKl%tB<CT0%nJLUPzTv2jFH&0A7o1S@;
zWj<wDP+yjcTSq3QGD~@n%fwc$0Snn0u!v=eqb!Rl%aYo&{Gm69uP!f3E9Yeyo0kNZ
zC6Tf$rz}Z-s7EzrsZx(tNNi<mW+l6olf_#(rF<)|VwqAY(`w3;_J?{@J0@C>(j|y&
zm<*OEQ%tm`oM<gew2l(3r$ih6P>*UyRHYtmlpwQZHnC)z#bjCKWLsFWt&}XAl5P7#
zJ*pd-s2*+q9n8{%*};PC6oc(52iwhp?V(_MDcHV0)T7#gRacMp|6Y$QO_?0F9_3c8
zM+Yd<L5g&!UcRf`5#&`@0|b?80A_0dVbhaOSq@W{BlYs#Z+*Z~i6t{*j<HL2T)bo_
z%9rdU%XEq|ou*7@>gBsCeZW}>B6H>(OLSgLbfKK+B1?3M5?!W5SL)@vYWskz5@Z(4
zHJ0qUn5>|j>;_AAlak${WVh?(yXyLYLJ2TSricamPYiaa9PBO&R!qU}QLvJF`R;Fh
zz<mkdSuqb--#x7AyGIo1F-3Y(o4x+8+u^4YX|iUXHDa6b=ao(Pf;HhwHdn7G*lP;*
zrcw2eu+##qiVxpPu-P#0*d;6#FX8+0CH%l*eWX~QDAs2YR<8!YqE`GO0cFd4Wud-_
zp}v<x{a~SfQm9`PN=kUCXTJu3s^Z2*(!ar3nK07Q6>u`rES#)#8JwJS1zclk5uCiV
z2u?u+=h}d9B10-lu$eMSEL#&XThns3W-OaBWmBPSsv@>Q4Zv2NC!0(E=1D6vMoqfH
zlP#pHc(Ns>Qm0fJ()ILY<>RoGM5fG{*3zse+elaPWLxP9PinI1YDdA^Q!p**dV2D=
zvr$`u&4STkSFwY56+4!%VkZ_$mtu9MSbEa+^kkLeQC|Yek}+VRx`?3+%b|=|sIC;M
z8-+5iuP3WLA-hY!SurLooT(VjtQ^jqg|nb=mK4sazMib^khGRyvu12qHd`@Uk8(CU
zmd&2BaVQ(AuP6U@PJ)CdZ5S@=Nr$SQbfi>0DOE4&n)aur|F~c7T`5iMeW8<drTfA@
z(iQiGec22-Qx+G>;#yx8_I|lvWtNIwxqs!7;Kr7O0W6C<Wf@3WJZjJKhvs2Wb$J<F
zIWI%lybNVohEWzz$}+shb+dYwDz(W=VmI3|BiP+MQoNf-mG5S6mT5F)8bg_Uq-)x@
z|BZ<jCg0zQtW6j{mdIaB6i`kS$P$gEL_w4&Sh}Wt``?JFRHbneWTwn`mMla}Hlds>
zlqH);$-*cZPr9akdrinhC5e#$Gh-&PV0<yy<Z`fZ7A%5-O`%{@rEA)^{|&6VdNl3#
zdSq?ROlRxSjH>l$CPkV>k!IJ+ca<B2$m(i9ROK2lhpho~+4MwHmU)zAe!YD6TTie+
zV#zF+7<S2G#Y?uZe90EEOmUQHF=bj(FW*&Z2;wD(ESaS&(K0bnLOD?)OSGI4B~hXk
z_3~Y{-M~r-GAkyTB});Lttux=Wyw}kvNTGTUN7HO*9v4vfLSw{EZ7<`*xGWibu8F=
z3buiQZLF8?{?-R<lJK1kvzhf>R#o3^p-5XPQg&_j`oHd%w@IYQmf0@N_U1b(n{X#<
z!d+~xc2lrD6l||_O?&e{f>rV1zTeqwOql)b66T1PFt>aO53pDVDb^v1l_y=(-uxd}
z)QSQLC{qTrP(%!sUk-Jcg*rl^j#8*&(lzbP{{dCSjmIV6%$O4_+(|LqsdBi}EZi9i
zcb3APldfrRUPCyMA<s*&nKKtywu@r6OXX~rS+*;b?J8xvCSB9s{2y%9dGh-2p0u%G
z3Rq9xsOrg^l<F3xx?NvSRz403B{F5n6p20gUu93;VLf@5O;<4myGOxF>g&nh&c^!^
zY*x$zb`>9rSMgE#Dn4eho=~i(6zf@iJz3>=d@ccH&AectUW%b!l|#K|q25raw-oAK
zeLY$230W!uXT!W_;Xa7rK9<9MV&OhhxGxm$YkfUg-68o+g3Xrs&a(Xwv;8b*`^B<J
z$%xn*$%xpbW$Nk4znzmZGQWA!)`XFjsqmznOchTyrd0BjN<pTk{i)eM?w1wC2;%!f
zC7DY1g-v8C?hBjBRAi_bWl^RqD)nVy@0V38vsCoT&1EW;1T~q8lF))>X-QesDT_w!
zS^m&Gw5l#Itt;oH4V#y?EQ=;(X-8Sw%ha@PR?kwUHfhQHwwrBD8Eu)0-K-<S?&c0M
zWxKf}%hZW7=~AZ7GBxen|HecMlb!^T8Kch<8HkCxloJ`UL`IaTD<$eCQ`5fvZ$wq9
zlCcDtIn$jbGZB-SmXn#WWagC2f|6Ov)U<D}37M!QSxJCdFxD)XjTp?f9IOWmW=Fy7
zDHunlrhWV0z^bc9NTMEDGQid&Zq<6^K#?3NQqOw%u5yFWtGXJ{yK)V1VrxJjHa&eQ
zi!)_$sh975>j_*XmduLj$1Yia@shcfFWCT=$(=F{q)Z<5@?DjNV2}inH8Yqc8X_hd
zT23^KCGw<1!zqzhy?j?~H!wni%!V1sl8q9Rd6$!oX354-G9OCjTQA>L*9!PafY~zs
zELeaTEU+AGEDIJy!GbB+xO(~SZ+*b{%Dxl5V%XHggb87dH=(NWLMhWk$`n?c&Hk?&
zW?tnym0_4Nj11eIPpWJ~K5N6tY_`HFSp+4UB2&}u{EuW+oH$hy&WxGHuHtm@D$Xch
z#hEPEEXp;Taz)D2v_1a^7j<KlB$YWchozb;riw17n#WSjr&J3lRg6qc`}2QLRdHmj
zB%KAbkfmEBri&}5Tg=iep>*+-ZmCR78}u5|iCno%63&uIVBr$QaLdc#l32JE6mBJj
zOO~l=hyD+^>YSM(=}aqT73<8@s?J<ZvC=43dVQT)`Bco1%#}5hDR$<X%FbNNI&&SH
zu=SK|10~y7UuXVyMsAXXvtc%~3z;Qe$Svgyxs~P0rd-=7*Y^54v&uQSLz2pt*~wDv
z5>xFir`p3(?WI)vDAoS@I<wldGDm{W)P%`p=?;kL4wlm$V(IcIoq*Ee`Z}|^^O8uy
znKJn-++i`?k#e}BEZi{)cbvkVsIN2sc4nTGaHgplbBcB5>8j2=L$S_MtaBnPBS|bD
zzQk|9#ST0#nG<v7LRq*lK69}=C(5o0^O;MctSE>Kc$v-16$)^b0$dXT7}p4(iqWo1
zGFUJLWem!E=0*j>JmzNky4+&p-=^^kY5XEl{I0c#PYv~-WOPgB4jcWhIC^n;^m}ad
z5*qzJjs8Fs-J%xJs~G2@B!w08h^2Terg&0L@sy=_Mk$_CiWedZ^Eyz74DwPE!kT%-
zLcA73yeWrx%R;=P5Tz92y$Hgt77*1L;e(_RY?zO%5k6Hl!e<)(3l0CZrbeiI6n&GN
zyDjsbHNua|M)=7Z;TJo9DOvj9nk;>AO}2(c`0d0Iy;D$C1~U^zR<<Jla<Xjx8_SmE
zUtYE%z5<P}NaHKX*3bx5PL?K;(M_4AZ1iU0=*s2MRoLjNG<tIyT}`%zMyU1#X(35r
z#<XN9)WsAU<rJ-0iq@2(4W(!+TSFsMcT#9dLYOn{Scvvw2(5AmZ5Bd@LUf=I9c621
zguk5#on(JEf|&)QD_dcN&azdEphv^&)9?neH8g@*B_ni^oVz7sC|h9!BiTwu=qg)b
zgl_EojVVBP3Sd%GBbZe+f~h2f6=TNc-&~x3i}L(ivhl5Gd}|utrlv+Pt84^Y$>`Qh
z4>r1;IJ$j#bPgLG(da;<b8Bh@v+9iCAW31vII<Ky#T32DDSERMPL!e#rRZBzBm7Pw
zGJ>-tge~L3Lb!?{`jtcUXCd4u!~hE6UQ;8O{ly3aC5&Kh!g#Po7*y2=gK797H2hFe
zcujHmz~E4ykkJ!-!oq|Tt=M$KB<5;v%6OK^tMZxQ<#Ux5ZG10w<BwqHJCX(*MFV;Z
z|DHUP87+#dD3lT|Dc?u*CU49bk)UX}kMJ7<PvOs`7+>MHGZ{b8V9|T9g$olAJTa8<
zrzxG}DSGd?inm;10_fO4;osvbe%?Z=+P7q4#?qldm4~WVKh)TS38sU`)nu?KGoB6(
zsmWk7W&#}?T9d)%%tSgktR{mk7#<zW)MT(FGl>r7*JQ92Gnoz!ugPF*CV~#0Qj@_p
z%v3sfT1^JqGSlhc8I=cX{Ap24n3;6+tQw6rWoFaSku@4^#zfK4b1IK+%=(F$OMe<&
z`KQgrZWF&n5i^etoX-xFp5&=4Q5Bd4bW}`bxi*ruqmpbay<lNwS?xbxQ1QZ9W)ZzC
zu5z5eygk%!Z#cm$rdKbie04R&zD93j#KcprrIod;>Zm33@-nKH&>(82Od?fVURlkw
z=4#^GUM7hOt*9(i&AC*Q_r_sXQmN!R&hWR#iJ26twyGi2gqtyys;#c9W?N@9kzg7X
zOs~AV>y|fDCWA_4)_<m=TPJ1>Ra@H-YQkx*qiXBxpSSX6n%O`FH&)JDmzoGxJY>Uc
zq5_+1?eJ<I^I@{6)|Oi5s@g|(nXObUyVkk-)2~XHZB%f30|^R~zk>?ytSo3>A3>4k
zE~>e^@`Bbmho;OPs<yX5rcCXyj|%Q@AVJ}LbEsf$gXFNn6bGp0!OA(*slh{;%prPR
zUS*b!HF*fsgb~myapfzkvEiRz&B_oekY8EAx;6s8owbLl+>w7QQ_bJff;mbBk2RE_
zDRZ0(p7=-IsnP3ynUhrO)IaJ@4PV30oTh4L>OR%q>&{s!c&>p2h5P$F6}<2d&(!KY
z3z&;k^U^=+P7Pm=&0MBxR~lr>wC-G`g4Y^IP&nV~RIs2ya#&G!Zcxpem2+6l12cbp
zxgK+iYTT|%A*k+1E~Iirb;(=xr}BSP@J?Ow_Q!+yF4Zh<I8EVN+@qQ$l{KNBnj+cz
zRQ5sTC9Y3CO__&O@KHm}otot_)qK)$n!<@crJBzgDxVdWc}`_tRL*COj$BjbCB5=h
zt&iM_7sN2Hslc0BJFWVcjxukl+`C#k?N6_YWlE{w`-T!UWj;{BkF`E>t9c0#^NDJG
zuJw^y?aKg}FI4So-KY9{MfpYrzc-MeaG(C5f<J3*slUAJk@-b6rQ~XR<W~EVE~b&3
zNKIO<A*M{Lhzu2!Z6HD6eC4QMW4VUNVMYCrr<w|KwLNkxJ>#EIq#8<cbtnYY9l1@Y
zT+_Pbt@<Ok85LBnOWyu?<f>3j)rQj)u0?aIsU}z3BjWF`h+<k$*_Lv(J#zo_q7Fu#
z3Tiae+^JbwQO(v3rzxCx8>-p1q4HT_8BHqNPOi2`t{KywUa2Kl+atH)Ikk*770{`*
z)2e^<3Dbefb*#11{`A5crV|y^Z74xgrZW}PldJ8KTg|hO8GWi{AXnQXx7t?{FkPsc
zVcn<tdqpv#f?XR(P`FRKQ9<KcTk0>boM5_BO_N$5xz#=^oH3<pW(_iBT1CvMphW`-
z3g>G{1+5w+hZXh1nrhnA`pB*HL_x-uYV@c}A*k-iwWD(Ob;(=xM=pm7qPpbmk4G+0
zO>V<!3fIDcYC6{Xi1_={shOTswpXo>+&?`Rl<7?cof>NH)GU3dX5WU>6i(ckYPvL3
zJ}WHaN@e?1&Sy2hyZ!x%B20gJnOkL=4mEgwi3u}+Ug%!=Lfx8NC`#)<daXz8uNBU7
z5WRMA<!kk8c&+I2A@uU0mD6Ncqsu1+h(0`UBEt-$3ZAvim?<-yUg%ZZS<{U;f?hka
z_SXuhJc?fHUE7RRY^>4r@-el|n5j5pK2*WCwiz>H{OE=LwVk!-X?RQky*9A+*9xaR
zmR=iF+l-Yz>52)amyfG$#>~VS8&4HNDg(3>D^z*f6*Ga34XreG@+4nlcUO4@`E5)5
z-Y`;9qCf6`{QS)3!v8|cKR?rVDnDSN=%X>69Gg}B08P>F#6ki@s`EJa+lZdCBKq@Z
zRb?qu<x@Gkn>ac4`0X<}Yx6a!(I(tx?=4aq??_&M=$7AgSj+q@eTDp&npXLqqub=`
zKK)9H7c3{MkL@C<ik`&q+z0&q1HruvkK<wOvIN@a^+@uJrJU%^2k=^@gJ`>VC-Ary
zg^Lw6;r6Po&|&m-lv45pg<p(;neW2^rg*_bn;<ZfUJeoCZ-K#jE;Q;k3yyT&2IN&7
zZ1lVY8@q0T8()KAowYYKxw#mQWX}hqZ#j_4OGWFnv|!xD^~fY{8)~d^1$j81709oX
zf%eg@z&T72^tH2zS!x<tcvA^F?usL-VOK~mdp+=Mugcx2BMXaSE|4y7K9dyg5b{Q?
zX?~%6J92#5Lc(u+ml${+CLR1v!OB7o`ZjwJzB9TLWIvvcJqp@#MJp6=xWfU_2cC%j
z{B4EWRb3%dC&$)*T%mqm()q38b@IbEmO$F;0%)Gp8sdf;lE*3EVS|4ot}U{ovG?aC
zC5e2d``6u}BBlLXY~l9M{M+`}uJ(p=RXPp-9AE(N?#qDoL3Q|vdqdlJ4eqz^x@g<Z
zT$E&J14(C|qNe92LZ)_G7@`>rpE3@i!Pn+Ox+4O8Jx#FR(iLXE-G?sfEdXx7Ni^2i
z4iZ;>AS${_z?^Y`mIfW+$(a$z@zf~P)V>g%8PziXx~UoV3v~h0?MLj+ZIFY0Lx+H+
zwIA&6G!EWQ_khTCO(5yBV*ca4Q^3loJ?ZE>5lnQQxl_h>;Cct$BEMojVkXEL+U{|M
z=mFYX<)9?|C1WL-Gu{w++-?f~&)aj;LT%w}=dKX<QZ_#!O&b*E%H<zzIShi{$#P2t
z10h`x!N-{o(LlSOSk}N5BF;{O<;PTkX}=Jjq&^{Ag6F}8RV~Pruy7*%pA!_wE<(X=
z?IE+3EUbF1gL;q9=RO*I6`xY-%MG59Lf&2b$#HS;fR5(>ktZjIbESfdNuu=+;@?FJ
zloTGI*^#pOTZ4gng}()6nO=hRT~EXHM;YjPm<EW}*<tdWHKMofkNfK$Q<bv%eVv61
zZB~am5YxO&Q51Vd!4O?hFonPFF;%G^zsDBtaH~4jpkhFH>+LnfJG&#`v)jpb<$j0k
zJ)}YDd<hw#)(0+0U&K>!q(D`*Z{B%E7uI?-gUzejph>GXlSkV5d2y#Bv09E4d{&Z%
z?<tdEwU;01xiAuKLGC1`YfGXNsErg#w4mjQxzOdF6p`)i09Qj^p`a%++|kFK1&WUC
zK=#!Y5~yYdSMOc4H``W7mJWW3zqG!G#x&tV@`Ih&bkJe}S`{UD84(B%FBRd!X@yAD
zt0%d5-UWvCy@kKbX-O_W6xN!fYwTCcD8v2?Crply#c>Cd$m|{|Xiw+wf{*-ukk_FR
zcS}=i?7T`kZ}*b{_{lyq(la?0+L;wYTRCso7q=R|8SjK(|31W4*MVG|ss#PNnZx@-
z!$|n0AUHVH58Y1N3A+l<!cvDf_;UD79DV8%dHZZBejL=349HC-TkIpq0B%dJiPlpx
z^u=Dm)P2YEVzjkMQn4pivL6C@QBvg5#n-s2M+Z=wFrR$bZ%A$qH-%ykE9j&b0=qMN
z5bOR=@uE2jkdk*43Qj1H#@{#Lb@5~2L7)YUy!zdK|FV|ESNlg^8;wpl`_(4!y_G~(
z>{djb@nu1;i<gnARSx!!-$CX+`GB^Bt{@*!JAhu(arYhaSY<y44E9<cTIsZtJpLYL
zZ?dTq^lzR;68pQ84WK|!#_hZ;>C5=x$6VBc%!0FdO*qAqcar4mgcR7GMcFg8VMND~
z#1B@GlV<aX_Hjqpmf8tYQ(C~{h!2NUq=ylo)E1C**#nf<Il!XRqe+hovcz@B9n|ws
z2>GbjldO|jhHiT`Bk$4_@kJ#d5pQl1$M1SX`^FyJbB!B`d#nMm_kVF(Jl&0gKIp@h
zfp(;A={Dq@xf%q$JP0mYhx{jplcNP$@Oso92+WrvK|fEzJ<U1f#J=^!vTX)fMx_Ya
zhPr{&jiGsm@1G$m8`PnxeZD|DcO?ogm@RPGwcgJC;8yr{X*`Nr%Ox&X1%eqbu3)`p
zYsjbf?Le{gCXqIbMIo0`;N%@=@^EoJGG?Zs1wC9zi%ACwZZjWZhQy-}-kY%RuBP^@
zHg^YI?_<Q^Bq7Y_;rQ~)CS*y{D>Bkz0&16i0~=e(!<eE)NU^^?-W>jbD8#peMvf0q
zhHMsDte=4H_Zfw7h&`lgWCB;d328pV6y#l20&lJj?2LE-i)Lit#T{*7uwUoASHDI<
zV5S@_FOG&gR}aHEc?Wzne>7MR4h4;4_C(ER9N9Lc0IQ^ZM-f<;tQ@<NJcxUWJv8SN
zPp>}EdF&vZ?AU~)$`ip`&+g=m#w~R1Wh%kKwb^zh8hKxSgyhr)k`?}Q$ZPYDXwWrT
zq*<y#dWUugf4dkMmnn_)@2!M0r`CeL?-3Ybcb#LwoeldtJ;NQbBGNCBL4L;l@Tf-C
zaMZdFzGl=1KUP*C?v7hHAMQ-X&)z56cfZ*WS{_P6Z)di|U50HZamRy+!!r}&u&O(h
zhP;J2=VH)q#UMy%<4xx5y3J8<pAH#oUlM83oGjV6j!bN~kmRIn!l9by;Om4>c>|Yy
z#JWfuTP%+!aT?>u#2h|}JklTE%J;yV>{Us2vO2VyYYg(UpV@nO&Vv!9>SRk_UG$+(
zZ$zSw+52Q~!tLB1lBeDKp>>LGu<7*x;D5h_o{u!*b}k+WQMdM^{Jp)A>T7*?FgXGn
z?`VsDE<2CEt2@HaE5c3etc*?_=|dtyWZ=hRDRODw1hU#Q2_LfEinkdWgP;c@Z%oy(
zOA8;|eaw8+*taWj>Us{th;7*adq&=k8Ar()$Bxj%I1WM@FXXhn{RGZGIhj}J(G`AV
zwu9$Ia?t!tJaOx`#csURc4BqV8H$HSlhXCCu|=X9q^GOHwTCxhl}0c&<?$fHTOJe+
zMv~?ac99!O?&wT!7g9PWjck561N@er5rhmd6?iRPO*EZP!3e$-aW;R5!t74s)0_4p
z=S)vXJeq*szq^8GDDFq<<L83lY-89RejkmU`w1Smm%^{MMnc;4O!ASRDd<)(j+nmp
zfPJrCWBu_<kmB$$0uBE~u<^(f^sLt&eC=~E;k?<4GbhB7^%rN5wEt$Hq8JWZ`RgdW
zc|RHZowkRLS4#0^JdkAV7z4Shov_8-B-s6O8Kf^hfX4a173>)|AI2CxhQ*KbNvp3X
zafEX_^61@ef~Nn#BW?BZc8zVw?$kBBQt1O)<Gv3m8X9uljUEZ?_11Atc}>F`RF-fa
zX(pq^k&TIwc{VXhUx0!~s*{z(9WvsFlV9Pk@LJuH=nS>B`+oNXdgs*<8zD8Kb#)rv
zGCK)2hhK#4#ve%Q$Rs>(@&XwC(u<h+bcd$@eF7AgV&A&20XQb4kt^f&Ah~lzc&S+t
z%6@1L*Zdm6up86S=WnMuTG<=%$#3%TdA|*e8+R`cJg*b!l5FH6<3euwYoR8Oa<G@o
zQk-Ml0p=xy!_~1shE98$x9{i?+-PqJzBK1KD*dVg`|ma;dh50ljj4x`S!+E~G=I4u
zV^}nbTetya<)rd`nskNlX3e2><H0a%kUZovgmeGp7zpv@lCSpOBx2G#4s-Z`{fz6f
z`J%pKZ0N?jQss@u-}(;6-}??TCr4C|dmqu)XoXd%Z=*Y?xO4&RNjU&YS@R&a?Ev_3
zA`xsf0c5r>26Iu)3<rZ+(H`g|8wqWv4S>w-09g666m>0hLCt;n(3P0Oz1JIIWxvhP
zs?Ra>u`~f399qJ+9$2Wu0gL(BVAReQ_G*4b#$TqustJ)GXP88LzK%Kq&q(16BmdEN
zNzC88{_|hFQJJ{47=yll&?6(-oyUp;pJO+ZC_HB4C&AUQK=gXfQi1-a=l09oRZxKZ
z5Ki91IY|3ZF~{lcQY_uB9iZZeSbDBCvD?y;JQ%7=`h?u(Y)nkXhHhLGVslf_?yUw<
zJ~auii%}-8nrJ|9_8$B0PIGaa&fe%-zh*=<|4&CJ$Jc587WJQh$tzXv^QF6poKc3L
z)!B6P%6$f5^8SOIW*s?!DobI)3q9DRFc!z!xr605H)Nop4F`I6f;Xd@<}LlSkeueb
z0%>iIKTRlr``#l_&&F|ZGByCLk64l?N4}D*Mc?qAR`O)K&n7r^LWY#OFF>Eo$8+#8
zBMu1rUeWx$yKY!sr}<mffBxZC54fSZ3Ebo<N4c}Ik8-~bnC76f%gN!`8byajuCWfw
zjz8v(`sBq;&NXo8yKoNo&eVxqQ19+wxnVMQz^efc&qoGu10S5`Zj0K&9hBY3q0MO}
zhZEXu9Bwu<a@hH1Ki4>a6?dxoZCG%1IXC5im&3b*XSm<(f*mqBnq0@6@ATew*tzjL
z>NJ0=`p>^X?tiT7e*bekWC!UqXa@=pzK1^LUl8Q)G9#TjM<V0n&KM$llk<_eB(p;x
z+3M;8t+Fzamb@CaooGNTH_iZy-RofI=oGkkc`9-69*MgfDZs@E*Wlg*71C+8IxLy9
z6=t}m;XA$J&<FW=QdpRcbEe4?pL9)9RMH<Q=IEfHQ@!AA>O>OeJcN)J!k#@k3N&4=
zU?=6)NHd`axirBIayqXPEZ-!H9JIUU_dShJ&o^MNeRd5DxwwxLI#reHe>N57j97r@
zHd7%Nk9xtB?f0=o-f|LOdIQF&cf(sp9wj$zxv1%wv#8nSPq@p<vG8j57ktF}2TsVE
zM^YcHgn_Teb4-JG;y_LsO4L{m@e!-QwY3B0`L%?p!)B9y57wimW948^kY4_CGd?!5
zmIwb9K_qcaJl1NK2%3FEiOcE3B)U-y+531U={qx(v<z&94(Xed=yg{)Q-U|bks~|N
z{GZWy$fczy|7s@Z`uA+4uRH-f#;!&u7Of;XS7X6I;~nQj|E8qnY*jMQt^>O4X@CM8
z;_-nS2jFw-E~tlEFH#bA2K~3UJ6!tm>rlUqDX{WQGER&djs|ZTM}B-h45t}Aj&grr
z((AxJ;^v{ly=&@AI@(SlvbTz{QjsrCid#rrCVaJ*PAbJV^SY8yMSZ02D-hgVwF0vH
zT5;4a&f!Q|$V05%NC;dh4cfn~;n9cy<b5U|J+@Rst!~ATRMidmNm>)M5p^M+zuu#w
z9?#KPUkB)Q$bkg+GJ@l{Ird93M}7{0XeI#Bdg_Xo&Ql}%el@~x+vyYDj}OSzbO2u3
z%ZeDCh!%J)YX{Z}2cZ9OX;|?jG;gTR7kFUDg}xE3$z<t$!hOQUKBof-?Ae8a6jKBq
zv*n2YbW3<0{|klOaTP4nS_x~%HG$C)iKJ(O8bkyxAj?#9Ib(15lJ@+T#6{K~e79zR
z(bHrUTab%xx-5a`8@rQbGm1#oh344OKp9!Q9CK*eyixGTeF?fW%K^F+H%7xo$6&uB
zy}@%@H~1NL0PkPd3`~VRT8|qZMAJ7CIq$cGB_L0}yIsXSr`tlGSGmX~R|f5|(1o@-
zdpVAi6X1>Adi0-9cR{G%Mr^+~5s!+Q3B@ha$!-7LcwT5KN}t#RmMr@L+w+srgA;v8
zM!Ft+J@^>y@7sf{dtr@DUM(OOrln%9jXH3qSuRQEKO$U#58RkjguP5Rqol3J@x%2=
