From 4ef9bfc6c1b93538578670e26c95b2f687e24382 Mon Sep 17 00:00:00 2001
From: Griffin Bassman <griffinbassman@gmail.com>
Date: Thu, 9 May 2024 13:55:59 -0400
Subject: [PATCH 1/3] fix: set mac CI to version 13 (#4691)

* fix: update images for linux CIs

* try to trigger on PR

* fix triggers

* try to force run

* try deleting pipelines

* readd azure devops

* test

* test

* mac-12

* mac13
---
 .github/workflows/asan.yml          | 4 ++--
 .github/workflows/dotnet_nugets.yml | 4 ++--
 .github/workflows/vcpkg_build.yml   | 2 +-
 .vscode/settings.json               | 5 ++++-
 4 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/asan.yml b/.github/workflows/asan.yml
index 5571eb8af37..2165e9f65fe 100644
--- a/.github/workflows/asan.yml
+++ b/.github/workflows/asan.yml
@@ -21,8 +21,8 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        #os: [windows-latest, ubuntu-latest, macos-latest]
-        os: [ubuntu-latest, macos-latest] # Temporarily remove windows asan
+        #os: [windows-latest, ubuntu-latest, macos-13]
+        os: [ubuntu-latest, macos-13] # Temporarily remove windows asan
         preset: [vcpkg-asan-debug, vcpkg-ubsan-debug]
         exclude:
           # UBSan not supported by MSVC on Windows
diff --git a/.github/workflows/dotnet_nugets.yml b/.github/workflows/dotnet_nugets.yml
index 3beaa1821f0..8b2466fe1e4 100644
--- a/.github/workflows/dotnet_nugets.yml
+++ b/.github/workflows/dotnet_nugets.yml
@@ -23,7 +23,7 @@ jobs:
         config:
           - { os: "windows-latest", runtime_id: "win-x64" }
           - { os: "ubuntu-latest", runtime_id: "linux-x64" }
-          - { os: "macos-latest", runtime_id: "osx-x64" }
+          - { os: "macos-13", runtime_id: "osx-x64" }
     runs-on: ${{matrix.config.os}}
     steps:
       # Setup for build
@@ -143,7 +143,7 @@ jobs:
         config:
           - { os: "windows-latest", runtime_id: "win-x64" }
           - { os: "ubuntu-latest", runtime_id: "linux-x64" }
-          - { os: "macos-latest", runtime_id: "osx-x64" }
+          - { os: "macos-13", runtime_id: "osx-x64" }
     runs-on: ${{matrix.config.os}}
     steps:
       - uses: actions/checkout@v3
diff --git a/.github/workflows/vcpkg_build.yml b/.github/workflows/vcpkg_build.yml
index 629b2ff11c9..9a763817d60 100644
--- a/.github/workflows/vcpkg_build.yml
+++ b/.github/workflows/vcpkg_build.yml
@@ -23,7 +23,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-latest, macos-latest, windows-latest]
+        os: [ubuntu-latest, macos-13, windows-latest]
         preset: [vcpkg-debug, vcpkg-release]
     steps:
       - uses: actions/checkout@v3
diff --git a/.vscode/settings.json b/.vscode/settings.json
index f7a94e718de..9844de0b25c 100644
--- a/.vscode/settings.json
+++ b/.vscode/settings.json
@@ -6,5 +6,8 @@
             ],
             "url": "./test/vwtest.schema.json"
         }
-    ]
+    ],
+    "files.associations": {
+        "limits": "cpp"
+    }
 }

From 5c77d729d04b0efdf5ca8af1d90c67fba285756e Mon Sep 17 00:00:00 2001
From: Byron Xu <byronxu@microsoft.com>
Date: Thu, 9 May 2024 15:06:28 -0400
Subject: [PATCH 2/3] fix: VW Slim (#4674)

* handle feature scale correctly in VW slim

* fix namespace copy bug

* update tests to work with current VW

* update VW slim CI job

* fix typo in script name

* fix CI errors

* Update build_vw_slim.yml

---------

Co-authored-by: Jack Gerrits <jackgerrits@users.noreply.github.com>
Co-authored-by: Jacob Alber <jacob.alber@microsoft.com>
Co-authored-by: olgavrou <olgavrou@gmail.com>
Co-authored-by: Griffin Bassman <griffinbassman@gmail.com>
---
 .github/workflows/build_vw_slim.yml           |  39 +-
 .scripts/linux/build-slim.sh                  |  19 -
 .../slim/include/vw/slim/vw_slim_predict.h    |  48 +-
 vowpalwabbit/slim/src/vw_slim_predict.cc      |  36 +-
 vowpalwabbit/slim/test/data.h                 | 484 +++++++-----------
 vowpalwabbit/slim/test/data/cb_data_5.model   | Bin 271 -> 236 bytes
 vowpalwabbit/slim/test/data/cb_data_5.pred    |   4 +-
 vowpalwabbit/slim/test/data/cb_data_6.model   | Bin 268 -> 230 bytes
 vowpalwabbit/slim/test/data/cb_data_6.pred    |   4 +-
 vowpalwabbit/slim/test/data/cb_data_7.model   | Bin 361 -> 346 bytes
 vowpalwabbit/slim/test/data/cb_data_7.pred    |   4 +-
 vowpalwabbit/slim/test/data/cb_data_8.model   | Bin 472 -> 471 bytes
 vowpalwabbit/slim/test/data/cb_data_8.pred    |   4 +-
 vowpalwabbit/slim/test/data/cb_data_9.model   | Bin 286 -> 283 bytes
 vowpalwabbit/slim/test/data/generate-data.sh  |  37 +-
 .../slim/test/data/multiclass_data_4.model    | Bin 174 -> 144 bytes
 .../slim/test/data/multiclass_data_4.pred     |   2 +-
 .../slim/test/data/multiclass_data_5.pred     |   2 +-
 .../slim/test/data/regression_data_1.model    | Bin 100 -> 70 bytes
 .../slim/test/data/regression_data_1.pred     |   4 +-
 .../slim/test/data/regression_data_2.model    | Bin 116 -> 86 bytes
 .../slim/test/data/regression_data_3.model    | Bin 131 -> 101 bytes
 .../slim/test/data/regression_data_3.pred     |   4 +-
 .../slim/test/data/regression_data_4.model    | Bin 159 -> 129 bytes
 .../slim/test/data/regression_data_4.pred     |   4 +-
 .../slim/test/data/regression_data_5.model    | Bin 168 -> 138 bytes
 .../slim/test/data/regression_data_5.pred     |   2 +-
 .../slim/test/data/regression_data_6.model    | Bin 147 -> 117 bytes
 .../slim/test/data/regression_data_6.pred     |   4 +-
 .../slim/test/data/regression_data_7.model    | Bin 116 -> 86 bytes
 .../slim/test/data/regression_data_7.pred     |   4 +-
 .../data/regression_data_ignore_linear.model  | Bin 144 -> 97 bytes
 .../data/regression_data_ignore_linear.pred   |   2 +-
 .../data/regression_data_no_constant.model    | Bin 92 -> 75 bytes
 .../data/regression_data_no_constant.pred     |   4 +-
 vowpalwabbit/slim/test/ut_util.cc             | 150 ++----
 36 files changed, 356 insertions(+), 505 deletions(-)
 delete mode 100755 .scripts/linux/build-slim.sh
 mode change 100644 => 100755 vowpalwabbit/slim/test/data/generate-data.sh

diff --git a/.github/workflows/build_vw_slim.yml b/.github/workflows/build_vw_slim.yml
index 3a8f8371340..9acaf124411 100644
--- a/.github/workflows/build_vw_slim.yml
+++ b/.github/workflows/build_vw_slim.yml
@@ -22,10 +22,45 @@ jobs:
       - uses: actions/checkout@v1
         with:
           submodules: recursive
-      - name: Build VW Slim
+
+      - name: Install dependencies
+        shell: bash
+        run: |
+          sudo apt update
+          sudo apt install -y xxd
+
+      - name: Configure VW Slim
         shell: bash
-        run: ./.scripts/linux/build-slim.sh
+        run: |
+          rm -rf build
+          cmake -S . -B build -G Ninja \
+              -DBUILD_TESTING=On \
+              -DVW_FEAT_FLATBUFFERS=Off \
+              -DRAPIDJSON_SYS_DEP=Off \
+              -DFMT_SYS_DEP=Off \
+              -DSPDLOG_SYS_DEP=Off \
+              -DVW_ZLIB_SYS_DEP=Off \
+              -DVW_BOOST_MATH_SYS_DEP=Off
+
+      - name: Build VW and VW Slim
+        shell: bash
+        run: cmake --build build --target vw_cli_bin vw_slim vw_slim_test
+
       - name: Test VW Slim
         shell: bash
         working-directory: build
         run: ctest --output-on-failure --no-tests=error --tests-regex "VowpalWabbitSlim|ExploreSlim|CommandLineOptionsSlim" --parallel 2
+
+      - name: Generate test data with new VW executable
+        shell: bash
+        working-directory: vowpalwabbit/slim/test/data/
+        run: ./generate-data.sh ../../../../build/vowpalwabbit/cli/vw
+
+      - name: Build VW Slim again
+        shell: bash
+        run: cmake --build build --target vw_slim vw_slim_test
+
+      - name: Test VW Slim again
+        shell: bash
+        working-directory: build
+        run: ctest --output-on-failure --no-tests=error --tests-regex "VowpalWabbitSlim|ExploreSlim|CommandLineOptionsSlim" --parallel 2
diff --git a/.scripts/linux/build-slim.sh b/.scripts/linux/build-slim.sh
deleted file mode 100755
index 04cc7d3f399..00000000000
--- a/.scripts/linux/build-slim.sh
+++ /dev/null
@@ -1,19 +0,0 @@
-#!/bin/bash
-set -e
-set -x
-
-SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
-REPO_DIR=$SCRIPT_DIR/../../
-cd $REPO_DIR
-
-rm -rf build
-cmake -S . -B build -G Ninja \
-    -DBUILD_TESTING=On \
-    -DVW_FEAT_FLATBUFFERS=Off \
-    -DRAPIDJSON_SYS_DEP=Off \
-    -DFMT_SYS_DEP=Off \
-    -DSPDLOG_SYS_DEP=Off \
-    -DVW_ZLIB_SYS_DEP=Off \
-    -DVW_BOOST_MATH_SYS_DEP=Off
-
-cmake --build build --target vw_slim vw_slim_test
diff --git a/vowpalwabbit/slim/include/vw/slim/vw_slim_predict.h b/vowpalwabbit/slim/include/vw/slim/vw_slim_predict.h
index 0d58099c214..6b8f2bc3574 100644
--- a/vowpalwabbit/slim/include/vw/slim/vw_slim_predict.h
+++ b/vowpalwabbit/slim/include/vw/slim/vw_slim_predict.h
@@ -44,31 +44,32 @@ class namespace_copy_guard
   VW::example_predict& _ex;
   unsigned char _ns;
   bool _remove_ns;
+  size_t _old_size;
 };
 
 class feature_offset_guard
 {
 public:
-  feature_offset_guard(VW::example_predict& ex, uint64_t ft_offset);
+  feature_offset_guard(VW::example_predict& ex, uint64_t ft_index_offset);
   ~feature_offset_guard();
 
 private:
   VW::example_predict& _ex;
-  uint64_t _old_ft_offset;
+  uint64_t _old_ft_index_offset;
 };
 