zM19m!?C4;Mi}#!a%Z01pmDF{t-Qy%Ky{sX~yzvZH?NCKaN52qsEAmDbeGj05fj%(h
z<V*CeNS2sOH6q<?&SBGjTaoKa9kNuX3F;bXM4q`M!-!FPNa4w^93!8{oHXOnc(Ans
zz9h4pH2b;?&pL7u+G#C=11(%hul8Gs{tr*w{=5Z=vbjT2Kc+)y4+YZaXC~517v}Bk
z8e|%}6U&`RK&O9jpiNE+F}53v6%Vw7BGX*h>EwiG?(Rr3uN}lmneB=5FeT*G>mbhT
z{0a{+l7e@m<sc>XJqr1ykA8;EAzum?W6h46uyN@rXj61k(D~FXn6P>c&Q6~N0ha)t
zY3Y$0=Z1irPxt({cXyDr-^UBiSZk3R{yfr^e;r-EB10k%EP~#VMfg#e9_U-{Aofvv
zu&sq7xthQb!yP+O?^6Id((^b43ajy|<?*PX|0UdiPa^tyY(D82E{m>SDgynhtMG#E
zox$*iJ<${_1LLI@&@CsO^F4Jn`f^>5tiQe$N~9N{fs>Mvx_$zQ&blG6)4wV($a8=-
zGW%eh%T3(=@nnc@Y9$C8Jy#%oJ03f4KZa(1(ZWq~qhLh8OL$`XMe-wauRvLPE(s`{
zkL2#%z*Ch9(S;8kaprPq;-oc-^tzr2zF80Rq8hJ(vtL^Y%u4o?d28aJB;|~tL(w8Q
zyT=-D$qGXAzsGQ_qvF9WX&jz@Z?&LTd^_@PLl2yJB_6FF!$A%*Bhbw&U5VwUc<gws
z7+>132<|C&gd4aQw(2k*R^AUM?~HQt9*+w`1Dj4krM)rf=FdYhQ!WU2ow5YnJE!n-
zrEd145}K14w~phY#=FQ{w~^$f={WG#T8p1d+Cp|N9Yj2@PXj^Ehgkh>Fo~Yn9$B5y
zCW(2yA@$v2cy!`5+O}62Qnn^Q&a_>`Hb9Q_`Sl2g6f70o4ZVN|oM{8o@Ad<!qJ?Ol
z{588UehgXHK@IW_cLSrzP2kwqDcIRkhR8m(K`KgCV7I3!ydR;4^P*mIPBi<)8NF>f
z>E`TC?q75y*}>m9dLz;~Bkm0%&cBX9m%|Jwo$Z?6L7jl(^zLY{?P{d*Spf=1ECRX3
zk+37;V4g~cx%lfP7v%A31v+=u6@RjsfQ!1BkZU`Hb!2o?f!)pn_>_q|QFnfTY(3i%
zOh#Y}*N&*Ab^+c#+?cfP6U}MQa}m5DZQ=Eje8Kzv10bb9gS^V^OP&Rp3OYy47R<=$
zO0=VcK~HH8wmTyQ@K&FA#)snTVO?QqTSdsf><r^_Z{RjcJ2>u|0;I7&4T>HQB)9Ib
zhdZS|@$rDQD0$`~<TKZh9F1rNQ&$I)k%5!ZxaP5><fRPgqhB0R-;tL(u6v#Ce{AaC
z|4em$hC<Z{T$!*Lvd+50dB=`m(ZLqdzlMV0B_0U7vcLW@zLbu}&GLeG2|Hj>!6fkX
zY7a+SA3`l#ECs)yE9hU>i{?Kvh`g#(|7TnO`G0@MA+r1gPQgbtVlZSiDI8=;j-79V
z|GTG3dbln?k{ngP3fr7VZZ?zQh*wV>*!TfjcYP-I;qDS_uyW$uxV-D1*VFr7l@*)*
zJ%1Icp1;lC)X5R*Ieo7w$J#*A4*UEmMI*u5@HO<zm=C_6li^UR5<Ck(2)9FTkoQw^
zKtn?||IUAJ;he23*Zz@yepDxQ?qa7-`NOPF0&}Q0XgnSPni(DPM<mB0s|#;oa@;U(
z$ZYxi3+_K5LO+IV51R}7@^WDBie`d`rB7h%{JGe5J-~R6Q;-+;8g8_HMtXH>mLEHR
z9tvx-m+V%_5|oU{N7r8d5ZHU(%2QjHNaVitA>)k(;ui`V2=48PminK?Av!I=dxjj@
zdLR;3%pC(3jEnF*&qcE$?+6Al?xe)nkUQ#}Cu#ILnOw;T#u~@(;oJ~?Zi|o%tkwMq
zx-}^aPagILOiGWFX^%DYPcB%89`v||U!9%_g}pw*p3;0cHuxky$r;YQW7;SGzC{vR
z7m|Rxmjd~f@)~5XZh+A@lCZuFCNe>C#Hyq_Y4m+q{><cj&WM!$q;>L7_*uLMK2FTW
z4<BtrM*n?)E}IYH>uVPyk|0C(e`Z2avgo6$BmeO`pWpYt=$Cu--T!Bu4xtyDoJrzD
zdx6e^M&NSLfH=k*a<{pk!jX^KlhPxW9QkSb#Qa4Xv6vQ03Jss2H4~l4>3MevvE7Oi
zm*sGLk6(vAUi;Bo&xgdjeLQ%#P3Aa^-b8pe_mZR_4S0WGJ||CF0}pMH$JuzM1-E_4
z4AjfE7~lBXf!_Zv*fzGPF8&u8puY3(%pdE3Rc1LjBm_Hboch?Ii@cgcK+ApHaicI-
zcf&b{882FK7nwS9)8jiiObRUJ`n(S0`l@Pj$N$&L;TzZ4VZCA$cbNNThsTXha%H;j
z<@%kEa_~;S#U0B$<mSsW4p#XoaLIW**J#xM?(!|mxt$W%IlLHq4qoq8<9_|QlHUL7
zv@RV}r};O)<6jMN|7&Vq_xs;TA4AC)Ir&3brs^c>Qi#B8;c`x+WqmkvO(sI@$z^2Z
zgI_2rc_l7h8jM;`?FW9v&hYD7IN}AL7X)c4L;nvyacNvQ5||D|<`YV>d6!M(P<M<Q
zFWn^Q5*<#MJK6X!-=3rAwi$vaT!G1Bo}ru5jtMs1Y6|9u?O?zCW;Cx~EKXVU3x}3S
z!8x_D@Ly&lvW7W9ilWl-_!0{!XlzABsl;=JOqGWqLwhptLk!AISdDBF7Q&>pMlf96
zjq{;>226~yhSljE^1qnNknrw9p=*cMpyaBG_6%J^Tuh&#m*;cwnz_qKg7ZP7*8K=t
z6cK?PW-lahZxGqxdzxbqtRpZUHXZfPZHbigi*c(te)uryO8RyGgnc_iLc-+^@b2vK
zLwCNWB7K1jiS4eB7x{M+jJVXB6c?Z8tZiotE<+2@fgoM-So^XdI6jt)dy+|hZIdD!
z>|#muPfr|kv@wag(ioU`o$%&<?&L*>P{GHKw)kfA#;6%rj`Wj#1>MiJ#%CsUgznz^
z;lO!Ua5R_?cg+aFsfuLOi!MY(Wj%VQrA6+k=+OGL;ryFpqK}%+uWkKmfXDAP{<Man
zjSOMG;yJW!YXl5!rUr9@Q$hE*G#GyRhVEs|gTmCdQ1o&bJf0GZmIcoTe@|=5y=kDA
zs*RA$<{$fA>342RC&$L++yk`F*}uLaM~aOG;Lq)y$;;>Q_+UyrUb$}wDqWn8lDuQE
zxAr?+Fiso3C1{cH1tWmBKLLy4r`T-lSEu+^;`meavg}{n|A97Ga>(u3;b5fq6Q8->
zUXcAJ`_RuBj%3U*2jc(Z0=~D|2cM9MCmQFXVOMc$8edWMmY`1Yt;O+QSS{lw*xU#0
z;#J%+UVQGJ)pKBW#D4A@RT+nIm*TkI-QU7QRX48tz*Mf@PhakC%@z*s7iAqp@mDz-
zmewi0jW~XT)Gyn**DsxB^3eY0P89p#6G~A`gAF?efZ^FwFeOrjEZ7~71{7$LKr1z2
z{*zJR+u5AGTgM8PS}!6oz2`%!<~H*A)IiS8n^O62+>F6wX;aY1wL>=tIdJZ&^3lWn
zF>u_x6velV#GJJ`c*!gevf9s(_&;t2cQ$C^ftTlCu5&v2gf9!!mTkgQI?sdiSEEoj
zt=afzv+FpnpAMO)dsi@Q@p4YYeZm=l=R!bHJn)RB5dDN$j%h>z-WqWm9g!J?=9}f<
zx9e}ByFb^$Ve4g_9fdP-qo>o+%sY<ot79}d>Sv5?Pd-QYZ%6~&zl?VzC6lmpf0%v$
z5XPNvVy^3E?BII>wjpzTd*KPe<d?p9vg0*;Vp1A0h+hlJ4lVF@o5m#fQxrMq(1a|E
z%|In9y4VdqWdUJ+17NQn5cjW7P+#R(Fe&H@jpw=u5}ZHaQ5H+Ed4MAf?Yb3p+0h)j
z?T;j1#_T4;L&kvaj8o{mduO8a?iQ!(=1oxWUrS=9(wWTqutMOK`35~0Zb$svuf$Ch
zw&FF$Tj168qd4W06#AU?khm4}fwiLt)B7(ogWMtiu3wV<w^;-9-%eRBAa^Q4pr!1G
zu67-W)_qVT{yI}3b#5mTKYtCV^y)!YKD<pdcbpSs<R2gcD|=XG`U!%<V$q0LbutKT
zC1zoLAbR=<V$x|fnRRF^cwh5C+oyIQ>ej;{$Lf~A)D>{MQ6F*9y@$YSwI6=HDnQvi
zHxg#EAwCfkhs=UFWc7o?Sc#hoAHMg1$xH;KUDG2M){H~W8SRPo<JGiJ`*wQ5yUoH(
zZ2pJ8+Wwv=VV|n;heQQ>2ZF}~O&Dkv1P9l2f_Gh<K*MDeq`%@r#5xNYuk{W^-xt6~
zmj&=)oi42MRDzRR7t;ED;CfhkUFwIKsknX!RA-^E=F)^2dj?B|b|Cwj8A3<*j^Gos
z66G~pBhZa&O&(==;<VdE=q@J*CwJ&dx?Z|X>!$`DHNH;q&BXB=NAw}hpY*WTd9(x_
zI%&&Uqy7;aURaDL`efj&*mHSuFRtKG8ZVIUp)TazLv@&xE{EI{wCMa}6^5;@Q+#uA
z{Gmxs4k`)e+!+(SxX+5LxHrG0at~TIcW9TS>ELDE&SAe(7w-3ElN=_DlXe(y@`>xY
z=010-{e0TDemnI!zE1Hi8n%9$S=PONM`%10<oYL*h>=Z5yx%<BYU+7T$qrrM9}eT3
zuziJkuI)km7rQ~>m8lSy?oZSgoJXhAE<;SQGyal37f-O-jXFgyhFfkU$?JiMq+`h?
ze08rcG1)7Jl(R>mZaw-zR}(8_5M>BCdkb*0W=Du1aTxJ*T8lpwzDHdLb%7hVrOB{U
z3COfZDYkh0fundd1;ThEp?~{(I4f&D+WV<7`Oq{SZ!<_HGP+T?zs@wUnQ|PtoWI3s
z+1wASx$(d;<P|bJ_8c8Fj3mTlA~96ABQx)IgrVBwU_jF`#C)b64BsaUdTucSeQQLF
zFX$2J9vUd;@ody3U;x>b!{E*)3Am|iG@9_<0lx0c!;kmf#a=GMN!TTM%-x;<W~<k7
z9!z$^267WYFQqY*j8DOMj~i~CHit~vSt!`qV;0#KqK`hyErrI$P04L!j3Yk=5Vc>)
z=-k+ty!-R_pvztD$uAoY3Ab=1`6+MmI#2RN%MC)|veh-rIWz#q&fFs4#A7&DvYZn+
zWhI_17)riuv;eXu4ITUM1KwojK;l|jkgsd?4q0`+Onorzy~@10^q*!8@cYA`<L$t9
zbU%2P?+G{ND}ZsVGgz7W!!r4`u-h>O{m|5b4{4DQ(PAWgKNto};zomCpD#$neJ*TZ
z|2p}nQva21)|Bj>@Ek`z{DlL&cjF|bt7zuAOw{agH{3BonrNxMz@DM4V2h4E%q+-6
z%XjoAx!aU!Jy&Rv>|Ce#mHIC|uV^xUS_x^rYbt!ya)L&~xx`{n0x8xr1~O<koUxcg
z4ExL^OVjR>&vR8c?;i`u!BLy({I6xc_p4KUTXFn#oz1!TpTs&Gix1^$%uC~5E_HN>
z{vqSA%}dvz^PLIY!LC7Ex6ik^wsz_cm!IzDZnnzdnw%L=>sN<m(K7#zuOijtxA?;H
zDPOT|{`L6g-yZjDXIH9Fm-(BUG;IAeugCLu182mcM$S&iTe!~4mt3?<Ipz#rg?lhJ
zVkQ|$5@BTzML2WxoB&BHaBgg#P9FZ0gVWECqCHOrLd+;t+JCKU?0AkVB;)>L9e(e>
zEE=Hyl8jTcFA=x%hErGTvVSbA?4SRCe5?QK{*{bxUCsPOdck))2K;;fsz}-WrY9^%
z)^*(nV7>r6Lu}xvgC^)@6{3p;F|hW+9XPmR5Zu0R2+!`=g6{IAFn#%Q;5sgc-Luz#
z)AZGJ{_~aY7u03`ww2~zC~reF0#@Vgw~~6?GoMSk?Ct}Fvh&c$oc1Iw-;cEVPZ50y
zjo^eS9Y&k2Rnc4iGBhu;nD(DHzD=&d;|I!8O{>NicApAtx(2L!j!pLLBkPVW5v1fa
z%3nL~ES`Ven?zd-f&5(;@UlC72(*YGZnNhR*~!Irc=mC~kZn)qmB!#HC+^TZ3{ASS
zfh$#O|B7x4>h0g17oG^Rk8;6aj!oZ;M>5DLXeV}wQYSH!TcfTgcL`*c{Sruf|G>|r
z7NUKsgV5*nMfPtsdG!3*^lSC0I;@}Q_JLi$^TTtXWUuf(U9qEklV#CxXzd!*S{}ik
zsZ-#BJRk0yT+QinJ{^rqnZr3}dXB?c8zubz0$TS}IzAj!m-uFt;&Y|N{@K=4r?3wG
z5#0{d**|+XEyt;EGO^!+cR0LY7FeI{!VT)u3|4&dfR@?&VNd#Iawq%^-m+>weCjTR
zV;)75qY;V7YeW=DiijaIoAIF6qqjISGZFH_FQD7+29c|4=i472o+R+!S%lR__>g1m
zy5V5e@gz|BDS4{Dg~;W92mSPBU_3OIkh^OTv3xA}()%E8(z6S`ncE9y6k3s>zAxbY
zvtzJw%x03j<Ai+%(-Pc~7Dv&$F=P)L$jP(}hGZiSM4lc<jxL|c`Qf%*;CQtaSy8--
zkURpb@)F5_xBi^r7oKC~+eW~dv>&G3)gpJ(79w?>4qQK(b-=X(e9SK#n`~%;)~<3C
zINECC7hEZUlU6a&J$C?Q42#CkI$XeU-xFb?vN4(2dJGx)BAGD1ZXs@`CIl;eLMeg}
zI56x8l4+?<7MJ!SE!;6&T%}E9KKVla_SR53_9d1&=Z<CbUn0I%9zI|ih_8D}!w9$W
zgsU+ND_vPYl$ynpV@EV#u!|ij=<qf#+<Y~5$QleKDI>@rmFvi4(gN~q&J7gz_7vuA
zmL@L_+yFGnlNj$a#r9S`$&Y>+P&jxGI;+FS3-kTZoqjQJFKs(+yqgEB4d%dsX}2Nw
z>MLYBE|#1*Hw~`~(;-=p-N3~$L-6%LZ?sD{5zqYP2D|t1NRdhgv5s6w#uy#J&s!fw
zcDxj{1r-S%Z(N0S##)lzGPjXo_B+n8-SXIb!c5}0wUywqbxR!6JsT^|OeOpKheO1t
zDR_#`JaVtwesX{JY%*iec9iP(7CG$i2K#px;3N0Ar0teT<h|N@((y_oa0s|g6hrf2
zgI5CDuHJ^^B#nmM4$mO`h$m>PC<yvp>H*QpNob<{K|$%tX*e@%B}{BL2j5?6Mmk?=
zLW=#mfX}pj5aT}^G)4?3bEizlFmX6KowJhsJk$lIXD5=O%bwy-HgRxvcs5zsY&RC1
zABlBe+(C*9Ly5QRdydo3jW`HRBWKgDpwi?t&WKG*3Gb>3GH%uuHCfwMu<G`Fd>UDz
z2<9p0;I2OCyxek<^f`yz)!PD==10jACu{P&=qWn%UFf5?USz=33FO(<%P?Cv8x1)e
zON>^u#UUHhI2`{X-0k{e68(NRcD>UY#*f+xN2kb=-QVUx0>_Pn{|q6IR<1$Zf``O@
zjR1OVy9RU4K1Gfn1L4S{g)lm)BigyOgmc5`3`%LnL;W>X@yDSP@Zsd!NIz;JTqZhD
zlDGrujYuOylr9VI9U|!KU^xJEK#RY$Y4kjphLefPkOg?K1Au1kOtNC6Hw=AmPPFg#
zh6CoiAp86XociG&y5Dp$94#zHB+eNhKC+x7hit}2wDQP|EIGKJ>qt6Xy8sXKM&kEc
zD&%X^>1g87M^GYn0{2+B1THplg~5*sQ1I=&`1$HQ?4(vIn3?$<w{DyYS({eiH(zJq
zAcMQmM>a`t<xMvjHN}x*>{o!z2R=mVhnJA%ONNp=%LwPqu>{CnJ0EP1W#O;IVPIiB
z5atDY!^^?4I54q*qj$Lv1i73grWS|E20=@>>k>qUe$<Ae&y+c7?Jo&(oA!s`^JgG@
z_XqnOYptPYgduT%nuS8Mym6X^4A{-eKt(FDq@De4=#-_xeQTZt3em~<WS5C>Y0Xz8
zeR>(O9Wn-n&fZGy{Wq03Kh3u<by^6MUJir;^9($UrzN=GqXT}0w-LRPcWAleChT~X
z11tE6d1-@V;p)_jcvWCD6m;E#jCvdp9D0xnla8LpeO3ubSEZv!+2;g4oxPni?IGcO
zFG<H<r*m=9g>K~7>>pUIcr9Ak?kpMSdV|wzv+kje3%F3y#z&wqG@BT9x-FQUoD6IK
zBXIpp6TJA#Nsh^_L9i!Rg*1LS7q!)zM7VOh;J9u}lHbFes9NtuTC+C`8CK)(gVn+3
zvMwHXvpXJQGaq1%9I@b7qs(vX@xYG#@P?`R&@pZq>Ncw>iCb%rqV~s<qr+3l&ygch
zzWFv%C_9!k-E@oCg#*$x_9SK}+9H!XnZzbA3tzgg26t>b3!cr2$NUxVF<LtTc_peK
z%Tb|l#Uhdf>Ziin@lhl<SOG88Tto~KG;nW~Wsn`RoD7UONpus{$r+1Gbf@KP82Zuy
z`nfkIy$j+vxjj{2O}j{F6O7UE!UMQ*h&2qGe33kuVGR~B{^aA)e%R!2Z$V3w5WuM~
zFkf1VtdjA<^Pcy_ik+{cgA1CI#y*#Er^jLB$kJh;bSas0EJF>Itk6LH_BWxgU+{T5
zFh@%C$r$pr>wg;J{$)Y*AnucihkCd;ufRiBXp@|`y6E1~CgdXI2{NpQz)LTJqn?k(
zS9c8&<nE6kTC)ek4CRGrr&(`u(^;Fu+kX_Sf91%z6S<k3Sauy<|9KSKuAKnATkFHY
z*NdR@ZX1%Y`yKMn+{)>-a1$Q0cN{So6@^AQ_JUR!Y1r$+3DW;kB%DkqIPvm1G(d47
z*?wX;Shlbw+s#Vwt_@RA-nXe_+Rzasq4^E$a%dquD$pb|wtXTWl@)*)(;c74xJ({A
zNE2ND@<Fh#NR9+4KNc7n>_yg|`*6?uN+e%OlZ5r%1Sh6M3vTk2^7k8zCq0%kaNzT4
z&St+TywzThjETA>ICi`bZhk5lq?atkZ-1U5$4%Rzgq*%)I?m!mbvQ`&ClJC*{EAv#
z97ATdTrDW>n~HpHAkyw-b1;AR1uL(RBG0B+LMP>cWP#5;l;3PVDO67d=~H3k;rjRJ
z!u91OrM(PjT~fpa{^!tX-DI@y$W%O8FBR^6c!bx_$w4=dSztGZ(V)<4JXpn9;E}m&
z!F9$^a*T6NKxD!>?&@EOSz;vC-`R-!vBw=8J~b7FSoK3piz#`F)E;wYc`QLK+YBH}
zwl5+-$DYQ0Rb0?UvuB)arxL+F?qob#p_^c#^*Z9$#E@jVgo4tr^PIxAtB7CQB(!l_
zJep(V111Rp1e6f0h!>D`8{VRrt$VP7^?sbQ-U<%%89~-R-w17AU*a&iGTfhy<T!CY
zirkl(!wKir3!MDU6y^2OfOmaw3RdbUkr|)Qp)r|L(48Z8BosJ!p8aypD7!Vt^s*kr
zK8Zkit=x#ht0ma?z5zrpZpz&e@(VBQV1=(7+bt-~xQC{ymk910bA<_OTN00uk(jf3
znZQrx9h$WAJ=*Xg6W>ucz<t!^N#~Nsh)BC}!V@jX;hoK4oO2>PpV<%DO=%2Qg>}2J
zksn_A<_PRKqYHtnc9Z2Kgbd1P4^1qZz!1a5DE7rM&>ue*j`?Px!9GJcs+^uA>Be5p
zGp#1zH~B91@;-%{Je4A8{Wl@UEpxE^wGzRN;)TR8qZ@oOibT0*N8wfDH{jWw_u~hL
z*As#2F0=&iA(M|S#*c@`pk#+o`0n$BG+n+1)LPFar%$_)bH|lo#tJWpy%mp=hF?T0
zZ;i)J@s|&MEaa2P@1^0(kyvuo+LX*_QAk__Ey$~n9!UGVAvWH5i<8jyAr`ouBd7q#
z#}4h_ZL28Q`*AMxJ)KHqx}DC;9^Mbu`31r>scU(KB?Hir%VR;w@VmXkXC0I+*AaXT
z-=YppN1;5E96%w7Xhu*VDwyB~vy;QnNU63Y@!1;GGG-Z!9<N92*4z^mNc+O`7hGa6
zTL}d%H6-J|bR$Hm6^XIk$8ql03&KC1BsmH#NJ<+mn4S|Zuw83Q+Ge@{U)qZ3ZI;V#
z&eP<)S&@T(WsN3DgAO4%pBXUcRy^tV1i}5NOz7s~h_@9qfp;^bq3_m2Vk|SAEE>3q
zM6I(!{pEL&xN~wG8^;GI*<dkw)Fu}9yBPwZ*XMw$yeqssnhZSc8-hId)ueBUEp9zQ
zz|rMskr-bqG8gM23%fUXS4vCLB6r6BnETQ|s@|@B5}L>yQA&eUBt$s-Ui%o6G$;+y
zph-z3L(yO!GS5RYg)$3~p|kI0NF^GT<`SZLPHFu2^nQBZ=fm%Pd;0J{U(T1a?|rUy
z-+S%ry4KqFX3p=KNkrIjM9|=m4|W2$D(-_7yk^MGY(&ZO5ZW1#0DVr{aOwC9q5{Ua
z|6>B?yeWs8>NuPz%Y(8b0Vs9p85!v$3$o><7#4hibPv~p&rJcSB1mGw4~il)FA5jF
zo<hiV1&nK%&m~>x#?>-EK}tECizsSg`sPUDj1!Bgyj~o+Z|OtdPd`gP<{YNp26C+E
zL`TTbHAL6*0dRAA4Y7B%LKBTn^gf^qyIg!>qKz~Rb}whVWFOJV0Yl(bP$vW>{93n*
z*FcEPemV+7z*nx7yCgb_+t_9T{?)d)A-ab-EzqO}Jb(JK(3`ey84E*ho}tV8R^j8$
z;TU-c$y%)~6sNY5j_2kCOIMKVnFHKepDRRXWfd;DF&tHAzb7wd_tV)u?nL6$7l`jM
zqy6?@>5!|UaNv$I%oAD)HJ>X;vu!vm+-Z!)!@rPe6H2%X+Aqiu!;7S#`&X{_GQtSm
zFw%`D8UDZ&I^|Rbwkrw2^%Jq2S)4Q+ZnY-+Cmw`gX;E~TmrSQTU54Z18d0^S3Mb{<
zCHuC&A=Z0_W5MWPMzC8N3?0L;=}|Clkm&{{_$awKRF0IjY{vjIbNbAr4!fq+kl|x*
zP{XclI)2<y+!?Hf)d$<?#kx#t82X6ZlirWdJi54w)B)gq&n1knDOfhugK6zvoK=-W
z=?+QKJ64=sEPNO~cgW!Q`Oaj0kO8u9kCSa}pNYc5P|yh4h>JiIg%w9b@QZa+Mnw@G
zNZkX2MlmQz(Sp@y6=>tWao`m{6UwZIu#rc0a1UjMq1KLG?(M~`OsRqzzKm#xZ`amC
zMTs^H$%?15Uu{9f_K&oSSTa9Mx8Q_*lQ6S<A*zU`!m*PVG4swR?m(MAK2}a6X8os$
z>wG!E-hCUO-*z#0PYHreOJ2|kooC?LK6Nrr`~}s^8`<!z)E3irJ7E9LKs?(&6BC2|
zIC&}f?Z3P|<RtXZ{lovxKj!=e{-Gnk4$k`I!Lb#IaA2?=ZWsxJ<jn+#R>%drqq9LS
zzYMgmxkGGGHaN>?gSWsATo)~XC8rm{58oXSEmuP77sZ2&!x<2gOol68V<E0F2u8YW
zffUvhv=gq8`#yd!+MyH<mE}NBsuqa9P=E>VgMjz7<hPGgqg<-*C-i^rpZ!<)=Rd~3
zUD1PzEl}eL>TfZUf(3Z)!(GnK;SN=}ah(e?91VHPD@e_$UV8STKgm-rBt{RrnCPV}
ziF#E+8Uu^CFW2PAnfcXBzVmus@8(OSUZRec50o;8*4mMZv?qe<f|sPW#f!}Q?nua?
z2%1_b!d!WGoRgX*f^r_g+`sU%xy{{wg5TgT;1?HH#%BlSu@$w7Y{8c)WNdLK*a#Kj
z#>Yb&HVrul-bwl(?K+fo9P|fYWd$f|KMd#WC$c4G0VKaB_}9I;o;!PXB0O2}24ANs
zG-wr!<K}D~-q4qI2RF@=VxD#l#|<-HlGd1ZZt=a#;1;w8Vs2f9pte|U&Ygna>wlrr
zOpiaoKld--Z@qDv|9<s+zH-+A{>?rW{wSj^{;eQUzVOYdd>aQNeqG00zQfz;e2w{%
z{GOAPpY%eEpL?{Q4P$}tZl}%{tDC{U=J@Mdrh|PZ{)jzY5W!#Pxr{H{x|!d(X9Pb@
zHJabD_5y35euKTvljlE_8o?jD5Y2bh*5=EPoAA4jYF<&M*Pq}w{0sR1@AW51A>M!g
zAG#JMbN~2noIm**M!iph^(`6ja{fd(6KDjDnUmrES~-Zmvy6nvv~b_Kv$Sr1KUMuO
z9-j>NMLWezbTeth*<0TIu7kc+dqG&C|FeIs`#1dSDa{t(E>!@pnF4TmIUk;n4g`+}
zelY0B!Wj>B*gSq76a;0%%t`wpN9$Mq!&DIYk_Q@BBEfZ90PtVe1D7Plet8}aDfa^)
zz^eev#!rJI(p$km_86!Nc|fRJ3P@H*LgI}$_~4oXQ@aO<<Lfr?&anM`{lqS7y7dR^
zNB3_$f1o4lNSnOho(_9!&pbJz#C`aYNrSRO$cCR&$<`CEsp4i)YP%|n1n!Zf4|LX%
zo0p%F#B4Rebo(MMFED_p`&wYbuA97LKCPT%NE^*^O{Z+GgmK9|3CfJGr>ZMtabsK@
zx8BhlCAKP2O_N(>bM+E1i=D^J$SeKr!$+pAc={*!|3?2;=C+l*gj4G<=hGGH6C$7=
z=FFyhp3i{M>s`3VM#%XciK8x|KbdZKKkWJzPPr3R%w@$kT0OgvNgTK&P<r^98uiYB
zXR(1qYuOv(QRj|s(@M#ef=zh&<RWI$k!X53EtWa3+MKg3xk<Vkv|)`}Ipfr+!N`n-
z-|N4_XF<`Q;Me^d^S{%TjnO^9iYM5!uMOX_cP*RP14}u!u<{UFr5Mll-b-L@T&vi0
zm0b2qXgJ&QB8qj9_hX&EzlE6o<19J1m%UuRnGF&a=dV*tV$bgR&dPm@Wb5B6@cCOb
z+2bb`u`*(7*(2prtXTYU_I`yH-@HVP|M{jJn<zN^yAOb#=dSz-{=ebh{NMBEztwMa
z=llot8}Ygr>gK!}1&$7|RrLYAp??xTt+1t@QG6JLcmFCl!1A^jnzbql2Zkm=hH*XR
zTFoIway#r9eF;TlL!j!#1FHF83AvWvNKcuDVDBLnZcgz9><|~BTMW;V>m3B%);Azm
z$cM7?e!ukOY?z`EfI}qj;i0}IsPxbrT#cnbC%P7EURr>E>1ceT;XuzlNJ80+CfH}<
zjy8s|D00VxWd7j6**b4p-Zh!Lm-E1x%@g5e76Hc<HK&*RkA=O<s>nLWLaxGL96s)Q
zOB)V}gTE%ldz>Upd)!E`I0b>EXb9-*=EGI-Z@h7L%IV5?;Y9w$Oc<Lvk39G+f(^O)
zRKMjdupuq<=f^O3-_}N+*iFNRqCgmIeTg%f>*UStHrgI|nVIw~la^?VMfdDY^yYyp
zjE~WDVE2bpPdOoOKW7TYzj%(%5O>^mGy#h5xsxB~CgYG{FR<X#axkwyN#yrGVw_%z
zq2DJxS~NxiFRxz-UQ<pG>%uei^^Jq{)$3wvQ8I!(sUnR&r>0@R$02lc@gwr++ISe&
zw1J*hl;S3+#-OuT2OZ<&L5k*@agndLL+qPS@=2)*ySD29D-}i7ua2d*{Av)r?1lRZ
zM*(*&n$F8?;_?gUVy{{ZV|8By^#f$E$>23jytERgKHCb<mIT6!0x>wAevvuhe+CwP
zRfI*y&eBs4wRnTqk5HM71=z6a7**h1fY4q_TukDyT1yI3zxR+Uqh?aMn5nd)HVL91
zkAaPs$IzY0MKpcmSZE%mL3*zopl#o0;nN%O^tMwaJ+n#}e8U~!kZ&NFtq@EjUl_yl
zl^G<#=sEUE4Pfh@qr|7Sh1#avqPDY>DCO%B{T((?K3|Q#@*738pNlu>FYO|mEgsTU
zhZbV!<a+$5_KcL@3`39ii6Aeg2~TDO;k<x}m^5V)_ei&gxw}oA32GA%IhSJ0Si*xL
zYld;-3nb|h4~|)GxeToy=fm**vRF6M69OtlV98EpC=VQ@<MKbERp?j7bgB$9sQHCx
z2V5o79tXe|Yh&CX`JUOMI-9)8ZbjQFeKK5T9PmzThqk2_kQ`xwGZ$Bp2#dR<xyT9>
z^^BnA+-D-yox|CwNC1}F!l3;H=9IHC^K?cq&(`G@$xh5C0dgAn=IV9weytAdb&H{{
zdt>0btRpVgYJr7cBcMleI9ro!j_=ONpoVlSxmG^|u8PcuGd*jGiI5Q#OJ-t;o*Bfc
z?k3vu1E8wM!PZ@ruKFy6+8M*~Lv1#wsuR4oMGQ@v{74-64C<32z=2l{FB>CCK#?+d
zen|mUk46l<`;ji&lS;$~MA4Z`hV%MLFlBB49N91*6(*;X?O)!2rDzX5>$CzjxI8Ac
z<}1DF`4U?_Z_(?M(rEoD35Yh?3aS=Dc(0&}>rV+pX%BtyU6cvmKSZPQuj@9jF@=mQ
zuHiEJZ1FZPg&K7vGmFS{kjj+C>CN&WGPQ#zA~po1zV*{qhiG_Vx*kfTL{N~i3YA|h
zz@WHRZu+Vo?z`9~nzUdxXx(29bBGhL!-8n|hht=FT`UB}tMj7Drvp#T0DtZj(2w*6
zeZE<Ou8-G+YzHX}f1gM9%NF4G(G6gyv6)<Ow!xyi*38!?c|5wuf?igc560&PF{wZc
zghMA{lXDho_Qs>(`amjjV+u@|-$5ByThP3+03*c>L4Nlt^h__nrify??wf%2a0x_N
zM40Sp7Ld(Hevr~PM=|tGC_ETfNStn(VeZtQ<ZN#&Q89i-Z8uCNyAL%nhJEj;dE`9E
zx_=B?7xq!HPe$;q#+eCrB=C4oKealghWqNA>4NcXWUhA+hR@~0#6>%a%<XV$`?!QF
zx)P2Fx2<91?DJ$>QYF|eXr>$VJh{16a@6tIPt<?%mI!ZM!40*!&Uid@1#z)zI5K84
zgecm;6SL_wSJ#nxt*~aqp8eu%#+8t2H4pjn@kC*Z1-UbE6;ZwzhG(m<(aX<v!Toc0
zc$42$GvC}tL3I5hQZ{~~AlftsbKEw-ac>#&;k*Y_>+dA0GXl|Tqa+E6ctOW3aE3zx
zHQW%93&09rBdW)EcyrWOPAYvYsLOgXW2#2Naf5L3^WqvDQaGAqDoleDOGaYsvKPcc
z^E}$-e&PnQN8;&$#k6XdJy~RE$Cw&;f!FEb#KiX$k?YsvrY?^s9UE-Pk6QtPheLy~
zLT@;_1dpd1sXGSmXrlbN6QJ^?5z2H<M4$2ckiKj-EPG*qk-Sy3Z%QoLa-xE~Yx^ni
zS~`n9dlLy;bx#uOj6Qm_N1ai<nLs{R*5kV9H4x!8qyfK*Lwe9V()!&2Hdd#Q(t!<V
z7|4+yS4}V@GZ<gK^nonRHIP7VGa;5b%(+vmL2}0odM#rOc_u7ntY`a!7?@g+UANQ7
zW*7x-mesU<ARo5s&4!hN%cP<HFe<cqgG2sO9JjL`GUm0Dxa&!z%XKX|A|~3PJ7hn2
zM9ziLv)_{tb3f)-ml?<`(}MT0g>=6EVQLWjUwe*PEA5u}1OG_(Z}=y#r4Pai!$N2f
z3I%tWN^n|j2EG-)xW|=JP&$zUk+s{PvMCU<PaFcj**37PZyk)Bz5=w@<w5zYw~&80
z9GvyjKu^vT%yRQV(If@Vzx4n&jXD_GU=B@s5s<hv5^fDI0XQE3z0<#tZR?N2Nv%3K
zt5^0Lf7v2Mn?K=yf5SiN*|e9chv(7N_Cd@Y;|YYO&Y&+o<uetT;mppwi}X#9CsA(e
zA-g(~NT_ExS+waYV_|%ZdKr9UcK2n|N&YXHg_Xle<ux;MT}J`F5;OW~u?oI_Xh!89
z_L6s7Hqo%oM0&9!oaQdG=Xjx=%qC8R2COfq8wYRv_Ro%K2o3oY{C~qgnLg(sv9(C2
zyxIg%GtDJ~vJ-J(ZY&obx`3>B!^0z|CXuLFF`&3*1DYkuBBwEqyf;gRA@)z`RFi4&
zG3*VuD{mxP>m7z=^XGu2(Na$Pagab|rWnpK-3En<W>nA+2;(xVsKZPVQn9yzc;y@<
zMJ-pDX{lp>&;OmFdy@YI|KIRW_D6)V*XG}5Yj~}!8~2TkTvyEQtF2|XL|tWXE=ypE
zZw>4ArIR&}e#u@eeaptaZ(@BL#rY?UGFihoNq%cS!@m}t$gZd|XD?mpV4V-wuyK>(
z*gK|+*abeh?6yr=EPvJ=cH+A*cK%9TcFT1U_BqvHqu*>~H>D~6_K#k;CPw`;eo3LB
z{~iA?`Sa_Sy2$jO%EpOO?YKbiDAP78fb_M5liSO)etVN;fx!<|ejTmy53yDMA7g7<
zm`wWL#P%=fC%xJKfquGjWfW2AI7N;n*umF)f0*{c018&TrmwVras9{bgzshrH|%BU
zu9^pQIBUm!zQUvSrR5kn-+=@yp9E~oe!AQ(n5T3r8$G=(Xheb$My?YfU6$o=XPZ8p
zZgq$0&ATY?bs1GV9R<6areV9$Mk4nmlS-&0Q5_v=V%u0q0@76k2}T04a_A54X<rT9
z>?zBvqO+mNGY-Q(CSdWm)y#wyW%OR}US?)Z6LEN+LQ90RP}s7abD8}K&v|~L-)G(>
zt-&vOQ@#zu>f2&)X;LjUENLJ%q5V96+p#dQ*$GzJ$Z~V`7^3OzSBzz5G&S5+NAe^t
z3FbR`!b5F79O+m~k3`oH)oC)gQZ^Y!&e#r9YVGlTgEO^#yNvkkjDieJeQMs(#Tcz!
zj*(|4;aKZ+n7w`y6C!4bCb?atV}B#}smh41RODm%LJin&tp!b$3fScEf!<fngU#11
zK#MdoW8)ZjyL33-JSw0SM(c@Cgd+0OcY$I3Fx(ih6tbV4q2qjC(A~ABbhJn%q}c~y
z)EP5!_T*w5HAO(??yCTo!5Gk+X+b77%EPC(&*}U9Pf(|7Cvf*<KqoI3!#Ca{JNXZY
z`npO`S6W2th#We5bdg1O{kf%)`;G6t3n3~)#j(`G439~L!{#%lxG-=B@>la=ykRlQ
z3X72w86wc0F9BtF1V#x@rag&a#I!FM?unnKmp^tvN`5Xxzf&Nx-wN=j&MrZV$2c?(
z359&|bM(o+^E73oGKyWWWD4r@@rbu3$&ENcE)2L}#iT2kqnaS7t<s?p*u#ll(#A&7
zJg%#F0+AbUjOqMqkaO}TG1m>GwbdsnXBLBJA8aN2WZv`Em?aPo^Js2#X(lx;8ADGw
zu45*C*MiMMp3<g}B|O_TDp<By8c(e`fF&h!a7?5zIEn6HR+yx7%)~$p<-Gxc;%7m?
zxDs;GXAv=UxCT!aE`tR|Vqoym9Q|*vA$JbX#sya@$kpxhL3Nvn;A>_ABUIN$1;aOh
zv_=a#su+v`n|9F4iH3My`6P~BEQyP%UD3_m02MY@kyo#0z_!{nx;H`_78|nE<HKCz
zk4eMv=LrMsaokF_^5oDd?<duKG@rYhr3WI9e`%S8PpHk1VNBB6T`-|wJDyxJ7thL{
zfGf>$2)Cn1$lXKKdhrRmviJ;f)IP{;HCu(}q<_%mAAouX_rUZOqR`{z4eQ*8V$zE;
zaz9BPZ=D|k3Uyj&Jeg39oFF>gJ{ROi93y+D^njGyA&fnp1=q8V)54EsoYK{s#Jx@*
z);sMcFMG7%@GK4J(o_Sjo31o_k~KbwJdR!?Vqxd!DB=~703Hs9p>p&_D%{El(uPb!
zFXISsS$~a?+dVW5t4ZaDbZVg~O!r+~3PT@9agQe@lC1>}P_QeR*pFNXH5c1q#zrkL
zYpeoW)B&ITCVF+80emqqqaIIR5K;LvjAGMt^jrOa&Wf(!)hAB@nWwV=$4`Vk$L>*?
z0U5}<nof)M9>THT{9u{W5u9chOG6|IXv~EZT#m>gh}>=n<$fjP`S!WQ<zf@3F@HSG
z(b)$b=On4ZzEI+I`Z66F|CN;KdJ)|_Q)uMp8T8VdQd<6YfCP=chMm32xPHb&+89`Q
zdXHl{nS1gbQ4LU~TK8&5L6;S}JU1e;&P%u>TdfFhd>p+lyazPw)X1|H9LV*%fbVx*
zsJ$Fd3Ktw9_B&I_tiUwl=39@W4n>hM$yXQ~iDHcCTt>_D_T%aOkvM_G!F8b&dVQfO
zXXe&R3x4Sfv0ofIpmnmq;n_~ST>1fZ6XQ{L*-RArkVlsvoQx{FJ#m|G4tIl#5fqs?
zgWfK26rCYM3!kjPF3ka2;w6Q~WA?))1qC`R+<-=pT}?M-{-8p|8<}>G9B>#^h0eu$
zVZ?_>6jQ0C6+Z%TT3#_87p^A#yabqTYKyy$eI&7SdAM`V4p=Jukv4CRfK;}FB-Rdv
zmN))bFwPtwe%*w3M=7C+OeRxXqfM@}?eMasmOOn}O6?W(aen?&Vjg-3EEBx(Ku$Cb
zexl3RN)=EQw|RKVDj!F_xlR_Zm&bSZW{_eN3Xd;0<I1{72qZPg7<3b6W*FVJQ2|%)
zf6wX1ekE~LGHmRc6l{3@iAZtfWbpJ!($HCq?Gbj!&~0?n>P$NAR5;<+2Y~O)6f_a_
zz%dRHxIX<TFKmYrsg&*j|G5|8>CzMMbVD*pOj5xIG0tfHUWM*jEQd|Lu_V97ljL>@
zWBIyl-2HqfRXNs3-Z?&?T6_1xFlQF_PkX}CkrhX)?oHHg;u@&dt;ROzaC*%12i5%i
zlq4J;1xkyAP})6*z7Y<iZ%I6<mAg;2r>4X8g^wWE^)}gF^o^KyjzC4#FwTt&Ap57?
z;~YYQK+@5Sp4S)+8w|uz$VUz|w=PCwv5&N>@rrTWja{IAaShmv`9OE4>BDxJT;@V_
zKcqEULfa52Du3h|Bh@)b1UF9+=ha2nKlKiwvX0!1?kD7N^={B?E&t7*oi4;o{{#J`
z_c!VX|MZ+r&%0tMt_cLIy2IfA96+;Z6|Bm=OjavJ!{FMIe|*j-);9|x2J_*G<6iLU
zodT1Z2z-f9hbZw6FmU0&_AlN%MkM$T__N;Ms2}|MdH(xnWu^o3c5fXS+q;HriC)e<
z&5I-n+ZK|`^X=d<H~$~^=jtE7mRMGYoIX?9MWZsWQ_Yk1bcS0y)prpU7}M<E^Y8ie
zOy!^8{~Ps#h@+e6-0NkuF~t!-#f*k5Wi80-^9Kjl_so`sugEJ;BgpG=#bfT_RHn3z
zjH(|EB(j*->8t=9Q3JF-_zN{$7C|3JC)4-lvQRiw4$DeRm<OvgXzZvAS}1Xn$kS{l
z^}9T7RvL<W!z6LYq{G}%!&#WH_1SNZFRHV1^Pk}V8~rEh(|aIUMU4MpT0ZNZ8qIn)
z#jtrs?^*BSWcHlgWtbS9#J*Mb;b*T=XG8MG^S#zwW@k)1%}&3f#J}Mk!=LwjB)`9E
zEIX?06?<OHn%(1Oz&?sz$G%dx;{TXZ!S1cG;#Ye$vZ<Eh{H$?N?B?^k`Ct0W*uBj*
z?2U~2-}R&YQt_|<1pnXgPyUAfG58PkPeRrT2+6ktBh7g*YOD`zm+d7E=i15gXAW?+
zJOjoLaRKovWeBOQfzo3spv}gB^@Bv%S9Tcs=y7;(G7Xk*ssO#xsW7YdJ<K~-NW`_|
zVEeZKa8F4D%fMrhy?ZwdTapIp8NbGUCk8gX3!qI&o7~Jl^ZWXf(0!EiC-MIq{zdI!
zH#CUpA}g9j@k4JVUT|N+JG?`bF`BADe=gAE@~gIx<f;(n=m7=TFlHS+buydS8GCcB
zviB)Hewq8gD>pXoMOwtO;PzY<h1;XOh=0EW+PApV=B6>E-R(ZlhZa($;d7w-V=yV0
zA_UbMT8zZwUZNJA_8UKY-pb`q@c#|}!p0|ynNo6t9Pbc;vMrT@<FTj61&{gU=L~6X
zq<RML^S~}r<m*A~(q9>q!cux|!X3gaT1lF%aycypAyV&oh@0wJ&SX^UlTAaLnN7Be
z<o%Z*+P|lg1|FJEyz2_c)1f?~b0V9w`M81%LI$xjjN(c&PB5Y$(tqP$yI}nOf5!iB
z{>7Ys`4^|K7L=d4F%QBX(H&`iBr3ERMLHT`)9V^&%ij;(ce<eSixS&xX^bC(5-@c6
zELzU3!^VT2zt5c&Y&q&*udd|%L+r)>$GNpFOho<_dwn-!=<`%#UBcFau>Ep)^+thR
zbWH@-@4E_%THDD6trn8#^NGkkoD4suhO&9DzmlC>Hp8_mV_BbVL*a<dLc-0cgQdF(
zoU6S_rWDIT%$xwIll5Vh%tJ}ey<<?vq{74-O60nW6+CeXBu~W}$Q6AXVliO#JAeIb
zmH(6ehdF<v|Kl&@uQ@vZLH;^)*$?FQq=2(`Fq~Mg4Z<2xzqqVB;HW&P+v@~JLX)AX
zdn<gO76WB>l3`R~5m|j80dki#!FIou@MNF^QZ%cfufGXOO-!IQkU;;@beQII8KQ3c
zK>y-UnD(t28Xrdhr#=%}wLM_`(Huyx{7hCo{xAPgE9OzdpTz%f@Tam5>zMNU(X?#d
zP%JtshJq9YGAwN>&ar0c$mwcecl11U_@RJ5&*YG&T0hCf({c1_#yNWBQWKL04{3za
z1~O${3g@qp$-O_5LhFmqkpib?E;eu{T_*3&Gn#ja^NxQ{ixyv{k2)%8+t6-$cR?i)
zwNioU!iP`)3;+2s3IB`zU;iEd-{4PmuavM#T^4JUYe0Sii}*2?tf(mj*DWh>ZD2AT
zD|!+Y!e+sw$h(;IG>dzkD}kHiH-Ssx1dNOqLf^Di*tn;I=oUocn*%;@D@zIe{YJu&
z%n!Iebt-yG=P?QPQMAfs4+=gBLGQ&F>QtFaDjyR3_HE#I|Bcc&lL>!<|8MZ8Z<bHk
z5%YJmf?E!3=x%v_>v$>lvej1@@urWp7m;PtN*4384qRl{?FnW_4J5MPNAF@6SZ!dV
zzv;3I;wo8B)f#rVTsQj=yx0sLkCjh|<G(&Oo>i9P^B)g}u}Kpd)^TV9>oR3JyMgze
z9n=kEReH5phc0_|;SS5+`A4_=V&R|Q|Nn$P&G}FGllB5(8gRJ|pEOT^EtU?1el3Bn
zg$H5O&ztDq@|wu1X~WKJeQG+rk7RXOVbw_^`ruYR-8CkUTE=fA^X7Qdm)}!iJ`us}
z`vKfe?M1}Kv>Y|6J?QzQ60G+Pz$=m&u&rtZZ2ZCizQ50j{744adJRM279LP7=hQk*
zL!!Jk9u)dcy0@v)GmCW~=Z-%$O<RmVH<l8M2}7vK<}`?mJ3&K=B;mY<EL}G^y5Z_~
zU9hg!gYEwFVQc4by77uKOj5o=&iNSPj-wo2aw!D+TqPK@pa^B`vO!EH5bwxUL)k5V
zFwm>PhpV>0v#f25(|0HCh3hP4$^&Gi$Hr3CvSnc3WQY2<@*u_C9{MF+iRczX9NUr%
zX}gDm+?^oOFS&(m`xpTEvrcndLp%NacntE#77>fBZ(yyV6?rx63+WFIBdHPTRJHLl
z`YkUe<6jq&q9wv$zpay~@ij?`P$PKED<y?FB4BpsqM&ohJCfJeiY+74VgKq7^vOL6
z1L4QOJ!Cb>J_N}0jevV^dZ^o+K=8ZGqW{t;(q3E$PgCaO^7oz4wzHB64svIz2aA|c
z=MwUH!F$H^j4E^|^H403Pt9iDputD;VE-p^+A>2Aex&6wPbao<2SU;~ZJ~vNo)N>L
z!PAqTYyL*<nJgGJdKf+%r3g_K@~As`6_i$GlU?U}Xqo&}cz@tI?i*JJKRSNWgfl{1
zN0J%TnYj}8PD`{Fl7maPj)LFlqcpKv8^&vO<N8zg1r}eAgSVG2_r~QdT_6-o;)Xkd
zoc9q>46KLGCb~#=8^aqRNxW7epj!`CkfMkZPC>k#(SCe|+_&9`BR17T-l9V=;>1$S
z^|FPN5{Gc&y(6&B`8yq5;znjvQ2g4llz!g(iY^$}N_#w1!E?k5Zsqx0T;F4l%adc6
z@|`^6;U)Scv-LPwm-JGrQZ>9fw3z#{fxx!A0pRUZM^-Sa>Bx{0T9jqW7!?RWUbG0!