-class stride_shift_guard
+class feature_scale_guard
 {
 public:
-  stride_shift_guard(VW::example_predict& ex, uint64_t shift);
-  ~stride_shift_guard();
+  feature_scale_guard(VW::example_predict& ex, uint64_t ft_index_scale);
+  ~feature_scale_guard();
 
 private:
   VW::example_predict& _ex;
-  uint64_t _shift;
+  uint64_t _ft_index_scale;
 };
 
-/**
+/*
  * @brief Vowpal Wabbit slim predictor. Supports: regression, multi-class classification and contextual bandits.
  */
 template <typename W>
@@ -152,7 +153,7 @@ class vw_predict
     }
 
     // TODO: take --cb_type dr into account
-    uint64_t num_weights = 0;
+    uint64_t feature_scale = 0;
 
     if (_command_line_arguments.find("--cb_explore_adf") != std::string::npos)
     {
@@ -164,7 +165,7 @@ class vw_predict
         _bag_size = static_cast<size_t>(bag_size);
 
         _exploration = vw_predict_exploration::bag;
-        num_weights = _bag_size;
+        feature_scale = _bag_size;
 
         // check for additional minimum epsilon greedy
         _minimum_epsilon = 0.f;
@@ -212,10 +213,10 @@ class vw_predict
     RETURN_ON_FAIL(mp.read("resume", gd_resume));
     if (gd_resume) { return E_VW_PREDICT_ERR_GD_RESUME_NOT_SUPPORTED; }
 
-    // read sparse weights into dense
-    _stride_shift = (uint32_t)ceil_log_2(num_weights);
+    _feature_scale_bits = (uint32_t)ceil_log_2(feature_scale);
 
-    RETURN_ON_FAIL(mp.read_weights<W>(_weights, _num_bits, _stride_shift));
+    // stride shift always 0 bits
+    RETURN_ON_FAIL(mp.read_weights<W>(_weights, _num_bits, 0));
 
     // TODO: check that permutations is not enabled (or parse it)
 
@@ -261,7 +262,7 @@ class vw_predict
       // add constant feature
       ns_copy_guard =
           std::unique_ptr<namespace_copy_guard>(new namespace_copy_guard(ex, VW::details::CONSTANT_NAMESPACE));
-      ns_copy_guard->feature_push_back(1.f, (VW::details::CONSTANT << _stride_shift) + ex.ft_offset);
+      ns_copy_guard->feature_push_back(1.f, (VW::details::CONSTANT << _feature_scale_bits) + ex.ft_offset);
     }
 
     if (_contains_wildcard)
@@ -291,9 +292,9 @@ class vw_predict
 
     out_scores.resize(num_actions);
 
-    VW::example_predict* action = actions;
-    for (size_t i = 0; i < num_actions; i++, action++)
+    for (size_t i = 0; i < num_actions; i++)
     {
+      VW::example_predict* action = &actions[i];
       std::vector<std::unique_ptr<namespace_copy_guard>> ns_copy_guards;
 
       // shared feature copying
@@ -358,19 +359,21 @@ class vw_predict
       {
         std::vector<uint32_t> top_actions(num_actions);
 
-        // apply stride shifts
-        std::vector<std::unique_ptr<stride_shift_guard>> stride_shift_guards;
-        stride_shift_guards.push_back(
-            std::unique_ptr<stride_shift_guard>(new stride_shift_guard(shared, _stride_shift)));
+        // apply feature scale
+        uint64_t feature_scale = static_cast<uint64_t>(1) << _feature_scale_bits;
+        std::vector<std::unique_ptr<feature_scale_guard>> feature_scale_guards;
+        feature_scale_guards.push_back(
+            std::unique_ptr<feature_scale_guard>(new feature_scale_guard(shared, feature_scale)));
         VW::example_predict* actions_end = actions + num_actions;
         for (VW::example_predict* action = actions; action != actions_end; ++action)
         {
-          stride_shift_guards.push_back(
-              std::unique_ptr<stride_shift_guard>(new stride_shift_guard(*action, _stride_shift)));
+          feature_scale_guards.push_back(
+              std::unique_ptr<feature_scale_guard>(new feature_scale_guard(*action, feature_scale)));
         }
 
         for (size_t i = 0; i < _bag_size; i++)
         {
+          // apply feature offset
           std::vector<std::unique_ptr<feature_offset_guard>> feature_offset_guards;
           for (VW::example_predict* action = actions; action != actions_end; ++action)
           {
@@ -487,7 +490,8 @@ class vw_predict
   size_t _bag_size;
   uint32_t _num_bits;
 