z9>&tt0wow){gXRhr%t78d1U>Haiq$#21YG8L9#+SV0xz|*~kO{)mMkLfu30ZxPeL!
zIb!_j!YMq(Fy!^jO6<;GK-P+zlJ18K@i{XZ9`Yul%<i#}do}}OPlu8@p_|c_T?)Q;
zPr_!UE~;870jgh$K?1~a?e#IV=j3)=w*4tBE1Qc48>Z5o^SbcfS8H6oG#(#GctQEx
zrC9KNG_JlLNw=SNhm;FSpdTZ`9Q3~lnTu!P!ORe5>Z@bOX-~#aX(qHeYy#9&sgN#9
z9Xy|J&fQBZq$ef~!;y{lpl<dVyk}~_J6A8#cUv7iy80N;+X5P)bDwLUGl|+fcY(vY
zH6$lz3tYX)kvm_Xkkuy-f#l*1=<B=#kDXP+=fmF#YScwx=HpGU+xZG!%bJKdqLL09
z`<%C{-G=mi6lw^udry5Vg;=jy$_+oY2B_F;e^|LllzdIrr|vTHWLxDRso%U4MIUK0
z`f+)@ouX+pcvdD_9=4>sMOO5J%v98RF$R_E^ib1%I6KkK9T;{zZOeR1LljPNIYab7
zFc3x`<@!PNfsy!Fdl=HAjc{>}B-Qz8L~c~LB4J+AJ2l~Ox4fBdl5i)AFawKThEX+<
zqxk(oB+>5VK=q^>cK8nirEl9{$-*vTku{zenMa~p<q+zN1*o$j2SUGf@&@Ll;#<kH
z5PxYqeQ+G0yup&0vN92OZ4k!lQ=)j-T^gN4JL%Gd;V^hQjuV}EfVYt63ty*gg`5S$
zfVG+_*cZJPD|eb;*IZFp7W0xz6AvZ43t`yT6$oE9+~#V3<>fC&Ptr5@ieOxDB5dq7
zg5x6!(8)NP^!@lOI1rEkaVAaFz&eznzPqsfa~3ohZ>1{Tl5A?AG>YHXg4;WfLt4=m
z?2O+H{-vMEHL-Z0rVq%+1YIH>`;t3->p0F{w-h6Gxqx~1Fo^f6MZCL}TV0Vz8n=w6
zK~X`_rIkYE0&<A?zEj|$Sq4>h^<>Q2$*@OB7QL@opr-T+xF@^=w520yTF2Oirr;L1
z(`tx0i3Tvae+mQx;?BE6v8wzUeWyDCn;w-g50u8?%wbtnzP*HMd->Lz^WTxs8!u_b
z;bM5Sc@Ot2dXVPxY?#IR%VFw^AZ*s|#4EZf%&9$VP;%2FJiKl=oGFTCIMd7Im1+gd
z?%791KbS|9W;tNXf<SB)&LwSMuF$;kl>!^@NZ4^~Dz)O0i0b_=dV7RCDf*BPukDQm
z_bRJE_Qx|SZJiCD=bR&_w3G2;bviMQR|QeCQOrvDWB36BaQ8|=+lEz<Jdq9%TYTI|
z*!eO~tSpDitcxU)lQ!eMTbpogz-%V}{z{0<xkV;yKa63!EO3LPA3Z#G`sq&D3pg+{
zi*~HC1ph4=^lX6>2{s=fl4eds)Mp0tpHXfQm<M4+=4xWAEDEM~cfsm|%0PTn;X(6z
z+}D^%W|phK-EbeU%3Y6s=hjn8{$BX}ycniw$6)rfyF^|goT$#(2jx2jcysn8va=j`
z(Ni>jYe@fjlb8O1e>~@J_!nOWqQL3m8W=X_5JbH@49jzNK;*Lb<W9~8a4ZdnbDn&t
zUUCF%zXXB9S%26R`5DZtkAR1Q9tcI81{c+5Fd-oXq->MnY3Uet&5~WPYc>zQADawr
zY6GAWkpp8G4JhzSfaU=;NN)=TnTJJi>Wd3R6lDL_Kl7!`cK-?g`y2j6`x!etVq`>}
zW=7N8878<T?ImX;JBfDxqGH`;cSu*}eR?&wnfV-jn94uRC-+u<p_X}mRDZDvS>YT?
zZk={vx<7Uax<*}Pa(YgY3k!3&zO&D%^+^r1NK_*W4U|y*b3WB<Udh;;HYL0$Ldo0P
z#*I4^shh$5-}8U2S@+^U<Nr7RLjPa>MQ^|symBZ)5L+~kec4|Pq+tw5I_&+eD^k91
zD^mP*q~brsb;kb~SKGow{eK%*%c#}FnQR4Wo=5gxm<wIK8C30HFfI6Ei4AFiC=%+2
zBqtcJiA2(HAaK#m1i7kN*!bfo)l)qJDSM;I%rW7lVQ(UQka$UY!o^T>#yY%pdnk^#
zP=eHCfAqhbhj*(1OzKPs72Zyk8l0wL?X`GtiScio&yME|`-AnP|2O=L_s`z3XI|;@
z$4%4b&wM=8#HUV@-(K0t2ECljAJx~$E(zMl_iRt(`|V6*+41!(PezMBOgw;Z7JZ4`
zlsJYT=n~2<$~WNW{L)Y(@~rsX`x;nhjgu@U2Cz=Lr`V!6d%pW;AHIBvBA>Y}!(V&K
zhwuOK5_{CBm_H+5?05gp<2AJ4Pw@YZ`qN*~Px}AK{!?zfE_n4&Iq8F6_kzq26iJ(k
zszbBzpx-vKIz=5*?rT%~*QY>4Jcj;UH3^Rhy&^(RqwrPK6tFVfPDY$jWbV&ei|YsT
zn44QeaSc;VbT1{5?$kI~|FeZVZ<0XvRxN@}U82PPz;d$dR5n+ps0ChEVo>D30*oAN
zK}Gp&V1_5dX>m*3{cR+S|0*Ci4t}HwcCPsKh%>wq|3MBY&cQh&l*sFrYI0`<4~53K
z0+-H%170s^QC$P!i|AorEMiY*C@8+mB3gyFNkr^kG~d-h;@I~XZgqs}c8?}Sc1dK1
zZ4J&ADJ7HQL&<{;qXo6hShBBq1TMEu#|47>_%r1Ny>dL2+FqmNqEj~vADT^7s2K$L
zHA9$%IvQ<p!^`8}ke^;ZI4@o;+8oWI7D^ZC(jq^+`ZgI?d(R-trG;r7)xy-(x~Ooy
zmloZeghltZlBHgo$<2aBBoRK~zQ_jC&jMC?JZ7AGVn~gqGwjqqOKtA@p=q(Sz+@>)
z+a|~pc_#&UmAZ|9?Op2pKmxOJccGuRIzE!N0*OhYOoLJ#MvC~O`X2Fy$r57N(zYL0
zbTpAA7eu)r{YdV*HsbhGo!p(;RETPk`=vL0sb}R6n4RR0vj?XDZ~O^ZxUGiHygtax
zjf&>HjGSok4<+ooIu5+I1=1h3{aF8U51}tYKq0f7IWC+9@mpr0liYI>KV$`1^^SnN
zGEJDRYCwW@ACnV(`{3-)96a6lmEMz$#Ef=3Iz93`S}jc_arY$fVyf;h{+dqqdtb$I
z60fMmhyXASwIS~uj6ijy6RsZlk{k%ygJZ)OE>153Gf&z>$+i%ztTLrDz8hmkcQ<u#
z+08Wc#ZzqsT?lhE!j&PSuzAZY5RdT1<X$8XFNK5A+gsouluB-o97Xk0L~xjM2Ze3o
z;3YMbDydMqE<zEEk}l9MiQjQ@Whqzi@CGT(p>(X>NAjX&7&Lw!Pxaih$c-Oc=-~b<
zBwZnd1k?sYyqFRmVbh7y@>AqucplL?a~&SfNu}AQmZVf^1}vQD1KA%#$eXVLf*tE8
zpwhcX)Npt*21X0;@Ydm=c|wUhX_Wxx1<rVr-%1}Z)P}E915nsU8@9GcH5?r#4q<)L
zP}cU4l$x5st}`K+lqU}p<)f%&m@ihA93(YiIUrml0>U?x@paq;SiLTc)<zfd)+blf
zSKe<(XsHG5XzL`wdnCapzKm|Ux)+8WUBq=xmZV1(?WQLDLvZ8PMVxwk7BA|+32Hn%
zmEK+5AegbG4s9=KLtUA8gK+Nxn(re9cFH2){!*PJW(cUo6jfkG^+T4YGp5X4N)DGa
z(iunh&<>9_I_u>=bXk)|WM+F}&{0ctJu?YHhOEUp?~Q1>w~8EJA}(;l9=vzq6RFdv
zClll*<3yeB<k_qr^pn<HEPJQU`reWflnJ(=YH<zSu519aWPZ{Elhf$#ySJE|29v2k
z@RNd^3d!Ki;GORT(2smhYNCVCE=`#2DcC`0F3G`sE%AmAUy4Z10wvg=;*7f`E|SQd
zi%`n_FmyLH(-{NX$YD<{jPDkP3!k6D^6ieCve!yT-t<DC`+5Os#0Gk*{}<;BUJDoB
z+2Z|Y{$S!il<gWB3g6bH5#1w>B&VQ~R4>V<>h0@MwWXJ2mMYVhm;Q9HTcRP=Yy~EM
z`c9P&!~!ndft}B$(3JjhxINB+c(9IWz}-OetG3*+4U@s&R1XWylE5-hl$xJg2A7h=
zv9&%49}M-x?K>hscElYj6fE6fQz1s?of(eJ2V$VFE{GaVOT>hyZy@!Wl)z<#Gm_hP
z>C<;7QRCAFNEO>eT_se=$PuX+*fE;e6;)#2Qdb(`X#v*1baLm@^VDTj2${nZ#>)<u
ziT3$ea`3`!l6HADj+;IV*5(}`17D9)g@u8*cZCkoxsVKd4P>zGWFDCJ4kcy>kKtgQ
zF6i48Qs<ySdg|6`$dT`%lRvG4D(edfXPnU9BNT$N%P`4)1Lo{%BoKO-yvUWpGrkXS
zr{7t!a7Q4NCVr$v&db58<t&z-&Y?zz6VWa}ANRP9z>rEGvUYe7EwPsX$Iu|$)OSaa
zVWWUHo7Bib-)Ne!b^~5{7y*4&>*!X?1hVenE|h6{NDZsoxz(2QvHsC+WZg2UvP(A;
zHI5H+N7SNgV;{3=*fC_^)X|$>!c1&I4Tg<7NHhYT5J@dB_-5TpBt;dO>QP6T{qr2~
z>E~?b-DE$q@B*KDA2i0CJ1kbrOTqKc7l6|7Oe#b#Qqd2yu`xpx*If1l*V5x8*)EiD
z4)5rN)7doGSrdy})rh(EUed2!K{YgwlI2_V;jWJ^+7wTO<rf{$(IFVkt)`*+qnXt5
zusD1hF3rRqd`ZXbbEK&X`_W~}Y~&q^N6qvaPW*@=jWeBzN<#Md#yN=)gVA{Yrhu7M
z*h;mh+~l+dchgKyZ)_CXL`FT|#~d(SOdC{X$s7+$3>i5R>c-2HH#aU}^wf=5ys970
z&Y1UGKRh4TaOMy6lm6eRAG8!tf&%$F@TJWfEEbEvtpgb#et82K`Y8^)Xgf%@NP&&W
zdU&O*4~hy1Cr{VIba4Tx*5|?Ow0%(0z5_Z{>LBcsK4{IK12QfjiNtw5*k6|i-ZO<*
z-3^MMZFmSCK6L{<$8>OHLqP4o6mUAX8>$K%f8+PqW<KLj_}|~CA0%HjMb8s@M0C0s
z>WXOMGz(K+UWfw5U$Uah#!A7v=JUkmzzsoDstIqzumt+#iW_rfh9ub$RX`efC1igG
zL$^h32g!~w+7uK^!?m-CmA5308-Kz0%z>jsWxxkbj6%5EEt+&zO)qn1bUI@;O&ROt
z=h1Z{i+<}TspbUJKf(Vu>Ibt&JcGGT8whnYf!Pm+u%U7z*kF?naD9&vT|7sfy`B<F
zx|b+op@So>sXv5o#w((>gg!dy*Kp>6m&q5WeA)yOtmMv5xU%B|cVM3sj%^;pO08T6
zM6(7I^hN-0r9U1Z7s!^&X|TO$A5L4VidkoriTx%1@A*HjzT)Md;Qt%XuXOae^Cwmm
zvs-S?;;(wG%-6}7$Uo40i$89AJ1ZI4#CI{4;Oj-D@PF94@tYEZ`NtGe_;k+>e(S;s
z{Ora;R#Kvl^>uULOJB}pWvl#H!Q@tU@a_`6gs}%dsd*MZaq3L|y_3sLK8N(OeDx6i
zSh2DEvWnxZ$AszpWTj=l^^e35!T;j@;s1{RZ}=yFL;o252l}VWqXl?FW8sk6d!oB6
z3s#7>!H~AKFsnrmE?%&N11&uGBApBqW~hRecNqj5WWse)0*ll}LhiC2vTT7Wl+S+(
z#wr<LwZjZ1C*{Bd2Q3(}cmcdmR0UH(8Z@%%aM~^y(pro`|8OcOM7F|PXEiwH=l^^C
zRnG5t@Xzu8H~+%mU;f3?C5^DRM;MofOht{WwP3Wd`gc#6j&xFx+^?hL{vnpL{>PZw
z7A7<PFR_%cJVCcjy3Xx4zem@rL{hRpn_IZ|8@Z6LL8a$;ks<4@P^|}5WQxaKQl(l=
zC3Z{DhUj|YrCP-JevhVt)CwqQw<b$oTOsda3b_rh$dXe+ST~qVrQaN&n!CH`!Ku?p
zTC4<pD{Ta0TbGe<ovVrArZ^gFx8b*cG0R(b?H{aPgTLWlq{oZ{>xFGZBRPO5el;gM
zM5W2%-a}mT17CV|b`rzbu_S%<TCzd=5I1%{(rpWC8MEn|jei<M886$eLf4Cb=Tw)z
zV+KHn+9zl-wJQvnSKhHq?4USfRH8u>g65IrnjS&wqY$Q8J%Ci7v!bP!G>FsG56s=N
zUA%wcw=kG${wMhVhJVraGzNMkPLZC%$HcFsfpG8*^t)Qg?VUlS|9URT3iv@JKh~0v
zM<PUM^Tc0zY&^sd43M<7o5>TawUF}ZFbOhvMBdKKCh5;2$ffn`0OHj^!aj?<E)s=z
zN~7Um%~IkVnn+Tej*(-=ABb4dco?lC3?C(9e&4@DxAw>X3I4y)|M3^{mx11YkiTZ!
zaR$-7HBh;70prQ;h4RXGpk(d}8K0~{UZV(X^>|Qw=r&W`zX+nwr$X$^?O=Sq1|pNw
zU}$^>sI_MiF98KyGYKAcZUV=k6gar`5a{1J40>9rU~oVMY{j?1ysrcXrEkD{`&&@p
zZ~}(fABRqzqTl&TuPL<mPvZYK_>-lp2n{~|o%Bb~qr=Ph68ilxO|)#INeTD4^h+b5
zJHww7ztl~I%lPEseF4p?y-s+aHe#plQ|7T*46W_oO+-ziiQSDbjBZV$s#ZEsu;V*@
zde8vFPcDZ9!wC9PR2Y4$>zMxc4cx7<Gw2jY6SA~LjZVQmzuzOUbJ(EvpWy!+{OPub
z4u(I7CewywbK7(YbuX}htV|uSbD2cXzObMz7gaG-RS6<ZhT~6tY3!1?M7t9s$V!d<
zc+~Y9l^Welo3~sfs}$;}e1-&f^TS+FeP2sYYzssGFFWYmNN+OSV;XT5%c5&`t|oW4
zkAz>!ft(GSj5|hd|DAvLZXfq2{mTY_gFkr)mGX<5Bv>mg8@6!iHFnzbG*(RP7Q5oH
zD7*bYEgKmV$huq4WnZ+XLRhUXfA8u@R&s4S>)<WN&KjJ;9=8_Z8@cad4P~CPHd+tC
zDc6;C*gT#MF~7kM`_Ro=?mNnw>Rn~SQsr3ZLK$|>mTq>g=>q<wQwHpX$v1!Z-yEwb
znD;06|3BeR2LB0vGS!o#la6QOu~rG{v2QzSeq2LGzcYdrRcGnlGyAF3Icu2F6o4Pk
zjll-q95N%*8WM9dX^d?heYigf)73O#bc-K(8TppJ`k6_-O*97Kd&1Owq!qX9*(G2f
zL_wC_T>5F~QZ#t107}!NnXqtqHdHtity24_**=Ea9!0`wv{Q7l#>azx#4mj<eN-|8
zHpOfwdrz2?qg@WDGZ2ddLb=3zQzuO`vqznvC@g2DQoR>`L@hW7llvMOp)3vP9+<}9
z=e5+eVgq>o*aR}8{J5O&7eLv49PZ2ANJQ;YLFc0sv}z?`XzD{MD|-kQ#vdW^!2u98
zGnJ9sBaLHi!eRbYb-HHZ8u~K-Fs+M@<E{_tHe^m<>BI}WU_tUnXbd|}qD{Br;$M8S
z(`ZP;rX^G1v`-DV1>azP3|&L{^4l@)P$hTq=m};>?-F2gUy_wo!E{nOi<V!kAVt1{
z7HyKEX>JuXZJQGsu@=nFj!;aneU1UE<sq*<AJ@eQ!>k1>XsnGQ%_#XsW~z*Z4>bVk
zvk2$DdLGWX<Ba#Mufsy$N{oH^jyKtD1$a0$!9z_68eI_pF@=R?ePvJ`O|&iU5<KX^
zA-KD{yF+k-26va>esBry?(R;4ySqzpmviB(dUdProB1(&de+{(*4{NgW~QfGPxXpG
zO6Jn=`G6s?jn2VQkv#nyzz?el$n#_C6!(l@mnIc|$U!B-s)l2Y&_~CvS7aKDXNDgt
zzl`*_lv5?O(H@1bIUY(pi&?YwfT;3FIiPcVaWwju+!&I!VtpseNXaZ18W5>gwSEYH
zsD1j`im3{1!lW!c-SSdN9(Zasy7w`JR|yU_(vjO{4U|^!L?n^{sFJ#ngUE=)UaK(J
z4W_9I=tVxQY46J9i#j5$E4w*TBO|O)wkB<LI+1$q&8@LkXIoGl$z8-j9&GOPn&LYn
z8$LNTiSva2xw-(4$FDM1L5jTbjUaRo;HGjeec1Cc36lC#ZpeHQbWRExbkW~`@x0O5
zBRlJoN`Pe1wqWX_eQ&DfucEdqtk8IT4tRuDiFrGF99*^8=*<nCV(bQCyd>Pv?1D?e
z(zB2wxZnHSWdMC1heQGWsxC3#w_-C4^a3W!Z%5`(GoV#B-FJKX7=8e%fn703J%m01
z7Pm{Vw#YBScE}T<7?Y5Or<g9z7E{7a^iZ**i|7<yG^vhOgg&#r$>cP1#$o2zuWSav
zzT$2c?LEl%=e|;QSQ<biz369w@-GGddemBN;us14{AfSfs;R@+%5qDllh=S*&eq*N
zo$0I*6+{d<o1oz8^5?TlL&`WXRSu2NzgDp2(O)MD)VOeIsZ8jM2lz(oKLs28vw)n^
zc8g_YA`%%)(hZ-aCFJqY!zRK#M6Jq}*r9l5jk>VQKQ-B%maB$Op>T0P9OlbeBmxAz
z;e~hKDwBKSl$)#<GdQ|gBXlP6p_Wc-#@L4C(8oYV$zM_1IK>8#;2xHYN=e4qT;!o}
z`(8SBfj?`|*wf4w=bqM;z&uP|xo?1|H~s)na?SST^lT}{lPV!`E+mWrI?^x;ZOWZL
zwU0pG@XKtaO_tKh+}{IHv$Bp+IR}4Tdb$#8yL%$opbaR87kcbn4=;+jy)vC-0-)9p
zA;G)DgdK{h<R{|HG3m$*wO3W_{Srdueh{;Br@_9tlGyJwvW-=xLuABkVtPMmNq62N
zGY|ZUmaX>e<p-jYQwgq79Nqw@`s<S{W+!havaICv6RA@7xzT7<i9~C)>Bh@72YN;<
zyz!iYm`9Ak<~r+2?SF77Uod-bmhbF|8n#?#<)&DDV-)>p0(5>-Vl8tg+}`Rs@p3F=
zW))tgQ?I_8R>+!8?aH%xzkG!ezmcq?Rh?^5hwV~E()_8FuO;DHp&K!-+DDRq%@EI9
zBF-a8wf);%(Xw)YoNX_=PxRZjEw4MGZ~HV`zv4(&kZ^{a7)_Z+9uY=QaYI;9d71i`
ztg1%x;S_Fv1WvCyJ2SHeq*TYE=LT1|K=pgNpJ+wyZWP7c=iA!zMkx*rW^5Nn(oWCy
zeWe*dfkd(0r)2J2Lzz@}4)y!#L*E_2jJ;S8Q!Qet7Rj-=v@BysFex@f7I?p^C3-MK
z&ZK1;UfzruzaWL^O;z`GT*vjx;K!OHXU+(%E>&3-*||t7^-apVWs?J%8W4qE=@xbE
zWL2%x(W1>IqI1z`N37gg5^{3>vNXO^W#mvcClGkN7o=|53tzExO`t0gVBF{4AvEOV
znHNwP@@(@AO3tneTRd-g>&iQVQVSzz&-0W!yemblrT`Lq&a2_ltF)f~?!}o(I%FuK
zGkUphp$LIpNF4a7B6_c1<Z9r0#KpqsdDNsN>6FMsC!bm93qjSbtg+>#r#S9~%3*l{
z?lielq$dRv&*uq=B!4ht;V5X}v4DAOCBJ5JF!NXvO=C_fuyRL~4yfR_nXyPk(`+3#
zAO0->y#UpM{{SyeJ5F8M>^=`G)|K`5cDPz`WHU9JBr%zpU1-^{k(7j)2E@XTbWB<e
z_{fR);;!-tl+bAoU$_S(*6P_%u?&3u3p6vr&VSfZ1Z8DoUW~>D<J1%7b(;l-$p_Pj
zg~?7GBklw;hI;Pa08#AgT!}NsPhxr5_eEW<c=xi7wOh0fJ6i<!af8d(U?=rZ**b;9
zu0<7v7ekYbXT4s+PUhpkv@v72Ma|<V{E6>zDL}Qs_}due7k_rv0rGiE)$ab{t=ru)
zO`pJ*JWOwmEz-<O<!AIt%Jhsn#CXn;5j2=;c$v8crA&xv*tnw8@@9zAF^L@fGCbj5
z7)}C5D<M?s?RLcaK$l*Mjxygg7}qrczoy6)5ZeXj!{ve&($NjIE@IQm?@RbqH(1C=
z(JAgC0gOuHr0*t8b*pbIAxB^wJtXCuaDWZ8-gLz;GF8{6vFtX3Dw1v{U(DB-yU2z)
zsLTRglH0{*QUg;*dwW>!y0avrtEXLVSLT1H0<~A;`?{)J#4%wG-i^WcO?bCD`w$AL
z_b^XUBUl2bX3@M^y8i8%NZ;lfPoxw$-TY)LNP}@!V#OGE#qGVch^y+Odlpcg92Uj-
zrOmH00wrC~iA^q~ZVM#J2eau9{u1~VzIL0u^?Bq{a!qE4{IN0>(pp!gcGzcGE^CQi
z`ZXRHqiZacrUIS*>`iT<?}2b9$sux<Z&uJTfX>l<OHtAbh1me?xew!{Q@oTk*a^j6
zu^!ll6aI=OZ;8oR%Sa4U91|z2&G}27Hgg(7F%W6OwWdZxlflM{g$l)Ip{W7n+S&8o
zt@6HE>q{bNN0wr?PkWt+KT)V_<&@OAKzc44%M}TEF2=riBEGSX&%K$?li9KqgP0DM
z7lf+d6cT}**XjY_`pZx*K7+t1CZIX&<g@vJ@pCnM<j(uW-+E4<w^HL9PNTsNae4q0
z=|$EsSEy)s=8!b`Ob2j;@Z@mROky*{G<Gie&OCd+TKt0vi0Vod8YKOyq{o3ZhuQ=g
zDrM3@!V@R<{kPl>env5i3yu9UMb;UNQ0KuJIL@exQd1Fv^&yxATWs{C3?p%9(tUq&
zt*!m>0S(aJl)uFzx+gsYFPH?w&Ul#Y>gaeGCe;X+Jkchr+`Qw4#O^*ebj1WmTE5-M
zIA1klX##yj;SQgCUK_Ovi6_>sqg1p!k8Csr$S9&3S`moLuzCqVUKx>8h_#9BMsaZ3
zS88<OS>g7=0iGyP{i`WM)Z0i6hAEN&hPD^B3?(^=JFbG=HPkSwZD7*r<@&W;Zm?*o
z72TEAkSwG4wYGx;Nhh|P9MbH~&7t;U=&}EKUkDAUNn&uenv!d)kYy3^CAmB6i;FFX
zZXawX*$-$0IA@rm+R+{46QR~^9hb9R(^<AWR=5x-EdL_5cSY8>wH#UA%odA5S%HV7
zx$7}pH&Q7T5aC<XfpCSkLQL6Lpd@NG4?13zDgd=f@wwmXZm)n7aD2riB`Ax&st|xS
zO`3te8Wl{*)kN*wiVI=j!Do$0g?{kgQ_*y%ucj7Y`i2x@rGutmi+1nPh$>B=DUv+@
zU9_+XfyrK;b4O_dH+0>m$4E2KNGC-om8k|VW3n5jD^k-R<SAZZJeOS}C?)*O<BDni
z0Z&G4JLlgqB_@8dd}bhIoX9&)!I~Ym@1rT^a@R#1*C0I@k>)T0z@U<Zmhlv0R{BN?
z+B6I_#!4}}vWcZ(^A8MgivkR8ty@$7j$$)cK(gf-$VFuaBfz|ZzK61c9?d`jZL8T3
zf8%?c7bPedPBlqSK_u@ay&31s5{41yu-F!H2z3?``ZC;?E@WpUaskKGYKQD+-6c+I
zD;p-bTnzU{P$^vb7o3k@G-7Tnj>2-0-J#WlX6~n7$iq_-eB>zyLMO^kiCJ#luY`nP
zrB@fVAU)Zs*EYamI;L=xfdCuFPk)8e@zMRAe(HyP*E|dh3C9f=a#@3QbcRIF;9^^r
zfZS@XE3(ixa{C)gbk{=8bc`LO_89(QEY6b}u_Imr8%_dK#)I$|fl~(O#&Vp=Gfd5v
zyG(cnmnBFuY)vJhTC8;YW`ytYdWamfh_$A|=EmeHk`0>9<dYh%<Vn_ewXr)E#w2b~
z93)OldjzP-;VZ=Z5KPrMNFMy>3UzV{If-ysOUW5gz$h#FM4(q+z%LBLd1(j<wV$H3
zv2%uH6G~D_Y4rD%bi!AIvGy;u>(Vi6Om*d(*l=FY+OKXY_2jO9eG^@ZaWG$hM-bRa
z8l?;v7EBKM_b;I>8BLJCikJHj09*qLLjGwTFuLWvY$1v~M;axnkzZ@g;w*IdMg8SO
zqH5J=C;xN(<+hU?(!ASSB6l&i510bgbl5FQ_zE1y;u%V40XL#H4r0!MHye0*G`rd$
z#&Vx606%F~4>LH{yE3@RB8GP3th*E^M}H9zW|YM;66U4&8uUF*Y8?sv&@{JYU*XbK
znW;*Njn*?gWCU^X8<V{hkK=h$kbBhZKz;eYNK}^&hB5E8Ftw!lt?(K72A?zZx?{O7
zdaOj?)^lZXb?NXW+7M%jvXmO5%21@)JBz&cebNR7juQ^xb@x0nU0*M#uJ&(6mSyDC
zc8Dn%s?A1)7hSei5~Fbb3?`sNxLRzU2t1BA{^MdcjIn=+zgp7L4v1pJ{Y8n}{RKF_
zTGN-#O3W(-&qG7+56cTHyf8yyFZ2WDxY?AnpgERGxb)zt_pjCz&p)sX+3a12(v%jr
z0_bHDd<53L*<aHy{3JpLziMtCjUDX%x|lxJI8FiPaPnZhMDJxpZlS`)&bg3A(<_Ak
zFi+Mj-tueh+=(?*Ux2-i6eFD5?}e2eLEgKe^(7CmuxWs!CCDmH#Vo<^9q}8Z5!gUA
zubt$&zxupB<w2<_a!3?Pve|bF><9lqb&{U@DXOI3h6ZAu##c$2jDKeN{(2^BuC>RC
zE~!PC^Nn)R{;l2SwLmu9-g-=9XKfU6;v9}*gFTh}(U!_+{AX#WQH_1237z8^eXPbx
zQ6H_BZcDQ_?;L+?=*UQW$)ttJo|ouv_fCah_)i4fvx4H@9%_cU`N6!)<Xka-(PPq5
z5ZN!QYR1J@%5c6-Y2cVEeD&8$%HCR$(){8=5Y2aL4lL)Ow5Z;or~4K`Y8F#R%v{e~
z`oI<NJx_A?Nx-$m;*32^S6`k6Q}*)ph=TaHpuD%wOZI>m$=Axv(5<pVT(-v|EL*9|
ze#8QloRpX6xHA$`V8_0{(oZYu_%2SJSLRAM%Rk`oM4Yf5MBPEf-24OOqP%FQh}TL@
zCYTGX?owR}EsSSZ++aeI;UyWz#z?&O)0MH*wa*M)ymF`_C+vn!J?hT%SFFCo3&g{A
zc>VH->Txak5ob_`()U?t7BRcZN3FWV0u8)TeEghga`p)%+CiksevL5$f-_+qubxRZ
z0v$jte3X85z_3v7D)AcI9ANEaBC<SW5Yg;~GYr4nMEE@!dF4s~_bjn+ROlN-GpeAt
z>f7X2VW?&<>hBQYt-9Zj^6%hIxL$(j<=IHt$&Y=bqU=j#4piw+>WDrk&S=rTI+4+Y
zd6$b><vK%t8^Xc{enz0^Ew@)+L%zQdV>kn%)hA3f$XW=$tHuLmgMyUvgvm$bMuVoW
zyJ{BA$ZvfC2n;?v0pT+)iRw86>dA<`6*|}~>&<;5HAqrX+=!^nbG^mQtXo3P@JQ{}
zurgG!JxHy`_FmuF!XOWOIw<ii*76am8R?;#w@c&5E;dwr*AtKS3~)hjF`KA7lF!?k
zFPnL)>Zi31vqGe;Pc#eb8)-JB-NAy%Zh_ApcO$t40HQlmIq&W=bZ+|yt*aHWyz@My
zM%8-6W@`cLmqlAF&)Cx-TOPvUFUL{OR!3S=;wf<1)fbQ)Z?Sm}EHeZ8H2M}?GFi?#
z4{%=OjKO(~KinVoE|Z^PlX5}j2H!vzk8eWGGO*tA*}DzSF7{R}qE@%&TZf(-IeJ{n
zeHwQBnRV;o<tyRi`RnhL*S>$^j>@21dSRv02cf9Hh5ZsLrXZJ)MakT-+{GBAIdZ?Y
zHJ#lE{kY!<UcR@itlgi6_mr*mIgGd_V$}m^-CmTKrug(z{T875AWd{u{Isa@4Lm?X
zm%JMWmJQl8ESY{O?ane14f`cjRH>L`cv>sUdp?~}<wS1!rkV5d&aL<8u&t$}JFcna
ziJ9bhtv#~NQtc7#sXpJGFN}B|9K(d1CF^1`2)vSb+vtCZULfSYB)_fYV^U0OcIv@R
zuwWwVcBS*?<dKUE=X2Xp_)KT{)VEx^y3%(8S_>+V@4Hj7wFQ}9ntUB;`C!#4i-d?4
z^QDS5jS9H#Fw!||Srb=F`U@T*ob4?AjLmDII;z@*rrTAFw3j8Xg}W2iKE95+Y50*^
z0VI1mUvWx`g(-1OPN0%5%t+j2Li2Xr^oe6J0behrLvCl)VN2sEubI7J<`{L6aPiqA
zV!mt8dB#_2IXf7mAy}X3t64L{+q!Cp{5YL}WQ`;R{@{e5`90K!iwWDfp8!^EP5!n3
z)f1q5n!CmerZ<d8Jg-o9biATydc0O$@_OU06}Ew4od$g{gj|)=HHigLHrDI3iD%E?
z5l%zkoYVFXEsY2r!u$-l^>da<YrYy-!FmfDo?b6SrI;r*Z^J58{U{dWbMHTQv(T5$
z<-+9i1<qqTf4gnjh>yb_wTogV7>YhPZ*u#|U6L2@=x!^>?Nta4u+Txm7iHM)B+8!Z
zD<X)}=%%EcjNLk0Ec(*Sj^l@o-J@`71$3~0lJkYx;!sWyI%dsub0i>A{RiI!{otTx
zg`j*!kG%c2csJ)>nrqXTpJ>e>DpXd099Hr*IMke||83}ea8jaRyC2rrUb_+3jVNm?
zCD*_8aFCk!1#IEG1^Ms&Q$!&vbf)4x%-M=$9(8w}W_<BsX>!+&gQ}HnWx=n3=2P>B
z2)zA>_{PFNvHy}bnLz9>iaXsBkMp8hLk?fL>={#SWT0y`P^h`ZgRW<Z!&Wemdt(WO
z7v@p+@Q6g?hETYgi_mh9+*4g|sFxlmf}`q~MlMQwpi!>_*bZxzT)6RBT*!^~7K3Xh
zje;)4_bf+nT>X<ehx%LKahbFhbDq));B{NSj3fXpj{Vf^s3>1WKqmLR3)yXczv+E}
zg<yRdxR2f*u-AhtFtpVfG%@Gh>3c|=k~c4?6fx4P)1meC)7SJoB{gn*I;|H0U)wE!
zg_9ivcaHHn9V~d#fIHG)T?69N{%WP-zzuRqSwJT2Sl{=*|8`}|nu8-<&-_RAi<m%1
z@B}YA4K;2)f|vTI3)e55=~h8|)rlat3c*r*DbhtLT5=B)hMDWN7$W=$bbXd&I*X{2
zU}OPac+RKwFTBY3)O7kY;%8a>nB{!!v}&>!;Zto$4i^#`eZ=O!cv4s3<#!HD@+tYx
zZ0_`1@(d6>UnbPJ1Wvnx3xe1@8+C?v^nO*h4B3&Q9N|+$rJo1%OKRuid^Z{IP6GYi
zQ|@+rPg%au@^d`PMNQ~~m`*HA>2bV=nR@0J4(o`1U6CZKr!~sKsYAVPU&$L8R3)KP
zfUucaGwNBAYj7*1oui@Y^=eI1B2f}96|T$SttUHUW(vV)W!W0zr(&aA2CJL<NAkwq
zYq4OEAWX4<^{%xS;JL|<s%cID)k$ka|5B`>qtN+IyAL9Kar2b~nOp;I1>5r%d*3kD
z!8cbCcqKPPGCaw9SMFSDPSr8&chP$a7m!V3X?tF(hM0J(P;X0Xpxrw6<L-tnx6K~Z
zxVQ>GCQm1xdRP%C!EY{AqT(mwz(+nfn|*N_CRb~{O8eehbKxK)AbqJuL&)Gcmj~C`
zczG3(!PMS#?e*$?99Cjb3fg3Tmrqj)ayUaXTF(=y;O0-zYkI~&MQso<x1-1*I-MkG
z^+&M?YUha1Z`8C<Ys5CAeKEdS^LeaOV8ZUG_qRo?CV7#ZcYlIooSALw^A_^;dNprN
z&CFT}b3)b(D=vu+;P?6%lQ|MtGBMYeh5L6r$Y~XEvHee6$(PZKtKqifcbQkx`VSiG
zTq+R86<u3Tha$ZStMMJ&dkARvoLBuwrKGL397-99(x&a9`RCRPz*~f<k2QDvQx2(x
zLAKiNk?&mBLw?nlla<KSuM<93`92O?fUE$|#cWY81d*_ioUhz`386}}Fy0K4t9?vs
zcqsI3Ax7?1moYkwN_Qz%+#v6%RUF&TM}6jZe*V|EqZ=CV6dI>MM9>R0)lwOyPQgKc
zk|N4Z)7cqG*&|st@0b07J2>?(cYJ=pRX$&`;#OH5Q!nB~d<Q8}wr&aaM02abeQMnB
zwi1CY6&V(;ggHSmnff_#%h|SYHkP+mEC8p&dFO$>NfXh&V^^|+u%C3~LOEc&La#t~
z0hDsV@)CL|N#}HT?_jWZ$9vFy-o2k=y!!f0W!zHgjSaSwJmi%ZHiTY&Yxm?_V0d=j
zFUHQv_DAbvV-wqJC@u>NJ1YkZD;iu>D4bjf3ZWI>*qrmFA?q&54^g>j`L2)AdQ9vr
z?6w}0e#YTJ`c4A3I*#9-=<7if#d;GkDViJT=$lm&auZJ*Um5hbq-Tx#VDak4vp1_4
zDy1e4{VWc#x~d{O4%<9uKjxk|$M777oUzj^)^N{LJRN|=oj$AW`vtXrFlnc0tx=O`
z3_Dw4ahy&#J#OjSo~TnR`88QO71Ut)-K6v{ZffqK*LmAZHG9TFLmNPp6?btOT@exZ
zXzOjwLm8GEI(7T9Q;5Hs6NI~xTJ#%<Wp_1ak?B?`J<HQb*cRUzP89W}d(lIVS~O$I
z*(hQ-hvS4VDp%gpGX($XK{Z6ii(eEm^<G0}Z5^gFAa~q?=clT=ic)=na8wC1zGZs6
z+0TjBG7eyjnW2Ol{T*lhY7QqeTf)5}9$ic>t}-iQ7ti-Oy*)dedBZED@vt_pzT1hw
z14ZzJkEv)yS@S3nU6<celQ+kvqDl~cgU~^#F@}~BbM(n?n+@@<=pTrt?}un~<6%5v
z7mta6#V7jYA7XFrw%e!tpUGUvb5+Q-ENWHMGw3)U9>gwSVInJ|$!gsicK#;InkVr`
zQT~mY--=e0l^dk>Y-ECOS<gmhwTjGPh5r%ynUgyBt)&;Jm5z4Tw*$ZHn|X817V<+Z
zNmCkcJ>-D*>qonePLM2q`h}^5gxdMil<Uue@tLTS^aHy+ka#n;y1?%ukZ_QA4HKDF
zTg5)K!p@7s;vC`72PB_Ww)&lGml>CByk&-ik0v1mSAC_bP6g(NoI`i{?sGLwzRHGp
ze-!ohIbiM4;g2-Zn7iZ6uF5LMpN@<WTzLZYDpN=F&DFBIqq@Y9GzFuF#km;(l}&f(
z4_xU)XCis4{>e_ZF)rY&@acUlEpK2zn$2Wsb{rMS;%g`Y@}{xj!9+=HD&ti(3AXw{
z5Zl;N+9KI7xzx^!Qa-z$Z(s;6Z|m(&I>(kb-j6eaYNM@H7(Jo>sNeZ@O;3&BI5SKp
zRNb_>hRn#Ut_!DgqzXHdt_uB_V-1*~DUTCrd5h%Q1V3Y5Xq@Q}AgIL&K|uP$`L(|9
z+ao42sA02c%lISiUiOT%iwM%#(x~7LQ058ZBiQw1Xu($G2c1ztz4$=J|J+SQ5W3_(
z-Sc5}SC?Be1JnN4wioi@CjZvGnSjjmWP<q!1t=1|GxYlpPmYIBRbVuv@XpMD>75ah
zhOnC*AkwSO_Bw0H4!xQCdF2?O_LKFK+7VLOOL@^}Q!56hcglzCCs@gbrutVRXx6(y
z!J1r`zNQDu`{GlHIevA~|41N|D1-%k;2hpdb9W&>Wx$RWZ-stxj%8BpjkS0LEaBd8
zeu^<^Zwi?Q%zktv$V8KLE&B<U#46m<`~;&CuB=R?dLsOY7(*Ri=Gmq~`uh>VOnP0l
z+VZiuoFDTw<dcqaflmJ{o<(YH0w6G4<_Y67?k<k<rTKq7gL1bR*~adFcZ_gjz9xnE
zwM~fjq60BMx!{V<H<5p$gZdkqX7wWi)AVkFDjO8If{Z@G#><p=l@Wek?)O%zVf!%N
z3owA6$?^3^EW-d6ep=3OCmtk{pQ@)2FzFQg&sq|2Qv3t4J|!-<@_%}7!W9rXeU{J<
zIS$XQOo=Ih38c<*(qyRhky9^=Ii@n&Pnf|@-j?mBYCp(`Ky}MUz*xlD{HLhyhg7^b
z=8p{d?J^|QVSi%XI#p^oK8@xml^PUeR^a7>u8O_gr(V0B{AY#l2tSsZbZK%J$Lmga
zJHA_~$^PdbiE~$R{2$pjWz35IjB#<y3dMeqA)fX|cE~a}ubnRPJPuOMXSuU5|Db&o
zFFU>inGlmN?laTvB-BeEM*kx|F17m<s0Y;P^YO~{<{szP3ff@1S8(IW@MCTHg{|)H
z_44%|zXY^7^QeUfMM1LBi)B1o+C^X}{3eKg`Pz@AZZj<``N9wM&ktNl77_{*3=9ko
zY}>X)%R2)JT^k$>tQZ0e4DF-!r=6*pje)(9qtQ<%1}i5!+hApFg?^?loickDO5EQq
zBK;BVp?+0Z^7)H*jL52&IJ*7F6hZVRHoL-qp=T99*p<@&CtV`9P0v3fq5VnYuE)6g
zW0hn5?zZZ<*$Q80+D*Z*E60bYrf|mbeU2k3+I1L#ZfZK$FPkxII`o9bI;|jO-hyRA
zPTD<|`>d;Epckq!G?!R;Uk+W^V(sHk70AEh;1piMW=Nw*8_K2qw6<a${*68EO@^30
z@FnxCA`EEO*iB3fiQXqA`Etf!?Xx-P^7OxLD1y<kXEOd=<hJGg#YQ=^?xavUa|Qwt
zi<ewfHJ`oAGKX^5?)6PftDUfpcUk5sT|I-to^f9L<<fX42RfhPqS*mo8LK_RilxZ1
zsh`#;PmA8MTyT9=v{Wh3jd47fMYvb1v?z^`hjhK7cnx6ZqVI}`DWJKBrD>E73g~_*
zu>ykkYS%X^_aGH3O`InU35FGwWWgaYA^*1$%?s`@irHXbR?6UD2p^UBe_KJor)6vA
q>}Ka^&0z2RfBpV%1^?I2?IYU%S5GBbXqZn3#K-RTaRlXjp8XH(PXqt}

literal 0
HcmV?d00001

diff --git a/clustering/tabnet-classifier-tools/plugin.json b/clustering/tabnet-classifier-tools/plugin.json
new file mode 100644
index 0000000..6c65b41
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/plugin.json
@@ -0,0 +1,672 @@
+{
+  "name": "Pytorch Tabnet",
+  "version": "0.1.0-dev0",
+  "title": "Pytorch Tabnet",
+  "description": "Pytorch tabnet tool for training tabular data.",
+  "author": "Hamdah Shafqat Abbasi (hamdahshafqat.abbasi@nih.gov)",
+  "institution": "National Center for Advancing Translational Sciences, National Institutes of Health",
+  "repository": "https://github.com/polusai/tabular-tools",
+  "website": "https://ncats.nih.gov/preclinical/core/informatics",
+  "citation": "",
+  "containerId": "polusai/pytorch-tabnet-tool:0.1.0-dev0",
+  "baseCommand": [
+    "python3",
+    "-m",
+    "polus.tabular.clustering.pytorch_tabnet"
+  ],
+  "inputs": [
+    {
+      "name": "inpDir",
+      "type": "genericData",
+      "description": "Input tabular data",
+      "required": "True"
+    },
+    {
+      "name": "filePattern",
+      "type": "string",
+      "description": "Pattern to parse input files",
+      "required": "False"
+    },
+    {
+      "name": "testSize",
+      "type": "number",
+      "description": "Proportion of the dataset to include in the test set",
+      "required": "False",
+      "default": 0.2
+    },
+    {
+      "name": "nD",
+      "type": "integer",
+      "description": "Width of the decision prediction layer",
+      "required": "False",
+      "default": 8
+    },
+    {
+      "name": "nA",
+      "type": "integer",
+      "description": "Width of the attention embedding for each mask",
+      "required": "False",
+      "default": 8
+    },
+    {
+      "name": "nSteps",
+      "type": "integer",
+      "description": "Number of steps in the architecture",
+      "required": "False",
+      "default": 8
+    },
+    {
+      "name": "gamma",
+      "type": "number",
+      "description": "Coefficient for feature reuse in the masks",
+      "required": "False",
+      "default": 1.3
+    },
+    {
+      "name": "catEmbDim",
+      "type": "integer",
+      "description": "List of embedding sizes for each categorical feature",
+      "required": "False",
+      "default": 1
+    },
+    {
+      "name": "nIndependent",
+      "type": "integer",
+      "description": "Number of independent Gated Linear Unit layers at each step",
+      "required": "False",
+      "default": 2
+    },
+    {
+      "name": "nShared",
+      "type": "integer",
+      "description": "Number of shared Gated Linear Unit layers at each step",
+      "required": "False",
+      "default": 2
+    },
+    {
+      "name": "epsilon",
+      "type": "number",
+      "description": "Constant value",
+      "required": "False",
+      "default": 1e-15
+    },
+    {
+      "name": "seed",
+      "type": "integer",
+      "description": "Random seed for reproducibility",
+      "required": "False",
+      "default": 0
+    },
+    {
+      "name": "momentum",
+      "type": "number",
+      "description": "Momentum for batch normalization",
+      "required": "False",
+      "default": 0.02
+    },
+    {
+      "name": "clipValue",
+      "type": "number",
+      "description": "Clipping of the gradient value",
+      "required": "False"
+    },
+    {
+      "name": "lambdaSparse",
+      "type": "number",
+      "description": "Extra sparsity loss coefficient",
+      "required": "False",
+      "default": 0.001
+    },
+    {
+      "name": "optimizerFn",
+      "type": "enum",
+      "description": "Pytorch optimizer function",
+      "options": {
+        "values": [
+          "Adadelta",
+          "Adagrad",
+          "Adam",
+          "AdamW",
+          "SparseAdam",
+          "Adamax",
+          "ASGD",
+          "LBFGS",
+          "NAdam",
+          "RAdam",
+          "RMSprop",
+          "Rprop",
+          "SGD",
+          "Default"
+        ]
+      },
+      "required": "True"
+    },
+    {
+      "name": "lr",
+      "type": "number",
+      "description": "learning rate for the optimizer",
+      "required": "False",
+      "default": 0.02
+    },
+    {
+      "name": "schedulerFn",
+      "type": "enum",
+      "description": "Parameters used initialize the optimizer",
+      "options": {
+        "values": [
+          "LambdaLR",
+          "MultiplicativeLR",
+          "StepLR",
+          "MultiStepLR",
+          "ConstantLR",
+          "LinearLR",
+          "ExponentialLR",
+          "PolynomialLR",
+          "CosineAnnealingLR",
+          "ChainedScheduler",
+          "SequentialLR",
+          "CyclicLR",
+          "OneCycleLR",
+          "CosineAnnealingWarmRestarts",
+          "Default"
+        ]
+      },
+      "required": "True"
+    },
+    {
+      "name": "stepSize",
+      "type": "integer",
+      "description": "Parameter to apply to the scheduler_fn",
+      "required": "False",
+      "default": 10
+    },
+    {
+      "name": "deviceName",
+      "type": "enum",
+      "description": "Device used for training",
+      "options": {
+        "values": [
+          "cpu",
+          "gpu",
+          "auto",
+          "DEFAULT"
+        ]
+      },
+      "required": "True"
+    },
+    {
+      "name": "maskType",
+      "type": "enum",
+      "description": "A masking function for feature selection",
+      "options": {
+        "values": [
+          "sparsemax",
+          "entmax",
+          "DEFAULT"
+        ]
+      },
+      "required": "True"
+    },
+    {
+      "name": "groupedFeatures",
+      "type": "integer",
+      "description": "Allow the model to share attention across features within the same group",
+      "required": "False"
+    },
+    {
+      "name": "nSharedDecoder",
+      "type": "integer",
+      "description": "Number of shared GLU block in decoder",
+      "required": "False",
+      "default": 1
+    },
+    {
+      "name": "nIndepDecoder",
+      "type": "integer",
+      "description": "Number of independent GLU block in decoder",
+      "required": "False",
+      "default": 1
+    },
+    {
+      "name": "evalMetric",
+      "type": "enum",
+      "description": "Metrics utilized for early stopping evaluation",
+      "options": {
+        "values": [
+          "auc",
+          "accuracy",
+          "balanced_accuracy",
+          "logloss",
+          "mse",
+          "mae",
+          "rmse",
+          "rmsle",
+          "DEFAULT"
+        ]
+      },
+      "required": "True"
+    },
+    {
+      "name": "maxEpochs",
+      "type": "integer",
+      "description": "Maximum number of epochs for training",
+      "required": "False",
+      "default": 200
+    },
+    {
+      "name": "patience",
+      "type": "integer",
+      "description": "Consecutive epochs without improvement before early stopping",
+      "required": "False",
+      "default": 10
+    },
+    {
+      "name": "weights",
+      "type": "integer",
+      "description": "Sampling parameter only for TabNetClassifier",
+      "required": "False",
+      "default": 10
+    },
+    {
+      "name": "lossFn",
+      "type": "enum",
+      "description": "Loss function",
+      "options": {
+        "values": [
+          "L1Loss",
+          "NLLLoss",
+          "NLLLoss2d",
+          "PoissonNLLLoss",
+          "GaussianNLLLoss",
+          "KLDivLoss",
+          "MSELoss",
+          "BCELoss",
+          "BCEWithLogitsLoss",
+          "HingeEmbeddingLoss",
+          "SmoothL1Loss",
+          "HuberLoss",
+          "SoftMarginLoss",
+          "CrossEntropyLoss",
+          "MultiLabelSoftMarginLoss",
+          "CosineEmbeddingLoss",
+          "MarginRankingLoss",
+          "MultiMarginLoss",
+          "TripletMarginLoss",
+          "TripletMarginWithDistanceLoss",
+          "CTCLoss",
+          "DEFAULT"
+        ]
+      },
+      "required": "True"
+    },
+    {
+      "name": "batchSize",
+      "type": "integer",
+      "description": "Batch size",
+      "required": "False",
+      "default": 1024
+    },
+    {
+      "name": "virtualBatchSize",
+      "type": "integer",
+      "description": "Size of mini-batches for Ghost Batch Normalization",
+      "required": "False",
+      "default": 128
+    },
+    {
+      "name": "numWorkers",
+      "type": "integer",
+      "description": "Number or workers used in torch.utils.data.Dataloader",
+      "required": "False",
+      "default": 0
+    },
+    {
+      "name": "dropLast",
+      "type": "boolean",
+      "description": "Option to drop incomplete last batch during training",
+      "required": "False"
+    },
+    {
+      "name": "warmStart",
+      "type": "boolean",
+      "description": "For scikit-learn compatibility, enabling fitting the same model twice",
+      "required": "False"
+    },
+    {
+      "name": "computeImportance",
+      "type": "boolean",
+      "description": "Compute feature importance?",
+      "required": "False"
+    },
+    {
+      "name": "targetVar",
+      "type": "string",
+      "description": "Target feature for classification",
+      "required": "True"
+    },
+    {
+      "name": "classifier",
+      "type": "enum",
+      "description": "Tabnet Classifier",
+      "options": {
+        "values": [
+          "TabNetClassifier",
+          "TabNetRegressor",
+          "TabNetMultiTaskClassifier",
+          "DEFAULT"
+        ]
+      },
+      "required": "True"
+    },
+    {
+      "name": "preview",
+      "type": "boolean",
+      "description": "Output a JSON preview of outputs produced by this plugin",
+      "required": "False"
+    }
+  ],
+  "outputs": [
+    {
+      "name": "outDir",
+      "description": "Output collection",
+      "type": "genericData",
+      "required": "True"
+    }
+  ],
+  "ui": [
+    {
+      "key": "inputs.inpDir",
+      "type": "genericData",
+      "title": "Input tabular data",
+      "description": "Input tabular data for clustering"
+    },
+    {
+      "key": "inputs.filePattern",
+      "type": "string",
+      "title": "FilePattern",
+      "description": "Pattern to parse input files",
+      "default": ".+"
+    },
+    {
+      "key": "inputs.testSize",
+      "type": "number",
+      "title": "FilePattern",
+      "description": "Proportion of the dataset to include in the test set",
+      "required": "False",
+      "default": 0.2
+    },
+    {
+      "key": "inputs.nD",
+      "type": "integer",
+      "title": "nD",
+      "description": "Width of the decision prediction layer",
+      "required": "False",
+      "default": 8
+    },
+    {
+      "key": "inputs.nA",
+      "type": "integer",
+      "title": "nA",
+      "description": "Width of the attention embedding for each mask",
+      "required": "False",
+      "default": 8
+    },
+    {
+      "key": "inputs.nSteps",
+      "type": "integer",
+      "title": "nSteps",
+      "description": "Number of steps in the architecture",
+      "required": "False",
+      "default": 8
+    },
+    {
+      "key": "inputs.gamma",
+      "type": "number",
+      "title": "gamma",
+      "description": "Coefficient for feature reuse in the masks",
+      "required": "False",
+      "default": 1.3
+    },
+    {
+      "key": "inputs.catEmbDim",
+      "type": "integer",
+      "title": "catEmbDim",
+      "description": "List of embedding sizes for each categorical feature",
+      "required": "False",
+      "default": 1
+    },
+    {
+      "key": "inputs.nIndependent",
+      "type": "integer",
+      "title": "nIndependent",
+      "description": "Number of independent Gated Linear Unit layers at each step",
+      "required": "False",
+      "default": 2
+    },
+    {
+      "key": "inputs.nShared",
+      "type": "integer",
+      "title": "nShared",
+      "description": "Number of shared Gated Linear Unit layers at each step",
+      "required": "False",
+      "default": 2
+    },
+    {
+      "key": "inputs.epsilon",
+      "type": "number",
+      "title": "epsilon",
+      "description": "Constant value",
+      "required": "False",
+      "default": 1e-15
+    },
+    {
+      "key": "inputs.seed",
+      "type": "integer",
+      "title": "seed",
+      "description": "Random seed for reproducibility",
+      "required": "False",
+      "default": 0
+    },
+    {
+      "key": "inputs.momentum",
+      "type": "number",
+      "title": "momentum",
+      "description": "Momentum for batch normalization",
+      "required": "False",
+      "default": 0.02
+    },
+    {
+      "key": "inputs.clipValue",
+      "type": "number",
+      "title": "clipValue",
+      "description": "Clipping of the gradient value",
+      "required": "False"
+    },
+    {
+      "key": "inputs.lambdaSparse",
+      "type": "number",
+      "title": "lambdaSparse",
+      "description": "Extra sparsity loss coefficient",
+      "required": "False",
+      "default": 0.001
+    },
+    {
+      "key": "inputs.optimizerFn",
+      "type": "enum",
+      "title": "optimizerFn",
+      "description": "Pytorch optimizer function",
+      "required": "True",
+      "default": "Adam"
+    },
+    {
+      "key": "inputs.lr",
+      "type": "number",
+      "title": "lr",
+      "description": "learning rate for the optimizer",