-  uint32_t _stride_shift;
+  // log2 of feature scale, rounded upwards to next integer
+  uint32_t _feature_scale_bits;
   bool _model_loaded;
 };
 }  // namespace vw_slim
diff --git a/vowpalwabbit/slim/src/vw_slim_predict.cc b/vowpalwabbit/slim/src/vw_slim_predict.cc
index 5c094149fe1..dda0deb3513 100644
--- a/vowpalwabbit/slim/src/vw_slim_predict.cc
+++ b/vowpalwabbit/slim/src/vw_slim_predict.cc
@@ -17,14 +17,23 @@ namespace_copy_guard::namespace_copy_guard(VW::example_predict& ex, unsigned cha
   {
     _ex.indices.push_back(_ns);
     _remove_ns = true;
+    _old_size = 0;
+  }
+  else
+  {
+    _remove_ns = false;
+    _old_size = _ex.feature_space[_ns].size();
   }
-  else { _remove_ns = false; }
 }
 
 namespace_copy_guard::~namespace_copy_guard()
 {
-  _ex.indices.pop_back();
-  if (_remove_ns) { _ex.feature_space[_ns].clear(); }
+  if (_remove_ns)
+  {
+    _ex.feature_space[_ns].clear();
+    _ex.indices.pop_back();
+  }
+  else { _ex.feature_space[_ns].truncate_to(_old_size); }
 }
 
 void namespace_copy_guard::feature_push_back(VW::feature_value v, VW::feature_index idx)
@@ -32,32 +41,33 @@ void namespace_copy_guard::feature_push_back(VW::feature_value v, VW::feature_in
   _ex.feature_space[_ns].push_back(v, idx);
 }
 
-feature_offset_guard::feature_offset_guard(VW::example_predict& ex, uint64_t ft_offset)
-    : _ex(ex), _old_ft_offset(ex.ft_offset)
+feature_offset_guard::feature_offset_guard(VW::example_predict& ex, uint64_t ft_index_offset)
+    : _ex(ex), _old_ft_index_offset(ex.ft_offset)
 {
-  _ex.ft_offset = ft_offset;
+  _ex.ft_offset = ft_index_offset;
 }
 
-feature_offset_guard::~feature_offset_guard() { _ex.ft_offset = _old_ft_offset; }
+feature_offset_guard::~feature_offset_guard() { _ex.ft_offset = _old_ft_index_offset; }
 
-stride_shift_guard::stride_shift_guard(VW::example_predict& ex, uint64_t shift) : _ex(ex), _shift(shift)
+feature_scale_guard::feature_scale_guard(VW::example_predict& ex, uint64_t ft_index_scale)
+    : _ex(ex), _ft_index_scale(ft_index_scale)
 {
-  if (_shift > 0)
+  if (_ft_index_scale > 1)
   {
     for (auto ns : _ex.indices)
     {
-      for (auto& f : _ex.feature_space[ns]) { f.index() <<= _shift; }
+      for (auto& f : _ex.feature_space[ns]) { f.index() *= _ft_index_scale; }
     }
   }
 }
 
-stride_shift_guard::~stride_shift_guard()
+feature_scale_guard::~feature_scale_guard()
 {
-  if (_shift > 0)
+  if (_ft_index_scale > 1)
   {
     for (auto ns : _ex.indices)
     {
-      for (auto& f : _ex.feature_space[ns]) { f.index() >>= _shift; }
+      for (auto& f : _ex.feature_space[ns]) { f.index() /= _ft_index_scale; }
     }
   }
 }
diff --git a/vowpalwabbit/slim/test/data.h b/vowpalwabbit/slim/test/data.h
index 5011d0ca5e9..5f598c0df43 100644
--- a/vowpalwabbit/slim/test/data.h
+++ b/vowpalwabbit/slim/test/data.h
@@ -1,330 +1,228 @@
 