+      "required": "False",
+      "default": 0.02
+    },
+    {
+      "key": "inputs.schedulerFn",
+      "type": "enum",
+      "title": "schedulerFn",
+      "description": "Parameters used initialize the optimizer",
+      "required": "True",
+      "default": "StepLR"
+    },
+    {
+      "key": "inputs.stepSize",
+      "type": "integer",
+      "title": "stepSize",
+      "description": "Parameter to apply to the scheduler_fn",
+      "required": "False",
+      "default": 10
+    },
+    {
+      "key": "inputs.deviceName",
+      "type": "enum",
+      "title": "deviceName",
+      "description": "Device used for training",
+      "required": "True",
+      "default": "auto"
+    },
+    {
+      "key": "inputs.maskType",
+      "type": "enum",
+      "title": "maskType",
+      "description": "A masking function for feature selection",
+      "required": "True",
+      "default": "entmax"
+    },
+    {
+      "key": "inputs.groupedFeature",
+      "type": "integer",
+      "title": "groupedFeatures",
+      "description": "Allow the model to share attention across features within the same group",
+      "required": "False"
+    },
+    {
+      "key": "inputs.nSharedDecoder",
+      "type": "integer",
+      "title": "nSharedDecoder",
+      "description": "Number of shared GLU block in decoder",
+      "required": "False",
+      "default": 1
+    },
+    {
+      "key": "inputs.nIndepDecoder",
+      "type": "integer",
+      "title": "nIndepDecoder",
+      "description": "Number of independent GLU block in decoder",
+      "required": "False",
+      "default": 1
+    },
+    {
+      "key": "inputs.evalMetric",
+      "type": "enum",
+      "title": "evalMetric",
+      "description": "Metrics utilized for early stopping evaluation",
+      "required": "True",
+      "default": "auc"
+    },
+    {
+      "key": "inputs.maxEpochs",
+      "type": "integer",
+      "title": "maxEpochs",
+      "description": "Maximum number of epochs for training",
+      "required": "False",
+      "default": 200
+    },
+    {
+      "key": "inputs.patience",
+      "type": "integer",
+      "title": "patience",
+      "description": "Consecutive epochs without improvement before early stopping",
+      "required": "False",
+      "default": 10
+    },
+    {
+      "key": "inputs.weights",
+      "type": "integer",
+      "title": "weights",
+      "description": "Sampling parameter only for TabNetClassifier",
+      "required": "False",
+      "default": 10
+    },
+    {
+      "key": "inputs.lossFn",
+      "type": "enum",
+      "title": "lossFn",
+      "description": "Loss function",
+      "required": "True",
+      "default": "L1Loss"
+    },
+    {
+      "key": "inputs.batchSize",
+      "type": "integer",
+      "title": "batchSize",
+      "description": "Batch size",
+      "required": "False",
+      "default": 1024
+    },
+    {
+      "key": "inputs.virtualBatchSize",
+      "type": "integer",
+      "title": "virtualBatchSize",
+      "description": "Size of mini-batches for Ghost Batch Normalization",
+      "required": "False",
+      "default": 128
+    },
+    {
+      "key": "inputs.numWorkers",
+      "type": "integer",
+      "title": "numWorkers",
+      "description": "Number or workers used in torch.utils.data.Dataloader",
+      "required": "False",
+      "default": 0
+    },
+    {
+      "key": "inputs.dropLast",
+      "type": "boolean",
+      "title": "dropLast",
+      "description": "Option to drop incomplete last batch during training",
+      "required": "False"
+    },
+    {
+      "key": "inputs.warmStart",
+      "type": "boolean",
+      "title": "warmStart",
+      "description": "For scikit-learn compatibility, enabling fitting the same model twice",
+      "required": "False"
+    },
+    {
+      "key": "inputs.computeImportance",
+      "type": "boolean",
+      "title": "computeImportance",
+      "description": "Compute feature importance?",
+      "required": "False"
+    },
+    {
+      "key": "inputs.targetVar",
+      "type": "string",
+      "title": "targetVar",
+      "description": "Target feature for classification",
+      "required": "True"
+    },
+    {
+      "key": "inputs.classifier",
+      "type": "enum",
+      "title": "classifier",
+      "description": "Tabnet Classifier",
+      "required": "True",
+      "default": "TabNetClassifier"
+    },
+    {
+      "key": "inputs.preview",
+      "type": "boolean",
+      "title": "preview",
+      "description": "Output a JSON preview of outputs produced by this plugin",
+      "required": "False"
+    }
+  ]
+}
diff --git a/clustering/tabnet-classifier-tools/pyproject.toml b/clustering/tabnet-classifier-tools/pyproject.toml
new file mode 100644
index 0000000..5e64613
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/pyproject.toml
@@ -0,0 +1,31 @@
+[tool.poetry]
+name = "polus-tabular-clustering-pytorch-tabnet"
+version = "0.1.0-dev0"
+description = "PyTorch TabNet tool"
+authors = [
+"Hamdah Shafqat Abbasi <hamdahshafqat.abbasi@nih.gov>"
+]
+readme = "README.md"
+packages = [{include = "polus", from = "src"}]
+
+[tool.poetry.dependencies]
+python = ">=3.9,<3.12"
+filepattern = "^2.0.4"
+typer = "^0.12.3"
+vaex = "^4.7.0"
+torch = "2.2.2"
+pytorch-tabnet = "^4.1.0"
+scikit-learn = "^1.5.0"
+
+[tool.poetry.group.dev.dependencies]
+bump2version = "^1.0.1"
+pre-commit = "^3.0.4"
+black = "^23.1.0"
+flake8 = "^6.0.0"
+mypy = "^1.0.0"
+pytest = "^7.2.1"
+ipykernel = "^6.29.4"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"
diff --git a/clustering/tabnet-classifier-tools/run-plugin.sh b/clustering/tabnet-classifier-tools/run-plugin.sh
new file mode 100644
index 0000000..3f9e094
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/run-plugin.sh
@@ -0,0 +1,93 @@
+#!/bin/bash
+
+version=$(<VERSION)
+datapath=$(readlink --canonicalize data)
+
+# Inputs
+inp_dir=/data/inputs
+
+# Output paths
+outDir=/data/output
+
+#Additional args
+filePattern=".*.csv"
+testSize=0.2
+nD=8
+nA=8
+nSteps=3
+gamma=1.3
+catEmbDim=1
+nIndependent=2
+nShared=2
+epsilon=1e-15
+seed=0
+momentum=0.02
+clipValue=""
+lambdaSparse=1e-3
+optimizerFn="Adam" 
+lr=0.02
+schedulerFn="StepLR"
+stepSize=10
+deviceName="auto"
+maskType="entmax"
+groupedFeatures=""
+nSharedDecoder=1
+nIndepDecoder=1
+evalMetric="auc"
+maxEpochs=200
+patience=10
+weights=0
+lossFn="L1Loss"
+batchSize=1024
+virtualBatchSize=128
+numWorkers=0
+dropLast="false"
+warmStart="false"
+computeImportance="true"
+classifier="TabNetClassifier"
+targetVar="income"
+
+
+# Show the help options
+# docker run polusai/pytorch-tabnet-tool:${version}
+
+docker run --mount type=bind,source=${datapath},target=/data/ \
+            polusai/pytorch-tabnet-tool:${version} \
+            --inpdir ${inpDir} \
+            --filePattern ${filePattern} \
+            --testSize ${testSize} \
+            --nD ${nD} \
+            --nA ${nA} \
+            --nSteps ${nSteps} \
+            --gamma ${gamma} \
+            --catEmbDim ${catEmbDim} \
+            --nIndependent ${nIndependent} \
+            --nShared ${nShared} \
+            --epsilon ${epsilon} \
+            --seed ${seed} \
+            --momentum ${momentum} \
+            --clipValue ${clipValue} \
+            --lambdaSparse ${lambdaSparse} \
+            --optimizerFn ${optimizerFn} \
+            --lr ${lr} \
+            --schedulerFn ${schedulerFn} \
+            --stepSize ${stepSize} \
+            --deviceName ${deviceName} \
+            --maskType ${maskType} \
+            --groupedFeatures ${groupedFeatures} \
+            --nSharedDecoder ${nSharedDecoder} \
+            --nIndepDecoder ${nIndepDecoder} \
+            --evalMetric ${evalMetric} \
+            --maxEpochs ${maxEpochs} \
+            --patience ${patience} \
+            --weights ${weights} \
+            --lossFn ${lossFn} \
+            --batchSize ${batchSize} \
+            --virtualBatchSize ${virtualBatchSize} \
+            --numWorkers ${numWorkers} \
+            --dropLast ${dropLast} \
+            --warmStart ${warmStart} \
+            --computeImportance ${computeImportance} \
+            --targetVar ${targetVar} \
+            --classifier ${classifier} \
+            --outdir ${outDir} \
diff --git a/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__init__.py b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__init__.py
new file mode 100644
index 0000000..04b4138
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__init__.py
@@ -0,0 +1,3 @@
+"""Pytorch TabNet tool."""
+
+__version__ = "0.1.0-dev0"
diff --git a/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__main__.py b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__main__.py
new file mode 100644
index 0000000..c15b610
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/__main__.py
@@ -0,0 +1,348 @@
+"""Pytorch TabNet tool."""
+
+import logging
+import os
+from pathlib import Path
+from typing import Optional
+from typing import Union
+
+import filepattern as fp
+import polus.tabular.clustering.pytorch_tabnet.tabnet as pt
+import polus.tabular.clustering.pytorch_tabnet.utils as ut
+import typer
+
+# Initialize the logger
+logging.basicConfig(
+    format="%(asctime)s - %(name)-8s - %(levelname)-8s - %(message)s",
+    datefmt="%d-%b-%y %H:%M:%S",
+)
+logger = logging.getLogger("polus.tabular.clustering.pytorch_tabnet")
+logger.setLevel(os.environ.get("POLUS_LOG", logging.INFO))
+POLUS_TAB_EXT = os.environ.get("POLUS_TAB_EXT", ".arrow")
+
+
+app = typer.Typer()
+
+
+@app.command()
+def main(  # noqa: PLR0913 PLR0915
+    inp_dir: Path = typer.Option(
+        ...,
+        "--inpDir",
+        help="Input tabular data",
+    ),
+    file_pattern: str = typer.Option(
+        ".+",
+        "--filePattern",
+        help="pattern to parse tabular files",
+    ),
+    test_size: float = typer.Option(
+        0.2,
+        "--testSize",
+        help="Proportion of the dataset to include in the test set",
+        min=0.1,
+        max=0.4,
+    ),
+    n_d: Optional[Union[int, None]] = typer.Option(
+        8,
+        "--nD",
+        help="Width of the decision prediction layer",
+        min=8,
+        max=64,
+    ),
+    n_a: Optional[Union[int, None]] = typer.Option(
+        8,
+        "--nA",
+        help="Width of the attention embedding for each mask",
+        min=8,
+        max=64,
+    ),
+    n_steps: Optional[Union[int, None]] = typer.Option(
+        3,
+        "--nSteps",
+        help="Number of steps in the architecture",
+        min=3,
+        max=10,
+    ),
+    gamma: Optional[Union[float, None]] = typer.Option(
+        1.3,
+        "--gamma",
+        help="Coefficient for feature reuse in the masks",
+        min=1.0,
+        max=2.0,
+    ),
+    cat_emb_dim: Optional[Union[int, None]] = typer.Option(
+        1,
+        "--catEmbDim",
+        help="List of embedding sizes for each categorical feature",
+    ),
+    n_independent: Optional[Union[int, None]] = typer.Option(
+        2,
+        "--nIndependent",
+        help="Number of independent Gated Linear Unit layers at each step",
+        min=1,
+        max=5,
+    ),
+    n_shared: Optional[Union[int, None]] = typer.Option(
+        2,
+        "--nShared",
+        help="Number of shared Gated Linear Unit layers at each step",
+        min=1,
+        max=5,
+    ),
+    epsilon: Optional[Union[float, None]] = typer.Option(
+        1e-15,
+        "--epsilon",
+        help="Constant value",
+    ),
+    seed: Optional[Union[int, None]] = typer.Option(
+        0,
+        "--seed",
+        help="Random seed for reproducibility",
+    ),
+    momentum: Optional[Union[float, None]] = typer.Option(
+        0.02,
+        "--momentum",
+        help="Momentum for batch normalization",
+        min=0.01,
+        max=0.4,
+    ),
+    clip_value: Optional[Union[float, None]] = typer.Option(
+        None,
+        "--clipValue",
+        help="Clipping of the gradient value",
+    ),
+    lambda_sparse: Optional[Union[float, None]] = typer.Option(
+        1e-3,
+        "--lambdaSparse",
+        help="Extra sparsity loss coefficient",
+    ),
+    optimizer_fn: ut.OptimizersFn = typer.Option(
+        ut.OptimizersFn.Default,
+        "--optimizerFn",
+        help="Pytorch optimizer function",
+    ),
+    lr: Optional[Union[float, None]] = typer.Option(
+        0.02,
+        "--lr",
+        help="learning rate for the optimizer",
+    ),
+    scheduler_fn: ut.SchedulerFn = typer.Option(
+        ut.SchedulerFn.Default,
+        "--schedulerFn",
+        help="Parameters used initialize the optimizer.",
+    ),
+    step_size: int = typer.Option(
+        10,
+        "--stepSize",
+        help="Parameter to apply to the scheduler_fn.",
+    ),
+    device_name: ut.DeviceName = typer.Option(
+        ut.DeviceName.DEFAULT,
+        "--deviceName",
+        help="Device used for training",
+    ),
+    mask_type: ut.MaskType = typer.Option(
+        ut.MaskType.DEFAULT,
+        "--maskType",
+        help="A masking function for feature selection",
+    ),
+    grouped_features: Optional[Union[list[int], None]] = typer.Option(
+        None,
+        "--groupedFeatures",
+        help="Allow the model to share attention across features within the same group",
+    ),
+    n_shared_decoder: Optional[Union[int, None]] = typer.Option(
+        1,
+        "--nSharedDecoder",
+        help="Number of shared GLU block in decoder",
+    ),
+    n_indep_decoder: Optional[Union[int, None]] = typer.Option(
+        1,
+        "--nIndepDecoder",
+        help="Number of independent GLU block in decoder",
+    ),
+    eval_metric: ut.Evalmetric = typer.Option(
+        ut.Evalmetric.DEFAULT,
+        "--evalMetric",
+        help="Metrics utilized for early stopping evaluation",
+    ),
+    max_epochs: Optional[Union[int, None]] = typer.Option(
+        200,
+        "--maxEpochs",
+        help="Maximum number of epochs for training",
+    ),
+    patience: Optional[Union[int, None]] = typer.Option(
+        10,
+        "--patience",
+        help="Consecutive epochs without improvement before early stopping",
+    ),
+    weights: Optional[Union[int, None]] = typer.Option(
+        0,
+        "--weights",
+        help="Sampling parameter only for TabNetClassifier",
+    ),
+    loss_fn: ut.LossFunctions = typer.Option(
+        ut.LossFunctions.DEFAULT,
+        "--lossFn",
+        help="Loss function",
+    ),
+    batch_size: Optional[Union[int, None]] = typer.Option(
+        1024,
+        "--batchSize",
+        help="Batch size",
+    ),
+    virtual_batch_size: Optional[Union[int, None]] = typer.Option(
+        128,
+        "--virtualBatchSize",
+        help="Size of mini-batches for Ghost Batch Normalization.",
+    ),
+    num_workers: Optional[Union[int, None]] = typer.Option(
+        0,
+        "--numWorkers",
+        help="Number or workers used in torch.utils.data.Dataloader",
+    ),
+    drop_last: bool = typer.Option(
+        False,
+        "--dropLast",
+        help="Option to drop incomplete last batch during training",
+    ),
+    warm_start: bool = typer.Option(
+        False,
+        "--warmStart",
+        help="For scikit-learn compatibility, enabling fitting the same model twice",
+    ),
+    compute_importance: bool = typer.Option(
+        True,
+        "--computeImportance",
+        help="Compute feature importance?",
+    ),
+    target_var: str = typer.Option(
+        ...,
+        "--targetVar",
+        help="Target feature for classification",
+    ),
+    classifier: ut.Classifier = typer.Option(
+        ut.Classifier.DEFAULT,
+        "--classifier",
+        help="Tabnet Classifier",
+    ),
+    out_dir: Path = typer.Option(..., "--outDir", help="Output collection"),
+    preview: Optional[bool] = typer.Option(
+        False,
+        "--preview",
+        help="Output a JSON preview of outputs produced by this plugin",
+    ),
+) -> None:
+    """Tool for training tabular data using PyTorch TabNet."""
+    logger.info(f"--inpDir = {inp_dir}")
+    logger.info(f"--filePattern = {file_pattern}")
+    logger.info(f"--testSize = {test_size}")
+    logger.info(f"--nD = {n_d}")
+    logger.info(f"--nA = {n_a}")
+    logger.info(f"--nSteps = {n_steps}")
+    logger.info(f"--gamma = {gamma}")
+    logger.info(f"--catEmbDim = {cat_emb_dim}")
+    logger.info(f"--nIndependent = {n_independent}")
+    logger.info(f"--nShared = {n_shared}")
+    logger.info(f"--epsilon = {epsilon}")
+    logger.info(f"--seed = {seed}")
+    logger.info(f"--momentum = {momentum}")
+    logger.info(f"--clipValue = {clip_value}")
+    logger.info(f"--lambdaSparse = {lambda_sparse}")
+    logger.info(f"--optimizerFn = {optimizer_fn}")
+    logger.info(f"--lr = {lr}")
+    logger.info(f"--schedulerFn = {scheduler_fn}")
+    logger.info(f"--stepSize = {step_size}")
+    logger.info(f"--deviceName = {device_name}")
+    logger.info(f"--maskType = {mask_type}")
+    logger.info(f"--groupedFeatures = {grouped_features}")
+    logger.info(f"--nSharedDecoder = {n_shared_decoder}")
+    logger.info(f"--nIndepDecode = {n_indep_decoder}")
+    logger.info(f"--evalMetric = {eval_metric}")
+    logger.info(f"--maxEpochs = {max_epochs}")
+    logger.info(f"--patience = {patience}")
+    logger.info(f"--weights = {weights}")
+    logger.info(f"--lossFn = {loss_fn}")
+    logger.info(f"--batch_size = {batch_size}")
+    logger.info(f"--virtualBatchSize = {virtual_batch_size}")
+    logger.info(f"--numWorkers = {num_workers}")
+    logger.info(f"--dropLast = {drop_last}")
+    logger.info(f"--warm_start = {warm_start}")
+    logger.info(f"--computeImportance = {compute_importance}")
+    logger.info(f"--targetVar = {target_var}")
+    logger.info(f"--classifier = {classifier}")
+    logger.info(f"--outDir = {out_dir}")
+
+    if not Path(inp_dir).exists():
+        msg = f"The input directory {Path(inp_dir).stem} does not exist."
+        raise FileNotFoundError(msg)
+
+    if not Path(out_dir).exists():
+        Path(out_dir).mkdir(exist_ok=False, parents=True)
+        msg = f"The output directory {out_dir} created."
+        logger.info(msg)
+
+    params = {
+        "test_size": test_size,
+        "n_d": n_d,
+        "n_a": n_a,
+        "n_steps": n_steps,
+        "gamma": gamma,
+        "cat_emb_dim": cat_emb_dim,
+        "n_independent": n_independent,
+        "n_shared": n_shared,
+        "epsilon": epsilon,
+        "seed": seed,
+        "momentum": momentum,
+        "clip_value": clip_value,
+        "lambda_sparse": lambda_sparse,
+        "optimizer_fn": ut.Map_OptimizersFn[optimizer_fn],
+        "optimizer_params": {"lr": lr},
+        "scheduler_fn": ut.Map_SchedulerFn[scheduler_fn],
+        "scheduler_params": {"step_size": step_size, "gamma": 0.95},
+        "device_name": device_name.value,
+        "mask_type": mask_type.value,
+        "grouped_features": grouped_features,
+        "n_shared_decoder": n_shared_decoder,
+        "n_indep_decoder": n_indep_decoder,
+        "eval_metric": eval_metric.value,
+        "max_epochs": max_epochs,
+        "patience": patience,
+        "weights": weights,
+        "loss_fn": loss_fn.value,
+        "batch_size": batch_size,
+        "virtual_batch_size": virtual_batch_size,
+        "num_workers": num_workers,
+        "drop_last": drop_last,
+        "warm_start": warm_start,
+        "compute_importance": compute_importance,
+    }
+
+    fps = fp.FilePattern(inp_dir, file_pattern)
+
+    flist = [f[1][0] for f in fps()]
+
+    if len(flist) == 0:
+        msg = f"No files found with pattern: {file_pattern}."
+        raise ValueError(msg)
+
+    if preview:
+        ut.generate_preview(out_dir)
+
+    if not preview:
+        for file in flist:
+            model = pt.PytorchTabnet(
+                **params,
+                file_path=file,
+                target_var=target_var,
+                classifier=classifier,
+                out_dir=out_dir,
+            )
+            mod_params = dict(model)
+
+            model.fit_model(params=mod_params)
+
+
+if __name__ == "__main__":
+    app()
diff --git a/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/tabnet.py b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/tabnet.py
new file mode 100644
index 0000000..04d4c16
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/tabnet.py
@@ -0,0 +1,241 @@
+"""Pytorch TabNet tool."""