-unsigned char regression_data_1_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+unsigned char regression_data_1_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1f, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e,
-    0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x4a, 0x0d, 0xa8, 0x7d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x84,
-    0x94, 0x7b, 0xbe, 0x5c, 0xc5, 0x01, 0x00, 0x55, 0xea, 0x9d, 0x3f};
-unsigned int regression_data_1_model_len = 100;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x49, 0x8e,
+    0xa9, 0xc7, 0x00, 0x00, 0x00, 0x00, 0x00, 0x4b, 0x94, 0x7b, 0xbe, 0x5c, 0xc5, 0x01, 0x00, 0x3c, 0xea, 0x9d, 0x3f};
+unsigned int regression_data_1_model_len = 70;
 unsigned char regression_data_1_pred[] = {
-    0x30, 0x2e, 0x39, 0x38, 0x38, 0x30, 0x33, 0x30, 0x0a, 0x30, 0x2e, 0x30, 0x30, 0x35, 0x32, 0x39, 0x35, 0x0a};
+    0x30, 0x2e, 0x39, 0x38, 0x38, 0x30, 0x32, 0x38, 0x0a, 0x30, 0x2e, 0x30, 0x30, 0x35, 0x32, 0x39, 0x36, 0x0a};
 unsigned int regression_data_1_pred_len = 18;
-unsigned char regression_data_no_constant_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01,
+unsigned char regression_data_no_constant_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01,
     0x00, 0x00, 0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1f, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61,
-    0x73, 0x68, 0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64,
-    0x65, 0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x4a, 0x0d, 0xa8, 0x7d, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x8c, 0x1e, 0x0d, 0x3d};
-unsigned int regression_data_no_constant_model_len = 92;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0e, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x6e, 0x6f,
+    0x63, 0x6f, 0x6e, 0x73, 0x74, 0x61, 0x6e, 0x74, 0x00, 0x04, 0x00, 0x00, 0x00, 0xf1, 0x40, 0x85, 0xfb, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0xe6, 0x1f, 0x0d, 0x3d};
+unsigned int regression_data_no_constant_model_len = 75;
 unsigned char regression_data_no_constant_pred[] = {
-    0x30, 0x2e, 0x30, 0x33, 0x34, 0x34, 0x35, 0x33, 0x0a, 0x30, 0x2e, 0x31, 0x37, 0x32, 0x32, 0x36, 0x35, 0x0a};
+    0x30, 0x2e, 0x30, 0x33, 0x34, 0x34, 0x35, 0x34, 0x0a, 0x30, 0x2e, 0x31, 0x37, 0x32, 0x32, 0x37, 0x31, 0x0a};
 unsigned int regression_data_no_constant_pred_len = 18;
-unsigned char regression_data_ignore_linear_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01,
+unsigned char regression_data_ignore_linear_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01,
     0x00, 0x00, 0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x43, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61,
-    0x73, 0x68, 0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x69, 0x67, 0x6e, 0x6f, 0x72, 0x65, 0x5f,
-    0x6c, 0x69, 0x6e, 0x65, 0x61, 0x72, 0x20, 0x61, 0x20, 0x2d, 0x2d, 0x69, 0x67, 0x6e, 0x6f, 0x72, 0x65, 0x5f, 0x6c,
-    0x69, 0x6e, 0x65, 0x61, 0x72, 0x20, 0x62, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e,
-    0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0xec, 0x9d, 0x02, 0x9b, 0x00, 0x00, 0x00, 0x00, 0x00, 0xa7,
-    0x93, 0x0f, 0x3f, 0x5c, 0xc5, 0x01, 0x00, 0xb0, 0xd8, 0xe0, 0x3e, 0x5f, 0xd6, 0x02, 0x00, 0xca, 0xe5, 0x15, 0xbe};
-unsigned int regression_data_ignore_linear_model_len = 144;
-unsigned char regression_data_ignore_linear_pred[] = {
-    0x31, 0x2e, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x0a, 0x30, 0x2e, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x0a};
-unsigned int regression_data_ignore_linear_pred_len = 18;
-unsigned char regression_data_2_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x69, 0x67,
+    0x6e, 0x6f, 0x72, 0x65, 0x5f, 0x6c, 0x69, 0x6e, 0x65, 0x61, 0x72, 0x20, 0x61, 0x62, 0x00, 0x04, 0x00, 0x00, 0x00,
+    0x17, 0xa5, 0x23, 0x26, 0x00, 0x00, 0x00, 0x00, 0x00, 0x29, 0x94, 0x0f, 0x3f, 0x5c, 0xc5, 0x01, 0x00, 0xac, 0xd7,
+    0xe0, 0x3e, 0x5f, 0xd6, 0x02, 0x00, 0x1d, 0xe5, 0x15, 0xbe};
+unsigned int regression_data_ignore_linear_model_len = 97;
+unsigned char regression_data_ignore_linear_pred[] = {0x31, 0x2e, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x0a, 0x30, 0x0a};
+unsigned int regression_data_ignore_linear_pred_len = 11;
+unsigned char regression_data_2_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1f, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e,
-    0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x4a, 0x0d, 0xa8, 0x7d, 0x00, 0x00, 0x00, 0x00, 0x00, 0xcf,
-    0xee, 0xb4, 0x3e, 0xb2, 0x69, 0x01, 0x00, 0xcf, 0xee, 0x34, 0x3e, 0x5c, 0xc5, 0x01, 0x00, 0x60, 0x22, 0x96, 0x3e,
-    0x5f, 0xd6, 0x02, 0x00, 0xd5, 0x2d, 0xc8, 0xbd};
-unsigned int regression_data_2_model_len = 116;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x49, 0x8e,
+    0xa9, 0xc7, 0x00, 0x00, 0x00, 0x00, 0x00, 0x4c, 0xef, 0xb4, 0x3e, 0xb2, 0x69, 0x01, 0x00, 0x4c, 0xef, 0x34, 0x3e,
+    0x5c, 0xc5, 0x01, 0x00, 0x66, 0x21, 0x96, 0x3e, 0x5f, 0xd6, 0x02, 0x00, 0x88, 0x2c, 0xc8, 0xbd};
+unsigned int regression_data_2_model_len = 86;
 unsigned char regression_data_2_pred[] = {0x31, 0x2e, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x0a, 0x30, 0x0a};
 unsigned int regression_data_2_pred_len = 11;
-unsigned char regression_data_3_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+unsigned char regression_data_3_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2e, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63,
-    0x20, 0x61, 0x62, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x69, 0x74, 0x79,
-    0x00, 0x04, 0x00, 0x00, 0x00, 0x43, 0x94, 0xd3, 0x3c, 0x00, 0xb2, 0x69, 0x01, 0x00, 0xfe, 0x91, 0x3e, 0x3f, 0x5c,
-    0xc5, 0x01, 0x00, 0xfe, 0x91, 0x3e, 0x3f, 0x05, 0x7e, 0x02, 0x00, 0xfb, 0x42, 0x2f, 0xbe, 0x33, 0x1d, 0x03, 0x00,
-    0xfb, 0x42, 0x2f, 0xbe};
-unsigned int regression_data_3_model_len = 131;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64,
+    0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x61, 0x62, 0x00, 0x04, 0x00, 0x00, 0x00, 0xc5, 0xa5, 0x6a, 0x34, 0x00, 0xb2,
+    0x69, 0x01, 0x00, 0xaa, 0x92, 0x3e, 0x3f, 0x5c, 0xc5, 0x01, 0x00, 0xaa, 0x92, 0x3e, 0x3f, 0x05, 0x7e, 0x02, 0x00,
+    0x19, 0x44, 0x2f, 0xbe, 0x33, 0x1d, 0x03, 0x00, 0x19, 0x44, 0x2f, 0xbe};
+unsigned int regression_data_3_model_len = 101;
 unsigned char regression_data_3_pred[] = {
-    0x30, 0x2e, 0x38, 0x30, 0x34, 0x32, 0x31, 0x34, 0x0a, 0x30, 0x2e, 0x31, 0x31, 0x39, 0x35, 0x39, 0x39, 0x0a};
+    0x30, 0x2e, 0x38, 0x30, 0x34, 0x32, 0x31, 0x38, 0x0a, 0x30, 0x2e, 0x31, 0x31, 0x39, 0x35, 0x38, 0x35, 0x0a};
 unsigned int regression_data_3_pred_len = 18;
-unsigned char regression_data_4_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+unsigned char regression_data_4_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x32, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x69, 0x6e, 0x74, 0x65, 0x72, 0x61, 0x63, 0x74, 0x69,
-    0x6f, 0x6e, 0x73, 0x20, 0x61, 0x62, 0x63, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e,
-    0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x25, 0x5b, 0x51, 0xe0, 0x00, 0x2b, 0x21, 0x00, 0x00, 0x86,
-    0xc6, 0x9d, 0x3c, 0x74, 0x14, 0x01, 0x00, 0x37, 0xab, 0xc9, 0xbd, 0x77, 0x14, 0x01, 0x00, 0x37, 0x50, 0x43, 0x3d,
-    0xb2, 0x69, 0x01, 0x00, 0xb2, 0x63, 0xab, 0x3d, 0x5c, 0xc5, 0x01, 0x00, 0xb2, 0x63, 0xab, 0x3d, 0x05, 0x7e, 0x02,
-    0x00, 0x78, 0xd1, 0xa0, 0x3b, 0x62, 0xd6, 0x02, 0x00, 0x00, 0xb8, 0x4f, 0x3d};
-unsigned int regression_data_4_model_len = 159;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x69, 0x6e, 0x74, 0x65,
+    0x72, 0x61, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x73, 0x20, 0x61, 0x62, 0x63, 0x00, 0x04, 0x00, 0x00, 0x00, 0x47, 0x3d,
+    0xa8, 0xc1, 0x00, 0x2b, 0x21, 0x00, 0x00, 0x22, 0xc4, 0x9d, 0x3c, 0x74, 0x14, 0x01, 0x00, 0x16, 0xaf, 0xc9, 0xbd,
+    0x77, 0x14, 0x01, 0x00, 0x17, 0x53, 0x43, 0x3d, 0xb2, 0x69, 0x01, 0x00, 0x06, 0x67, 0xab, 0x3d, 0x5c, 0xc5, 0x01,
+    0x00, 0x06, 0x67, 0xab, 0x3d, 0x05, 0x7e, 0x02, 0x00, 0x3e, 0xd6, 0xa0, 0x3b, 0x62, 0xd6, 0x02, 0x00, 0xd4, 0xb7,
+    0x4f, 0x3d};
+unsigned int regression_data_4_model_len = 129;
 unsigned char regression_data_4_pred[] = {
-    0x30, 0x2e, 0x36, 0x33, 0x35, 0x36, 0x32, 0x30, 0x0a, 0x30, 0x2e, 0x31, 0x31, 0x37, 0x38, 0x31, 0x33, 0x0a};
+    0x30, 0x2e, 0x36, 0x33, 0x35, 0x36, 0x33, 0x37, 0x0a, 0x30, 0x2e, 0x31, 0x31, 0x37, 0x38, 0x30, 0x38, 0x0a};
 unsigned int regression_data_4_pred_len = 18;
-unsigned char regression_data_5_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+unsigned char regression_data_5_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x33, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x69, 0x6e, 0x74, 0x65, 0x72, 0x61, 0x63, 0x74, 0x69,