+
+import json
+import logging
+import os
+from pathlib import Path
+from typing import Any
+
+import numpy as np
+import polus.tabular.clustering.pytorch_tabnet.utils as ut
+import vaex
+from scipy.sparse import csr_matrix
+from sklearn.model_selection import train_test_split
+from sklearn.preprocessing import LabelEncoder
+
+from pytorch_tabnet.multitask import TabNetMultiTaskClassifier
+from pytorch_tabnet.tab_model import TabNetClassifier
+from pytorch_tabnet.tab_model import TabNetRegressor
+
+# Initialize the logger
+logger = logging.getLogger(__name__)
+logger.setLevel(os.environ.get("POLUS_LOG", logging.INFO))
+
+
+class PytorchTabnet(ut.TabnetParameters):
+    """Train a Pytorch TabNet model and evaluate it on validation data.
+
+    Args:
+        file_path: Path to the tabular data used for training.
+        target_var: Dataset feature with classification labels.
+        classifier: Select TabNetClassifier,TabNetMultiTaskClassifier,TabNetRegressor.
+        out_dir: Path to the output directory.
+    """
+
+    file_path: Path
+    target_var: str
+    classifier: ut.Classifier
+    out_dir: Path
+
+    @property
+    def convert_vaex_dataframe(self) -> vaex.dataframe.DataFrame:
+        """Vaex supports reading tabular data in .csv, .feather, and .arrow formats."""
+        extensions = [".arrow", ".feather", ".parquet"]
+        if self.file_path.name.endswith(".csv"):
+            return vaex.read_csv(
+                Path(self.file_path),
+                convert=True,
+                chunk_size=5_000_000,
+            )
+        if self.file_path.name.endswith(tuple(extensions)):
+            return vaex.open(Path(self.file_path))
+        return None
+
+    @property
+    def get_data(self) -> ut.MyTupleType:
+        """Subsetting for train/validation,extracting categorical indices/dimensions."""
+        data = self.convert_vaex_dataframe
+
+        if not isinstance(data.shape, tuple) and all(el != 0 for el in data.shape):
+            msg = "Vaex dataframe is empty"
+            raise ValueError(msg)
+
+        if self.target_var not in list(data.columns):
+            msg = f"{self.target_var} does not exist!!"
+            raise ValueError(msg)
+
+        features = [
+            feature for feature in data.get_column_names() if feature != self.target_var
+        ]
+
+        cat_idxs = []
+        cat_dims = []
+        for i, col in enumerate(list(data.columns)):
+            unique_values = 200
+            if data[col].dtype == "string" or len(data[col].unique()) < unique_values:
+                l_enc = LabelEncoder()
+                data[col] = data[col].fillna("fillna")
+                data[col] = l_enc.fit_transform(data[col].values)
+                if col != self.target_var:
+                    cat_idxs.append(i)
+                    cat_dims.append(len(l_enc.classes_))
+            else:
+                # Calculate the mean of the column, ignoring NA values
+                column_mean = data[col].mean()
+                # Replace NA values with the column mean
+                data[col] = data[col].fillna(column_mean)
+
+        if len(cat_idxs) == 0 and len(cat_dims) == 0:
+            cat_idxs = []
+            cat_dims = []
+            logger.info("Categorical features are not dectected")
+
+        features = [
+            feature for feature in data.get_column_names() if feature != self.target_var
+        ]
+
+        x = np.array(data[features])
+        if self.classifier.value in ["TabNetRegressor", "TabNetMultiTaskClassifier"]:
+            y = data[self.target_var].to_numpy().reshape(-1, 1)
+        else:
+            y = data[self.target_var].to_numpy()
+
+        x_train, x_test, y_train, y_test = train_test_split(
+            x,
+            y,
+            test_size=self.test_size,
+            random_state=42,
+            stratify=y,
+        )
+        x_train, x_val, y_train, y_val = train_test_split(
+            x_train,
+            y_train,
+            test_size=self.test_size,
+            random_state=42,
+            stratify=y_train,
+        )
+
+        return (
+            x_train,
+            y_train,
+            x_test,
+            y_test,
+            x_val,
+            y_val,
+            cat_idxs,
+            cat_dims,
+            features,
+        )
+
+    @staticmethod
+    def parameters(params: dict[str, Any]) -> tuple[dict[str, Any], dict[str, Any]]:
+        """Segmenting input parameters for model and evaluation."""
+        exclude_params = [
+            "out_dir",
+            "file_path",
+            "classifier",
+            "test_size",
+            "target_var",
+        ]
+
+        evalparams = [
+            "eval_metric",
+            "max_epochs",
+            "patience",
+            "weights",
+            "loss_fn",
+            "batch_size",
+            "virtual_batch_size",
+            "num_workers",
+            "drop_last",
+            "warm_start",
+            "compute_importance",
+        ]
+
+        params = {
+            k: v for k, v in params.items() if k not in exclude_params if v is not None
+        }
+
+        model_params = {k: v for k, v in params.items() if k not in evalparams}
+        eval_params = {k: v for k, v in params.items() if k in evalparams}
+
+        return model_params, eval_params
+
+    def fit_model(self, params: dict[str, Any]) -> None:
+        """Train a PyTorch tabNet model."""
+        (
+            x_train,
+            y_train,
+            _,
+            _,
+            x_val,
+            y_val,
+            cat_idxs,
+            cat_dims,
+            features,
+        ) = self.get_data
+
+        model_params, eval_params = self.parameters(params)
+
+        model_params["cat_idxs"] = cat_idxs
+        model_params["cat_dims"] = cat_dims
+
+        eval_metric = eval_params["eval_metric"]
+
+        if self.classifier.value == "TabNetClassifier":
+            model = TabNetClassifier(**model_params)
+            if eval_metric not in ut.BINARY_EVAL_METRIC:
+                msg = f"Invalid eval_metric: {eval_metric} for {self.classifier.value}"
+                raise ValueError(msg)
+
+        if self.classifier.value == "TabNetMultiTaskClassifier":
+            model = TabNetMultiTaskClassifier(**model_params)
+            if eval_metric not in ut.MULTICLASS_EVAL_METRIC:
+                msg = f"Invalid eval_metric: {eval_metric} for {self.classifier.value}"
+                raise ValueError(msg)
+
+        if self.classifier.value == "TabNetRegressor":
+            model = TabNetRegressor(**model_params)
+            if eval_metric not in ut.REGRESSION_EVAL_METRIC:
+                msg = f"Invalid eval_metric: {eval_metric} for {self.classifier.value}"
+                raise ValueError(msg)
+
+        # This illustrates the behaviour of the model's fit method using
+        # Compressed Sparse Row matrices
+        sparse_x_train = csr_matrix(x_train)
+        sparse_x_val = csr_matrix(x_val)
+
+        model.fit(
+            X_train=sparse_x_train,
+            y_train=y_train,
+            eval_set=[(x_train, y_train), (sparse_x_val, y_val)],
+            eval_name=["train", "valid"],
+            eval_metric=[eval_params["eval_metric"]],
+            max_epochs=eval_params["max_epochs"],
+            patience=eval_params["patience"],
+            weights=eval_params["weights"],
+            batch_size=eval_params["batch_size"],
+            virtual_batch_size=eval_params["virtual_batch_size"],
+            num_workers=eval_params["num_workers"],
+            drop_last=eval_params["drop_last"],
+            warm_start=eval_params["warm_start"],
+            compute_importance=eval_params["compute_importance"],
+        )
+
+        # save tabnet model
+        model_name = f"tabnet_{Path(self.file_path.name).stem}"
+        model_path = self.out_dir.joinpath(model_name)
+        logger.info("Saving of trained model")
+        model.save_model(model_path)
+
+        imp_features = [round(i, 4) for i in model.feature_importances_]
+
+        feature_importance_pairs = list(zip(features, imp_features))
+        sorted_feature_importance_pairs = dict(
+            sorted(feature_importance_pairs, key=lambda x: x[1], reverse=True),
+        )
+
+        save_feat_path = self.out_dir.joinpath("feature_importances.json")
+        with Path.open(save_feat_path, "w") as jf:
+            logger.info("Save feature importances")
+            json.dump(sorted_feature_importance_pairs, jf, indent=4)
diff --git a/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/utils.py b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/utils.py
new file mode 100644
index 0000000..115b322
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/src/polus/tabular/clustering/pytorch_tabnet/utils.py
@@ -0,0 +1,224 @@
+"""Pytorch TabNet tool."""
+
+import os
+import shutil
+from enum import Enum
+from pathlib import Path
+from typing import Any
+from typing import Optional
+from typing import Union
+
+import numpy as np
+import torch
+from pydantic import BaseModel
+from pydantic import Field
+
+POLUS_TAB_EXT = os.environ.get("POLUS_TAB_EXT", ".arrow")
+
+BINARY_EVAL_METRIC = ["auc", "accuracy", "balanced_accuracy", "logloss"]
+MULTICLASS_EVAL_METRIC = ["accuracy", "balanced_accuracy", "logloss"]
+REGRESSION_EVAL_METRIC = ["mse", "mae", "rmse", "rmsle"]
+
+
+MyTupleType = tuple[
+    np.ndarray,
+    np.array,
+    np.ndarray,
+    np.array,
+    np.ndarray,
+    np.array,
+    list[int],
+    list[int],
+    list[str],
+]
+
+
+def generate_preview(
+    path: Path,
+) -> None:
+    """Generate preview of the plugin outputs."""
+    source_path = Path(__file__).parents[5].joinpath("examples")
+    shutil.copytree(source_path, path, dirs_exist_ok=True)
+
+
+class OptimizersFn(str, Enum):
+    """Optimizers Function."""
+
+    Adadelta = "Adadelta"
+    Adagrad = "Adagrad"
+    Adam = "Adam"
+    AdamW = "AdamW"
+    SparseAdam = "SparseAdam"
+    Adamax = "Adamax"
+    ASGD = "ASGD"
+    LBFGS = "LBFGS"
+    NAdam = "NAdam"
+    RAdam = "RAdam"
+    RMSprop = "RMSprop"
+    Rprop = "Rprop"
+    SGD = "SGD"
+    Default = "Adam"
+
+
+Map_OptimizersFn = {
+    OptimizersFn.Adadelta: torch.optim.Adadelta,
+    OptimizersFn.Adagrad: torch.optim.Adagrad,
+    OptimizersFn.Adam: torch.optim.Adam,
+    OptimizersFn.AdamW: torch.optim.AdamW,
+    OptimizersFn.SparseAdam: torch.optim.SparseAdam,
+    OptimizersFn.Adamax: torch.optim.Adamax,
+    OptimizersFn.ASGD: torch.optim.ASGD,
+    OptimizersFn.LBFGS: torch.optim.LBFGS,
+    OptimizersFn.ASGD: torch.optim.ASGD,
+    OptimizersFn.LBFGS: torch.optim.LBFGS,
+    OptimizersFn.NAdam: torch.optim.NAdam,
+    OptimizersFn.RAdam: torch.optim.RAdam,
+    OptimizersFn.RMSprop: torch.optim.RMSprop,
+    OptimizersFn.Rprop: torch.optim.Rprop,
+    OptimizersFn.SGD: torch.optim.SGD,
+    OptimizersFn.Default: torch.optim.Adam,
+}
+
+
+class SchedulerFn(str, Enum):
+    """Scheduler Function."""
+
+    LambdaLR = "LambdaLR"
+    MultiplicativeLR = "MultiplicativeLR"
+    StepLR = "StepLR"
+    MultiStepLR = "MultiStepLR"
+    ConstantLR = "ConstantLR"
+    LinearLR = "LinearLR"
+    ExponentialLR = "ExponentialLR"
+    PolynomialLR = "PolynomialLR"
+    CosineAnnealingLR = "CosineAnnealingLR"
+    ChainedScheduler = "ChainedScheduler"
+    SequentialLR = "SequentialLR"
+    CyclicLR = "CyclicLR"
+    OneCycleLR = "OneCycleLR"
+    CosineAnnealingWarmRestarts = "CosineAnnealingWarmRestarts"
+    Default = "StepLR"
+
+
+Map_SchedulerFn = {
+    SchedulerFn.LambdaLR: torch.optim.lr_scheduler.LambdaLR,
+    SchedulerFn.MultiplicativeLR: torch.optim.lr_scheduler.MultiplicativeLR,
+    SchedulerFn.StepLR: torch.optim.lr_scheduler.StepLR,
+    SchedulerFn.MultiStepLR: torch.optim.lr_scheduler.MultiStepLR,
+    SchedulerFn.ConstantLR: torch.optim.lr_scheduler.ConstantLR,
+    SchedulerFn.LinearLR: torch.optim.lr_scheduler.LinearLR,
+    SchedulerFn.ExponentialLR: torch.optim.lr_scheduler.ExponentialLR,
+    SchedulerFn.PolynomialLR: torch.optim.lr_scheduler.PolynomialLR,
+    SchedulerFn.CosineAnnealingLR: torch.optim.lr_scheduler.CosineAnnealingLR,
+    SchedulerFn.ChainedScheduler: torch.optim.lr_scheduler.ChainedScheduler,
+    SchedulerFn.SequentialLR: torch.optim.lr_scheduler.SequentialLR,
+    SchedulerFn.CyclicLR: torch.optim.lr_scheduler.CyclicLR,
+    SchedulerFn.OneCycleLR: torch.optim.lr_scheduler.OneCycleLR,
+    SchedulerFn.CosineAnnealingWarmRestarts: torch.optim.lr_scheduler.CosineAnnealingWarmRestarts,  # noqa : E501
+    SchedulerFn.Default: torch.optim.lr_scheduler.StepLR,
+}
+
+
+class Evalmetric(str, Enum):
+    """Evaluation Metric."""
+
+    AUC = "auc"
+    ACCURACY = "accuracy"
+    BALANCEDACCURACY = "balanced_accuracy"
+    LOGLOSS = "logloss"
+    MSE = "mse"
+    MAE = "mae"
+    RMSE = "rmse"
+    RMSLE = "rmsle"
+    DEFAULT = "auc"
+
+
+class MaskType(str, Enum):
+    """Masking Function."""
+
+    SPARSEMAX = "sparsemax"
+    ENTMAX = "entmax"
+    DEFAULT = "entmax"
+
+
+class DeviceName(str, Enum):
+    """Platform Name."""
+
+    CPU = "cpu"
+    GPU = "gpu"
+    AUTO = "auto"
+    DEFAULT = "auto"
+
+
+class Classifier(str, Enum):
+    """Pytorch TabNet Classifier."""
+
+    TabNetClassifier = "TabNetClassifier"
+    TabNetRegressor = "TabNetRegressor"
+    TabNetMultiTaskClassifier = "TabNetMultiTaskClassifier"
+    DEFAULT = "TabNetClassifier"
+
+
+class LossFunctions(str, Enum):
+    """Loss Functions."""
+
+    L1Loss = "L1Loss"
+    NLLLoss = "NLLLoss"
+    NLLLoss2d = "NLLLoss2d"
+    PoissonNLLLoss = "PoissonNLLLoss"
+    GaussianNLLLoss = "GaussianNLLLoss"
+    KLDivLoss = "KLDivLoss"
+    MSELoss = "MSELoss"
+    BCELoss = "BCELoss"
+    BCEWithLogitsLoss = "BCEWithLogitsLoss"
+    HingeEmbeddingLoss = "HingeEmbeddingLoss"
+    SmoothL1Loss = "SmoothL1Loss"
+    HuberLoss = "HuberLoss"
+    SoftMarginLoss = "SoftMarginLoss"
+    CrossEntropyLoss = "CrossEntropyLoss"
+    MultiLabelSoftMarginLoss = "MultiLabelSoftMarginLoss"
+    CosineEmbeddingLoss = "CosineEmbeddingLoss"
+    MarginRankingLoss = "MarginRankingLoss"
+    MultiMarginLoss = "MultiMarginLoss"
+    TripletMarginLoss = "TripletMarginLoss"
+    TripletMarginWithDistanceLoss = "TripletMarginWithDistanceLoss"
+    CTCLoss = "CTCLoss"
+    DEFAULT = "MSELoss"
+
+
+class TabnetParameters(BaseModel):
+    """Parameters for Pytorch TabNet model."""
+
+    test_size: float = Field(default=0.2, ge=0.1, le=0.4)
+    n_d: int = Field(default=8, ge=8, le=64)
+    n_a: int = Field(default=8, ge=8, le=64)
+    n_steps: int = Field(default=3, ge=3, le=10)
+    gamma: float = Field(default=1.3, ge=1.0, le=2.0)
+    cat_emb_dim: int = Field(default=1)
+    n_independent: int = Field(default=2, ge=1, le=5)
+    n_shared: int = Field(default=2, ge=1, le=5)
+    epsilon: float = Field(default=1e-15)
+    seed: int = Field(default=0)
+    momentum: float = Field(default=0.02, ge=0.01, le=0.4)
+    clip_value: Union[float, None] = Field(default=None)
+    lambda_sparse: float = Field(default=1e-3)
+    optimizer_fn: Any = Field(default=torch.optim.Adam)
+    optimizer_params: dict = Field(default={"lr": 0.02})
+    scheduler_fn: Any = Field(default=torch.optim.lr_scheduler.StepLR)
+    scheduler_params: dict = Field(default={"step_size": 10, "gamma": 0.95})
+    device_name: str = Field(default="auto")
+    mask_type: str = Field(default="entmax")
+    grouped_features: Optional[Union[list[int], None]] = Field(default=None)
+    n_shared_decoder: int = Field(default=1)
+    n_indep_decoder: int = Field(default=1)
+    eval_metric: str = Field(default="auc")
+    max_epochs: int = Field(default=200)
+    patience: int = Field(default=10)
+    weights: int = Field(default=0)
+    loss_fn: str = Field(default="MSELoss")
+    batch_size: int = Field(default=1024)
+    virtual_batch_size: int = Field(default=128)
+    num_workers: int = Field(default=0)
+    drop_last: bool = Field(default=False)
+    warm_start: bool = Field(default=False)
+    compute_importance: bool = Field(default=True)
diff --git a/clustering/tabnet-classifier-tools/tests/__init__.py b/clustering/tabnet-classifier-tools/tests/__init__.py
new file mode 100644
index 0000000..4d84872
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/tests/__init__.py
@@ -0,0 +1 @@
+"""Pytorch TabNet tool."""
diff --git a/clustering/tabnet-classifier-tools/tests/conftest.py b/clustering/tabnet-classifier-tools/tests/conftest.py
new file mode 100644
index 0000000..defa87b
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/tests/conftest.py
@@ -0,0 +1,208 @@
+"""Test Fixtures."""
+
+import shutil
+import tempfile
+from pathlib import Path
+from typing import Union
+
+import numpy as np
+import pandas as pd
+import pytest
+
+EXT = None
+
+
+def pytest_addoption(parser: pytest.Parser) -> None:
+    """Add options to pytest."""
+    parser.addoption(
+        "--slow",
+        action="store_true",
+        dest="slow",
+        default=False,
+        help="run slow tests",
+    )
+
+
+def clean_directories() -> None:
+    """Remove all temporary directories."""
+    for d in Path(".").cwd().iterdir():
+        if d.is_dir() and d.name.startswith("tmp"):
+            shutil.rmtree(d)
+
+
+@pytest.fixture()
+def input_directory() -> Union[str, Path]:
+    """Create input directory."""
+    return Path(tempfile.mkdtemp(dir=Path.cwd()))
+
+
+@pytest.fixture()
+def output_directory() -> Union[str, Path]:
+    """Create output directory."""
+    return Path(tempfile.mkdtemp(dir=Path.cwd()))
+
+
+@pytest.fixture()
+def create_dataset() -> Union[str, Path]:
+    """Create output directory."""
+    size = 5000
+
+    inp_dir = Path(tempfile.mkdtemp(dir=Path.cwd()))
+
+    rng = np.random.default_rng()
+
+    workclass = [
+        "Private",
+        "Local-gov",
+        "Self-emp-not-inc",
+        "Federal-gov",
+        "State-gov",
+        "Self-emp-inc",
+        "Without-pay",
+        "Never-worked",
+    ]
+
+    education = [
+        "11th",
+        "HS-grad",
+        "Assoc-acdm",
+        "Some-college",
+        "10th",
+        "Prof-school",
+        "7th-8th",
+        "Bachelors",
+        "Masters",
+        "Doctorate",
+        "5th-6th",
+        "Assoc-voc",
+        "9th",
+        "12th",
+        "1st-4th",
+        "Preschool",
+    ]
+
+    marital_status = [
+        "Never-married",
+        "Married-civ-spouse",
+        "Widowed",
+        "Divorced",
+        "Separated",
+        "Married-spouse-absent",
+        "Married-AF-spouse",
+    ]
+
+    occupation = [
+        "Machine-op-inspct",
+        "Farming-fishing",
+        "Protective-serv",
+        "?",
+        "Other-service",
+        "Prof-specialty",
+        "Craft-repair",
+        "Adm-clerical",
+        "Exec-managerial",
+        "Tech-support",
+        "Sales",
+        "Priv-house-serv",
+        "Transport-moving",
+        "Handlers-cleaners",
+        "Armed-Forces",
+    ]
+
+    relationship = [
+        "Own-child",
+        "Husband",
+        "Not-in-family",
+        "Unmarried",
+        "Wife",
+        "Other-relative",
+    ]
+
+    race = ["Black", "White", "Asian-Pac-Islander", "Other", "Amer-Indian-Eskimo"]
+
+    gender = ["Male", "Female"]
+
+    income = ["<=50K", ">50K"]
+
+    countries = [
+        "United-States",
+        "Peru",
+        "Guatemala",
+        "Mexico",
+        "Dominican-Republic",
+        "Ireland",
+        "Germany",
+        "Philippines",
+        "Thailand",
+        "Haiti",
+        "El-Salvador",
+        "Puerto-Rico",
+        "Vietnam",
+        "South",
+        "Columbia",
+        "Japan",
+        "India",
+        "Cambodia",
+        "Poland",
+        "Laos",
+        "England",
+        "Cuba",
+        "Taiwan",
+        "Italy",
+        "Canada",
+        "Portugal",
+        "China",
+        "Nicaragua",
+        "Honduras",
+        "Iran",
+        "Scotland",
+        "Jamaica",
+        "Ecuador",
+        "Yugoslavia",
+        "Hungary",
+        "Hong",
+        "Greece",
+        "Trinadad&Tobago",
+        "Outlying-US(Guam-USVI-etc)",
+        "France",
+        "Holand-Netherlands",
+    ]
+
+    diction_1 = {
+        "age": rng.integers(low=15, high=80, size=size),
+        "workclass": rng.choice(workclass, size),
+        "fnlwgt": rng.integers(low=12285, high=1490400, size=size),
+        "education": rng.choice(education, size),
+        "educational-num": rng.integers(low=1, high=16, size=size),
+        "marital-status": rng.choice(marital_status, size),
+        "occupation": rng.choice(occupation, size),
+        "relationship": rng.choice(relationship, size),
+        "race": rng.choice(race, size),
+        "gender": rng.choice(gender, size),
+        "capital-gain": rng.integers(low=0, high=99999, size=size),
+        "capital-loss": rng.integers(low=0, high=4356, size=size),
+        "hours-per-week": rng.integers(low=1, high=99, size=size),
+        "native-country": rng.choice(countries, size),
+        "income": rng.choice(income, size),
+    }
+
+    data = pd.DataFrame(diction_1)
+
+    data.to_csv(Path(inp_dir, "adult.csv"), index=False)
+    data.to_feather(Path(inp_dir, "adult.arrow"))
+    data.to_parquet(Path(inp_dir, "adult.parquet"))
+
+    return inp_dir
+
+
+@pytest.fixture(
+    params=[
+        (0.3, "Adam", "StepLR", "accuracy", "L1Loss", "TabNetMultiTaskClassifier"),
+        (0.2, "Adadelta", "StepLR", "mse", "GaussianNLLLoss", "TabNetRegressor"),
+        (0.2, "Adagrad", "StepLR", "logloss", "MSELoss", "TabNetClassifier"),
+        (0.2, "RAdam", "StepLR", "auc", "CrossEntropyLoss", "TabNetClassifier"),
+    ],
+)
+def get_params(request: pytest.FixtureRequest) -> pytest.FixtureRequest:
+    """To get the parameter of the fixture."""
+    return request.param
diff --git a/clustering/tabnet-classifier-tools/tests/test_cli.py b/clustering/tabnet-classifier-tools/tests/test_cli.py
new file mode 100644
index 0000000..a0493e1
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/tests/test_cli.py
@@ -0,0 +1,50 @@
+"""Test Command line Tool."""
+
+from typer.testing import CliRunner
+from pathlib import Path
+import pytest
+from polus.tabular.clustering.pytorch_tabnet.__main__ import app
+from .conftest import clean_directories
+from typing import Union
+
+
+def test_cli(
+    output_directory: Path,
+    create_dataset: Union[str, Path],
+    get_params: pytest.FixtureRequest,
+) -> None:
+    """Test the command line."""
+
+    inp_dir = create_dataset
+
+    runner = CliRunner()
+
+    test_size, optimizer_fn, scheduler_fn, eval_metric, loss_fn, classifier = get_params
+
+    result = runner.invoke(
+        app,
+        [
+            "--inpDir",
+            inp_dir,
+            "--filePattern",
+            ".*.csv",
+            "--testSize",
+            test_size,
+            "--optimizerFn",
+            optimizer_fn,
+            "--evalMetric",
+            eval_metric,
+            "--schedulerFn",
+            scheduler_fn,
+            "--lossFn",
+            loss_fn,
+            "--targetVar",
+            "income",
+            "--classifier",
+            classifier,
+            "--outDir",
+            output_directory,
+        ],
+    )
+    assert result.exit_code == 0
+    clean_directories()
diff --git a/clustering/tabnet-classifier-tools/tests/test_tabnet.py b/clustering/tabnet-classifier-tools/tests/test_tabnet.py
new file mode 100644
index 0000000..0b4cf45
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/tests/test_tabnet.py
@@ -0,0 +1,151 @@
+"""Pytorch TabNet tool."""