-    0x6f, 0x6e, 0x73, 0x20, 0x61, 0x62, 0x63, 0x64, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65,
-    0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x0b, 0x1f, 0xb3, 0x77, 0x00, 0x74, 0x14, 0x01, 0x00,
-    0x2d, 0x4f, 0x94, 0xbd, 0x77, 0x14, 0x01, 0x00, 0x87, 0x9e, 0x37, 0x3d, 0xc5, 0x22, 0x01, 0x00, 0x2d, 0x4f, 0x94,
-    0xbc, 0xc6, 0x22, 0x01, 0x00, 0xd8, 0xd4, 0xf4, 0x3b, 0xb2, 0x69, 0x01, 0x00, 0x7d, 0xcf, 0xde, 0x3d, 0x5c, 0xc5,
-    0x01, 0x00, 0x7d, 0xcf, 0xde, 0x3d, 0x05, 0x7e, 0x02, 0x00, 0x59, 0x79, 0x78, 0x3c, 0x62, 0xd6, 0x02, 0x00, 0xa7,
-    0x34, 0x53, 0x3d};
-unsigned int regression_data_5_model_len = 168;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x69, 0x6e, 0x74, 0x65,
+    0x72, 0x61, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x73, 0x20, 0x61, 0x62, 0x63, 0x64, 0x00, 0x04, 0x00, 0x00, 0x00, 0x13,
+    0xd9, 0x47, 0xa4, 0x00, 0x74, 0x14, 0x01, 0x00, 0x1f, 0x4f, 0x94, 0xbd, 0x77, 0x14, 0x01, 0x00, 0x43, 0xa0, 0x37,
+    0x3d, 0xc5, 0x22, 0x01, 0x00, 0x1f, 0x4f, 0x94, 0xbc, 0xc6, 0x22, 0x01, 0x00, 0xb0, 0xd5, 0xf4, 0x3b, 0xb2, 0x69,
+    0x01, 0x00, 0x8a, 0xcd, 0xde, 0x3d, 0x5c, 0xc5, 0x01, 0x00, 0x8a, 0xcd, 0xde, 0x3d, 0x05, 0x7e, 0x02, 0x00, 0xa4,
+    0x7a, 0x78, 0x3c, 0x62, 0xd6, 0x02, 0x00, 0xfb, 0x39, 0x53, 0x3d};
+unsigned int regression_data_5_model_len = 138;
 unsigned char regression_data_5_pred[] = {
-    0x30, 0x2e, 0x37, 0x36, 0x31, 0x32, 0x34, 0x37, 0x0a, 0x30, 0x2e, 0x30, 0x34, 0x30, 0x31, 0x34, 0x38, 0x0a};
+    0x30, 0x2e, 0x37, 0x36, 0x31, 0x32, 0x36, 0x34, 0x0a, 0x30, 0x2e, 0x30, 0x34, 0x30, 0x31, 0x34, 0x38, 0x0a};
 unsigned int regression_data_5_pred_len = 18;
-unsigned char regression_data_6_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+unsigned char regression_data_6_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x21, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2e, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63,
-    0x20, 0x61, 0x62, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x69, 0x74, 0x79,
-    0x00, 0x04, 0x00, 0x00, 0x00, 0x9b, 0x9d, 0xc6, 0xde, 0x00, 0x33, 0x1d, 0x3b, 0xf5, 0x00, 0x00, 0x00, 0x00, 0x20,
-    0xdc, 0x90, 0x3b, 0x5c, 0xc5, 0xb1, 0x00, 0x00, 0x00, 0x00, 0x00, 0x8a, 0x19, 0xd8, 0x3d, 0x05, 0x7e, 0xde, 0x95,
-    0x00, 0x00, 0x00, 0x00, 0x20, 0xdc, 0x90, 0x3b, 0xb2, 0x69, 0x25, 0x3c, 0x00, 0x00, 0x00, 0x00, 0x8a, 0x19, 0xd8,
-    0x3d};
-unsigned int regression_data_6_model_len = 147;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64,
+    0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x61, 0x62, 0x00, 0x04, 0x00, 0x00, 0x00, 0x77, 0xbd, 0xf7, 0x46, 0x00, 0x33,
+    0x1d, 0x3b, 0xf5, 0x00, 0x00, 0x00, 0x00, 0xe8, 0xf2, 0x90, 0x3b, 0x5c, 0xc5, 0xb1, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0xb9, 0x21, 0xd8, 0x3d, 0x05, 0x7e, 0xde, 0x95, 0x00, 0x00, 0x00, 0x00, 0xe8, 0xf2, 0x90, 0x3b, 0xb2, 0x69, 0x25,
+    0x3c, 0x00, 0x00, 0x00, 0x00, 0xb9, 0x21, 0xd8, 0x3d};
+unsigned int regression_data_6_model_len = 117;
 unsigned char regression_data_6_pred[] = {
-    0x30, 0x2e, 0x32, 0x32, 0x38, 0x37, 0x31, 0x38, 0x0a, 0x30, 0x2e, 0x32, 0x34, 0x36, 0x34, 0x30, 0x31, 0x0a};
+    0x30, 0x2e, 0x32, 0x32, 0x38, 0x37, 0x36, 0x30, 0x0a, 0x30, 0x2e, 0x32, 0x34, 0x36, 0x34, 0x35, 0x34, 0x0a};
 unsigned int regression_data_6_pred_len = 18;
-unsigned char regression_data_7_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+unsigned char regression_data_7_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1f, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e,
-    0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x4a, 0x0d, 0xa8, 0x7d, 0x00, 0x88, 0xd8, 0x00, 0x00, 0xdb,
-    0xe8, 0x3c, 0x3e, 0x5c, 0xc5, 0x01, 0x00, 0xdb, 0xe8, 0x3c, 0x3e, 0xd2, 0x52, 0x02, 0x00, 0xb2, 0x6a, 0x14, 0x3e,
-    0xe8, 0x78, 0x03, 0x00, 0xf9, 0xc3, 0x89, 0xbd};
-unsigned int regression_data_7_model_len = 116;
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x49, 0x8e,
+    0xa9, 0xc7, 0x00, 0x88, 0xd8, 0x00, 0x00, 0xc8, 0xe9, 0x3c, 0x3e, 0x5c, 0xc5, 0x01, 0x00, 0xc8, 0xe9, 0x3c, 0x3e,
+    0xd2, 0x52, 0x02, 0x00, 0xc8, 0x6e, 0x14, 0x3e, 0xe8, 0x78, 0x03, 0x00, 0xe0, 0xc1, 0x89, 0xbd};
+unsigned int regression_data_7_model_len = 86;
 unsigned char regression_data_7_pred[] = {