+
+import filepattern as fp
+import pytest
+import torch
+import polus.tabular.clustering.pytorch_tabnet.tabnet as tb
+from .conftest import clean_directories
+from pathlib import Path
+from typing import Union
+import numpy as np
+
+
+# @pytest.mark.skipif("not config.getoption('slow')")
+def test_convert_vaex_dataframe(
+    output_directory: Path,
+    create_dataset: Union[str, Path],
+    get_params: pytest.FixtureRequest,
+) -> None:
+    """Testing reading vaex dataframe."""
+
+    inp_dir = create_dataset
+    test_size, optimizer_fn, scheduler_fn, eval_metric, loss_fn, classifier = get_params
+
+    params = {
+        "test_size": test_size,
+        "n_d": 8,
+        "n_a": 8,
+        "seed": 0,
+        "optimizer_fn": optimizer_fn,
+        "optimizer_params": {"lr": 0.001},
+        "scheduler_fn": scheduler_fn,
+        "device_name": "cpu",
+        "eval_metric": eval_metric,
+        "max_epochs": 10,
+        "loss_fn": loss_fn,
+    }
+
+    patterns = [".*.csv", ".*.arrow", ".*.parquet"]
+
+    for pat in patterns:
+        fps = fp.FilePattern(inp_dir, pat)
+
+        for f in fps:
+            model = tb.PytorchTabnet(
+                **params,
+                file_path=f[1][0],
+                target_var="income",
+                classifier=classifier,
+                out_dir=output_directory,
+            )
+            df = model.convert_vaex_dataframe
+            assert df.shape == (5000, 15)
+            assert df is not None
+
+    clean_directories()
+
+
+def test_get_data(
+    output_directory: Path,
+    create_dataset: Union[str, Path],
+    get_params: pytest.FixtureRequest,
+) -> None:
+    """Testing getting data."""
+
+    inp_dir = create_dataset
+    test_size, optimizer_fn, scheduler_fn, eval_metric, loss_fn, classifier = get_params
+
+    params = {
+        "test_size": test_size,
+        "n_d": 8,
+        "n_a": 8,
+        "seed": 0,
+        "optimizer_fn": optimizer_fn,
+        "optimizer_params": {"lr": 0.001},
+        "scheduler_fn": scheduler_fn,
+        "device_name": "cpu",
+        "eval_metric": eval_metric,
+        "max_epochs": 10,
+        "loss_fn": loss_fn,
+    }
+    fps = fp.FilePattern(inp_dir, ".*.csv")
+    file = [f[1][0] for f in fps()][0]
+
+    model = tb.PytorchTabnet(
+        **params,
+        file_path=file,
+        target_var="income",
+        classifier=classifier,
+        out_dir=output_directory,
+    )
+
+    (
+        X_train,
+        y_train,
+        X_test,
+        y_test,
+        X_val,
+        y_val,
+        cat_idxs,
+        cat_dims,
+        features,
+    ) = model.get_data
+
+    assert all(
+        isinstance(arr, np.ndarray)
+        for arr in [X_train, y_train, X_test, y_test, X_val, y_val]
+    )
+    assert all(isinstance(i, list) for i in [cat_idxs, cat_dims, features])
+
+
+def test_fit_model(
+    output_directory: Path,
+    create_dataset: Union[str, Path],
+    get_params: pytest.FixtureRequest,
+) -> None:
+    """Testing fitting model."""
+
+    inp_dir = create_dataset
+    test_size, _, _, eval_metric, loss_fn, classifier = get_params
+
+    params = {
+        "test_size": test_size,
+        "n_d": 8,
+        "n_a": 8,
+        "seed": 0,
+        "optimizer_fn": torch.optim.Adam,
+        "scheduler_fn": torch.optim.lr_scheduler.StepLR,
+        "device_name": "cpu",
+        "eval_metric": eval_metric,
+        "max_epochs": 10,
+        "loss_fn": loss_fn,
+    }
+    fps = fp.FilePattern(inp_dir, ".*.csv")
+    file = [f[1][0] for f in fps()][0]
+
+    model = tb.PytorchTabnet(
+        **params,
+        file_path=file,
+        target_var="income",
+        classifier=classifier,
+        out_dir=output_directory,
+    )
+    mod_params = dict(model)
+    model.fit_model(params=mod_params)
+
+    files = [
+        f for f in Path(output_directory).iterdir() if f.suffix in [".zip", ".json"]
+    ]
+
+    assert len(files) != 0
+    clean_directories()

From 785667fb1762149c610f44a6011c1df199ead149 Mon Sep 17 00:00:00 2001
From: hamshkhawar <hamdahshafqat@hotmail.com>
Date: Wed, 29 May 2024 10:27:43 -0500
Subject: [PATCH 2/5] fix plugin manifest & generated ict and clt

---
 .../tabnet-classifier-tools/PytorchTabnet.cwl | 172 ++++++
 clustering/tabnet-classifier-tools/ict.yaml   | 502 ++++++++++++++++++
 .../tabnet-classifier-tools/plugin.json       |   4 +-
 3 files changed, 676 insertions(+), 2 deletions(-)
 create mode 100644 clustering/tabnet-classifier-tools/PytorchTabnet.cwl
 create mode 100644 clustering/tabnet-classifier-tools/ict.yaml

diff --git a/clustering/tabnet-classifier-tools/PytorchTabnet.cwl b/clustering/tabnet-classifier-tools/PytorchTabnet.cwl
new file mode 100644
index 0000000..96f5fa5
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/PytorchTabnet.cwl
@@ -0,0 +1,172 @@
+class: CommandLineTool
+cwlVersion: v1.2
+inputs:
+  batchSize:
+    inputBinding:
+      prefix: --batchSize
+    type: string?
+  catEmbDim:
+    inputBinding:
+      prefix: --catEmbDim
+    type: string?
+  classifier:
+    inputBinding:
+      prefix: --classifier
+    type: string
+  clipValue:
+    inputBinding:
+      prefix: --clipValue
+    type: double?
+  computeImportance:
+    inputBinding:
+      prefix: --computeImportance
+    type: boolean?
+  deviceName:
+    inputBinding:
+      prefix: --deviceName
+    type: string
+  dropLast:
+    inputBinding:
+      prefix: --dropLast
+    type: boolean?
+  epsilon:
+    inputBinding:
+      prefix: --epsilon
+    type: double?
+  evalMetric:
+    inputBinding:
+      prefix: --evalMetric
+    type: string
+  filePattern:
+    inputBinding:
+      prefix: --filePattern
+    type: string?
+  gamma:
+    inputBinding:
+      prefix: --gamma
+    type: double?
+  groupedFeatures:
+    inputBinding:
+      prefix: --groupedFeatures
+    type: string?
+  inpDir:
+    inputBinding:
+      prefix: --inpDir
+    type: Directory
+  lambdaSparse:
+    inputBinding:
+      prefix: --lambdaSparse
+    type: double?
+  lossFn:
+    inputBinding:
+      prefix: --lossFn
+    type: string
+  lr:
+    inputBinding:
+      prefix: --lr
+    type: double?
+  maskType:
+    inputBinding:
+      prefix: --maskType
+    type: string
+  maxEpochs:
+    inputBinding:
+      prefix: --maxEpochs
+    type: string?
+  momentum:
+    inputBinding:
+      prefix: --momentum
+    type: double?
+  nA:
+    inputBinding:
+      prefix: --nA
+    type: string?
+  nD:
+    inputBinding:
+      prefix: --nD
+    type: string?
+  nIndepDecoder:
+    inputBinding:
+      prefix: --nIndepDecoder
+    type: string?
+  nIndependent:
+    inputBinding:
+      prefix: --nIndependent
+    type: string?
+  nShared:
+    inputBinding:
+      prefix: --nShared
+    type: string?
+  nSharedDecoder:
+    inputBinding:
+      prefix: --nSharedDecoder
+    type: string?
+  nSteps:
+    inputBinding:
+      prefix: --nSteps
+    type: string?
+  numWorkers:
+    inputBinding:
+      prefix: --numWorkers
+    type: string?
+  optimizerFn:
+    inputBinding:
+      prefix: --optimizerFn
+    type: string
+  outDir:
+    inputBinding:
+      prefix: --outDir
+    type: Directory
+  patience:
+    inputBinding:
+      prefix: --patience
+    type: string?
+  preview:
+    inputBinding:
+      prefix: --preview
+    type: boolean?
+  schedulerFn:
+    inputBinding:
+      prefix: --schedulerFn
+    type: string
+  seed:
+    inputBinding:
+      prefix: --seed
+    type: string?
+  stepSize:
+    inputBinding:
+      prefix: --stepSize
+    type: string?
+  targetVar:
+    inputBinding:
+      prefix: --targetVar
+    type: string
+  testSize:
+    inputBinding:
+      prefix: --testSize
+    type: double?
+  virtualBatchSize:
+    inputBinding:
+      prefix: --virtualBatchSize
+    type: string?
+  warmStart:
+    inputBinding:
+      prefix: --warmStart
+    type: boolean?
+  weights:
+    inputBinding:
+      prefix: --weights
+    type: string?
+outputs:
+  outDir:
+    outputBinding:
+      glob: $(inputs.outDir.basename)
+    type: Directory
+requirements:
+  DockerRequirement:
+    dockerPull: polusai/pytorch-tabnet-tool:0.1.0-dev0
+  InitialWorkDirRequirement:
+    listing:
+    - entry: $(inputs.outDir)
+      writable: true
+  InlineJavascriptRequirement: {}
diff --git a/clustering/tabnet-classifier-tools/ict.yaml b/clustering/tabnet-classifier-tools/ict.yaml
new file mode 100644
index 0000000..e7fa7db
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/ict.yaml
@@ -0,0 +1,502 @@
+author:
+- Hamdah Shafqat
+contact: hamdahshafqat.abbasi@nih.gov
+container: polusai/pytorch-tabnet-tool:0.1.0-dev0
+description: Pytorch tabnet tool for training tabular data.
+entrypoint: python3 -m polus.tabular.clustering.pytorch_tabnet
+inputs:
+- description: Input tabular data
+  format:
+  - genericData
+  name: inpDir
+  required: true
+  type: path
+- description: Pattern to parse input files
+  format:
+  - string
+  name: filePattern
+  required: false
+  type: string
+- description: Proportion of the dataset to include in the test set
+  format:
+  - number
+  name: testSize
+  required: false
+  type: number
+- description: Width of the decision prediction layer
+  format:
+  - integer
+  name: nD
+  required: false
+  type: number
+- description: Width of the attention embedding for each mask
+  format:
+  - integer
+  name: nA
+  required: false
+  type: number
+- description: Number of steps in the architecture
+  format:
+  - integer
+  name: nSteps
+  required: false
+  type: number
+- description: Coefficient for feature reuse in the masks
+  format:
+  - number
+  name: gamma
+  required: false
+  type: number
+- description: List of embedding sizes for each categorical feature
+  format:
+  - integer
+  name: catEmbDim
+  required: false
+  type: number
+- description: Number of independent Gated Linear Unit layers at each step
+  format:
+  - integer
+  name: nIndependent
+  required: false
+  type: number
+- description: Number of shared Gated Linear Unit layers at each step
+  format:
+  - integer
+  name: nShared
+  required: false
+  type: number
+- description: Constant value
+  format:
+  - number
+  name: epsilon
+  required: false
+  type: number
+- description: Random seed for reproducibility
+  format:
+  - integer
+  name: seed
+  required: false
+  type: number
+- description: Momentum for batch normalization
+  format:
+  - number
+  name: momentum
+  required: false
+  type: number
+- description: Clipping of the gradient value
+  format:
+  - number
+  name: clipValue
+  required: false
+  type: number
+- description: Extra sparsity loss coefficient
+  format:
+  - number
+  name: lambdaSparse
+  required: false
+  type: number
+- description: Pytorch optimizer function
+  format:
+  - enum
+  name: optimizerFn
+  required: true
+  type: string
+- description: learning rate for the optimizer
+  format:
+  - number
+  name: lr
+  required: false
+  type: number
+- description: Parameters used initialize the optimizer
+  format:
+  - enum
+  name: schedulerFn
+  required: true
+  type: string
+- description: Parameter to apply to the scheduler_fn
+  format:
+  - integer
+  name: stepSize
+  required: false
+  type: number
+- description: Device used for training
+  format:
+  - enum
+  name: deviceName
+  required: true
+  type: string
+- description: A masking function for feature selection
+  format:
+  - enum
+  name: maskType
+  required: true
+  type: string
+- description: Allow the model to share attention across features within the same
+    group
+  format:
+  - integer
+  name: groupedFeatures
+  required: false
+  type: number
+- description: Number of shared GLU block in decoder
+  format:
+  - integer
+  name: nSharedDecoder
+  required: false
+  type: number
+- description: Number of independent GLU block in decoder
+  format:
+  - integer
+  name: nIndepDecoder
+  required: false
+  type: number
+- description: Metrics utilized for early stopping evaluation
+  format:
+  - enum
+  name: evalMetric
+  required: true
+  type: string
+- description: Maximum number of epochs for training
+  format:
+  - integer
+  name: maxEpochs
+  required: false
+  type: number
+- description: Consecutive epochs without improvement before early stopping
+  format:
+  - integer
+  name: patience
+  required: false
+  type: number
+- description: Sampling parameter only for TabNetClassifier
+  format:
+  - integer
+  name: weights
+  required: false
+  type: number
+- description: Loss function
+  format:
+  - enum
+  name: lossFn
+  required: true
+  type: string
+- description: Batch size
+  format:
+  - integer
+  name: batchSize
+  required: false
+  type: number
+- description: Size of mini-batches for Ghost Batch Normalization
+  format:
+  - integer
+  name: virtualBatchSize
+  required: false
+  type: number
+- description: Number or workers used in torch.utils.data.Dataloader
+  format:
+  - integer
+  name: numWorkers
+  required: false
+  type: number
+- description: Option to drop incomplete last batch during training
+  format:
+  - boolean
+  name: dropLast
+  required: false
+  type: boolean
+- description: For scikit-learn compatibility, enabling fitting the same model twice
+  format:
+  - boolean
+  name: warmStart
+  required: false
+  type: boolean
+- description: Compute feature importance?
+  format:
+  - boolean
+  name: computeImportance
+  required: false
+  type: boolean
+- description: Target feature for classification
+  format:
+  - string
+  name: targetVar
+  required: true
+  type: string
+- description: Tabnet Classifier
+  format:
+  - enum
+  name: classifier
+  required: true
+  type: string
+- description: Output a JSON preview of outputs produced by this plugin
+  format:
+  - boolean
+  name: preview
+  required: false
+  type: boolean
+name: polusai/PytorchTabnet
+outputs:
+- description: Output collection
+  format:
+  - genericData
+  name: outDir
+  required: true
+  type: path
+repository: https://github.com/polusai/tabular-tools
+specVersion: 1.0.0
+title: Pytorch Tabnet
+ui:
+- description: Input tabular data for clustering
+  key: inputs.inpDir
+  title: Input tabular data
+  type: path
+- description: Pattern to parse input files
+  key: inputs.filePattern
+  title: FilePattern
+  type: text
+- default: 0.2
+  description: Proportion of the dataset to include in the test set
+  key: inputs.testSize
+  title: FilePattern
+  type: number
+- default: 8
+  description: Width of the decision prediction layer
+  key: inputs.nD
+  title: nD
+  type: number
+- default: 8
+  description: Width of the attention embedding for each mask
+  key: inputs.nA
+  title: nA
+  type: number
+- default: 8
+  description: Number of steps in the architecture
+  key: inputs.nSteps
+  title: nSteps
+  type: number
+- default: 1.3
+  description: Coefficient for feature reuse in the masks
+  key: inputs.gamma
+  title: gamma
+  type: number
+- default: 1
+  description: List of embedding sizes for each categorical feature
+  key: inputs.catEmbDim
+  title: catEmbDim
+  type: number
+- default: 2
+  description: Number of independent Gated Linear Unit layers at each step
+  key: inputs.nIndependent
+  title: nIndependent
+  type: number
+- default: 2
+  description: Number of shared Gated Linear Unit layers at each step
+  key: inputs.nShared
+  title: nShared
+  type: number
+- default: 1.0e-15
+  description: Constant value
+  key: inputs.epsilon
+  title: epsilon
+  type: number
+- default: 0
+  description: Random seed for reproducibility
+  key: inputs.seed
+  title: seed
+  type: number
+- default: 0.02
+  description: Momentum for batch normalization
+  key: inputs.momentum
+  title: momentum
+  type: number
+- description: Clipping of the gradient value
+  key: inputs.clipValue
+  title: clipValue
+  type: number
+- default: 0.001
+  description: Extra sparsity loss coefficient
+  key: inputs.lambdaSparse
+  title: lambdaSparse
+  type: number
+- description: Pytorch optimizer function
+  fields:
+  - Adadelta
+  - Adagrad
+  - Adam
+  - AdamW
+  - SparseAdam
+  - Adamax
+  - ASGD
+  - LBFGS
+  - NAdam
+  - RAdam
+  - RMSprop
+  - Rprop
+  - SGD
+  - Default
+  key: inputs.optimizerFn
+  title: optimizerFn
+  type: select
+- default: 0.02
+  description: learning rate for the optimizer
+  key: inputs.lr
+  title: lr
+  type: number
+- description: Parameters used initialize the optimizer
+  fields:
+  - LambdaLR
+  - MultiplicativeLR
+  - StepLR
+  - MultiStepLR
+  - ConstantLR
+  - LinearLR
+  - ExponentialLR
+  - PolynomialLR
+  - CosineAnnealingLR
+  - ChainedScheduler
+  - SequentialLR
+  - CyclicLR
+  - OneCycleLR
+  - CosineAnnealingWarmRestarts
+  - Default
+  key: inputs.schedulerFn
+  title: schedulerFn
+  type: select
+- default: 10
+  description: Parameter to apply to the scheduler_fn
+  key: inputs.stepSize
+  title: stepSize
+  type: number
+- description: Device used for training
+  fields:
+  - cpu
+  - gpu
+  - auto
+  - DEFAULT
+  key: inputs.deviceName
+  title: deviceName
+  type: select
+- description: A masking function for feature selection
+  fields:
+  - sparsemax
+  - entmax
+  - DEFAULT
+  key: inputs.maskType
+  title: maskType
+  type: select
+- description: Allow the model to share attention across features within the same
+    group
+  key: inputs.groupedFeatures
+  title: groupedFeatures
+  type: number
+- default: 1
+  description: Number of shared GLU block in decoder
+  key: inputs.nSharedDecoder
+  title: nSharedDecoder
+  type: number
+- default: 1
+  description: Number of independent GLU block in decoder
+  key: inputs.nIndepDecoder
+  title: nIndepDecoder
+  type: number
+- description: Metrics utilized for early stopping evaluation
+  fields:
+  - auc
+  - accuracy
+  - balanced_accuracy
+  - logloss
+  - mse
+  - mae
+  - rmse
+  - rmsle
+  - DEFAULT
+  key: inputs.evalMetric
+  title: evalMetric
+  type: select
+- default: 200
+  description: Maximum number of epochs for training
+  key: inputs.maxEpochs
+  title: maxEpochs
+  type: number
+- default: 10
+  description: Consecutive epochs without improvement before early stopping
+  key: inputs.patience
+  title: patience
+  type: number
+- default: 10
+  description: Sampling parameter only for TabNetClassifier
+  key: inputs.weights
+  title: weights
+  type: number
+- description: Loss function
+  fields:
+  - L1Loss
+  - NLLLoss
+  - NLLLoss2d
+  - PoissonNLLLoss
+  - GaussianNLLLoss
+  - KLDivLoss
+  - MSELoss
+  - BCELoss
+  - BCEWithLogitsLoss
+  - HingeEmbeddingLoss
+  - SmoothL1Loss
+  - HuberLoss
+  - SoftMarginLoss
+  - CrossEntropyLoss
+  - MultiLabelSoftMarginLoss
+  - CosineEmbeddingLoss
+  - MarginRankingLoss
+  - MultiMarginLoss
+  - TripletMarginLoss
+  - TripletMarginWithDistanceLoss
+  - CTCLoss
+  - DEFAULT
+  key: inputs.lossFn
+  title: lossFn
+  type: select
+- default: 1024
+  description: Batch size
+  key: inputs.batchSize
+  title: batchSize
+  type: number
+- default: 128
+  description: Size of mini-batches for Ghost Batch Normalization
+  key: inputs.virtualBatchSize
+  title: virtualBatchSize
+  type: number
+- default: 0
+  description: Number or workers used in torch.utils.data.Dataloader
+  key: inputs.numWorkers
+  title: numWorkers
+  type: number
+- description: Option to drop incomplete last batch during training
+  key: inputs.dropLast
+  title: dropLast
+  type: checkbox
+- description: For scikit-learn compatibility, enabling fitting the same model twice
+  key: inputs.warmStart
+  title: warmStart
+  type: checkbox
+- description: Compute feature importance?
+  key: inputs.computeImportance
+  title: computeImportance
+  type: checkbox
+- description: Target feature for classification
+  key: inputs.targetVar
+  title: targetVar
+  type: text
+- description: Tabnet Classifier
+  fields:
+  - TabNetClassifier
+  - TabNetRegressor
+  - TabNetMultiTaskClassifier
+  - DEFAULT
+  key: inputs.classifier
+  title: classifier
+  type: select
+- description: Output a JSON preview of outputs produced by this plugin
+  key: inputs.preview
+  title: preview
+  type: checkbox
+version: 0.1.0-dev0
diff --git a/clustering/tabnet-classifier-tools/plugin.json b/clustering/tabnet-classifier-tools/plugin.json
index 6c65b41..6ad359b 100644
--- a/clustering/tabnet-classifier-tools/plugin.json
+++ b/clustering/tabnet-classifier-tools/plugin.json
@@ -539,7 +539,7 @@
       "default": "entmax"
     },
     {
-      "key": "inputs.groupedFeature",
+      "key": "inputs.groupedFeatures",
       "type": "integer",
       "title": "groupedFeatures",
       "description": "Allow the model to share attention across features within the same group",
@@ -669,4 +669,4 @@
       "required": "False"
     }
   ]
-}
+}
\ No newline at end of file

From 37ea70aa739c6bc706a295f96337527350e56ba5 Mon Sep 17 00:00:00 2001
From: hamshkhawar <hamdahshafqat@hotmail.com>
Date: Wed, 29 May 2024 11:53:23 -0500
Subject: [PATCH 3/5] adding missing file

---
 .../tabnet-classifier-tools/.bumpversion.cfg  | 29 +++++++++++++++++++
 1 file changed, 29 insertions(+)
 create mode 100644 clustering/tabnet-classifier-tools/.bumpversion.cfg

diff --git a/clustering/tabnet-classifier-tools/.bumpversion.cfg b/clustering/tabnet-classifier-tools/.bumpversion.cfg
new file mode 100644
index 0000000..986af44
--- /dev/null
+++ b/clustering/tabnet-classifier-tools/.bumpversion.cfg
@@ -0,0 +1,29 @@
+[bumpversion]
+current_version = 0.1.0-dev0
+commit = True
+tag = False
+parse = (?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(\-(?P<release>[a-z]+)(?P<dev>\d+))?
+serialize = 
+	{major}.{minor}.{patch}-{release}{dev}
+	{major}.{minor}.{patch}
+
+[bumpversion:part:release]
+optional_value = _
+first_value = dev
+values = 
+	dev
+	_
+
+[bumpversion:part:dev]
+
+[bumpversion:file:pyproject.toml]
+search = version = "{current_version}"
+replace = version = "{new_version}"
+
+[bumpversion:file:plugin.json]
+[bumpversion:file:PytorchTabnet.cwl]
+[bumpversion:file:ict.yaml]
+
+[bumpversion:file:VERSION]
+
+[bumpversion:file:src/polus/tabular/clustering/pytorch_tabnet/__init__.py]
\ No newline at end of file

From 50f515ee88c92ef6a0a90d7c7a8b9f18ff507c88 Mon Sep 17 00:00:00 2001
From: hamshkhawar <hamdahshafqat@hotmail.com>
Date: Wed, 29 May 2024 11:59:23 -0500
Subject: [PATCH 4/5] change yaml to yml

---
 clustering/tabnet-classifier-tools/{ict.yaml => ict.yml} | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename clustering/tabnet-classifier-tools/{ict.yaml => ict.yml} (100%)

diff --git a/clustering/tabnet-classifier-tools/ict.yaml b/clustering/tabnet-classifier-tools/ict.yml
similarity index 100%
rename from clustering/tabnet-classifier-tools/ict.yaml
rename to clustering/tabnet-classifier-tools/ict.yml

From 844cf0dbc975850bed21773f2e3f69a33cc17d10 Mon Sep 17 00:00:00 2001
From: hamshkhawar <hamdahshafqat@hotmail.com>
Date: Wed, 29 May 2024 12:25:09 -0500
Subject: [PATCH 5/5] fix feature_importance.json

---
 .../examples/feature_importances.json         | 30 +++++++++----------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/clustering/tabnet-classifier-tools/examples/feature_importances.json b/clustering/tabnet-classifier-tools/examples/feature_importances.json
index 8e56739..6f03300 100644
--- a/clustering/tabnet-classifier-tools/examples/feature_importances.json
+++ b/clustering/tabnet-classifier-tools/examples/feature_importances.json
@@ -1,16 +1,16 @@
 {
-    "capital-gain": 0.2208,
-    "marital-status": 0.1949,
-    "educational-num": 0.1724,
-    "hours-per-week": 0.1041,
-    "age": 0.0789,
-    "gender": 0.0756,
-    "occupation": 0.0635,
-    "relationship": 0.0264,
-    "capital-loss": 0.0225,
-    "native-country": 0.0223,
-    "race": 0.0102,
-    "fnlwgt": 0.0061,
-    "education": 0.0016,
-    "workclass": 0.001
-}
\ No newline at end of file
+  "capital-gain": 0.2208,
+  "marital-status": 0.1949,
+  "educational-num": 0.1724,
+  "hours-per-week": 0.1041,
+  "age": 0.0789,
+  "gender": 0.0756,
+  "occupation": 0.0635,
+  "relationship": 0.0264,
+  "capital-loss": 0.0225,
+  "native-country": 0.0223,
+  "race": 0.0102,
+  "fnlwgt": 0.0061,
+  "education": 0.0016,
+  "workclass": 0.001
+}