-    0x30, 0x2e, 0x36, 0x35, 0x38, 0x38, 0x34, 0x31, 0x0a, 0x30, 0x2e, 0x30, 0x39, 0x39, 0x38, 0x39, 0x31, 0x0a};
+    0x30, 0x2e, 0x36, 0x35, 0x38, 0x38, 0x37, 0x39, 0x0a, 0x30, 0x2e, 0x30, 0x39, 0x39, 0x39, 0x31, 0x34, 0x0a};
 unsigned int regression_data_7_pred_len = 18;
-unsigned char multiclass_data_4_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
+unsigned char multiclass_data_4_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x49, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63,
-    0x20, 0x61, 0x62, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x20, 0x2d,
-    0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20,
-    0x69, 0x64, 0x65, 0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0xc7, 0x53, 0x35, 0xf9, 0x00, 0xb2,
-    0x69, 0x01, 0x00, 0x9a, 0x5c, 0xe3, 0x3e, 0xb7, 0x69, 0x01, 0x00, 0x48, 0x97, 0x17, 0x3d, 0x5c, 0xc5, 0x01, 0x00,
-    0x9a, 0x5c, 0xe3, 0x3e, 0x03, 0x7e, 0x02, 0x00, 0xde, 0x2e, 0x13, 0xbe, 0x16, 0x15, 0x03, 0x00, 0x18, 0x41, 0x44,
-    0xbc, 0x35, 0x1d, 0x03, 0x00, 0xde, 0x2e, 0x13, 0xbe};
-unsigned int multiclass_data_4_model_len = 174;
-unsigned char multiclass_data_4_pred[] = {0x32, 0x3a, 0x30, 0x2e, 0x30, 0x33, 0x38, 0x36, 0x32, 0x32, 0x33, 0x2c, 0x31,
-    0x3a, 0x30, 0x2e, 0x34, 0x36, 0x39, 0x38, 0x33, 0x2c, 0x30, 0x3a, 0x30, 0x2e, 0x39, 0x30, 0x31, 0x30, 0x33, 0x38,
-    0x0a, 0x0a};
-unsigned int multiclass_data_4_pred_len = 34;
-unsigned char multiclass_data_5_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x36, 0x2e, 0x31, 0x00, 0x01, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2b, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61,
+    0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e,
+    0x6b, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x61, 0x62, 0x00, 0x04, 0x00,
+    0x00, 0x00, 0x91, 0xfa, 0x13, 0x4a, 0x00, 0xb2, 0x69, 0x01, 0x00, 0xe2, 0x5e, 0xe3, 0x3e, 0xb7, 0x69, 0x01, 0x00,
+    0x98, 0x94, 0x17, 0x3d, 0x5c, 0xc5, 0x01, 0x00, 0xe2, 0x5e, 0xe3, 0x3e, 0x03, 0x7e, 0x02, 0x00, 0x21, 0x30, 0x13,
+    0xbe, 0x16, 0x15, 0x03, 0x00, 0x29, 0x40, 0x44, 0xbc, 0x35, 0x1d, 0x03, 0x00, 0x21, 0x30, 0x13, 0xbe};
+unsigned int multiclass_data_4_model_len = 144;
+unsigned char multiclass_data_4_pred[] = {0x33, 0x3a, 0x30, 0x2e, 0x30, 0x33, 0x38, 0x36, 0x30, 0x35, 0x2c, 0x32, 0x3a,
+    0x30, 0x2e, 0x34, 0x36, 0x39, 0x38, 0x32, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x39, 0x30, 0x31, 0x30, 0x33, 0x35, 0x0a,
+    0x0a};
+unsigned int multiclass_data_4_pred_len = 33;
+unsigned char multiclass_data_5_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00,
     0x00, 0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3a, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68,
-    0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66,
-    0x20, 0x6d, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c,
-    0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x58, 0xf3,
-    0xe1, 0x52, 0x00, 0xa9, 0x91, 0x00, 0x00, 0x54, 0x4d, 0x34, 0x3e, 0xae, 0x91, 0x00, 0x00, 0x68, 0x58, 0x70, 0x3c,
-    0x5c, 0xc5, 0x01, 0x00, 0x54, 0x4d, 0x34, 0x3e, 0x11, 0x76, 0x03, 0x00, 0x02, 0xc7, 0x03, 0x3d, 0x5f, 0xd7, 0x03,
-    0x00, 0xb9, 0x4e, 0xc1, 0x3c};
-unsigned int multiclass_data_5_model_len = 151;
-unsigned char multiclass_data_5_pred[] = {0x30, 0x3a, 0x30, 0x2e, 0x35, 0x35, 0x31, 0x37, 0x38, 0x34, 0x2c, 0x33, 0x3a,
-    0x30, 0x2e, 0x35, 0x35, 0x31, 0x37, 0x38, 0x34, 0x2c, 0x34, 0x3a, 0x30, 0x2e, 0x35, 0x36, 0x30, 0x33, 0x35, 0x39,
-    0x2c, 0x37, 0x3a, 0x30, 0x2e, 0x35, 0x36, 0x30, 0x33, 0x35, 0x39, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x35, 0x37, 0x35,
-    0x33, 0x38, 0x31, 0x2c, 0x35, 0x3a, 0x30, 0x2e, 0x35, 0x39, 0x32, 0x35, 0x33, 0x31, 0x2c, 0x32, 0x3a, 0x30, 0x2e,
-    0x35, 0x39, 0x38, 0x39, 0x37, 0x38, 0x2c, 0x36, 0x3a, 0x30, 0x2e, 0x36, 0x32, 0x34, 0x37, 0x30, 0x33, 0x0a, 0x0a};
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61,
+    0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e,
+    0x6b, 0x00, 0x04, 0x00, 0x00, 0x00, 0x2c, 0x43, 0x9b, 0xba, 0x00, 0xa9, 0x91, 0x00, 0x00, 0x31, 0xc8, 0xc0, 0x3e,
+    0xae, 0x91, 0x00, 0x00, 0x78, 0x85, 0x00, 0x3d, 0x5c, 0xc5, 0x01, 0x00, 0x31, 0xc8, 0xc0, 0x3e, 0x11, 0x76, 0x03,
+    0x00, 0x68, 0xf2, 0x6b, 0xbe, 0x5f, 0xd7, 0x03, 0x00, 0x01, 0x06, 0xab, 0xbe};
+unsigned int multiclass_data_5_model_len = 121;
+unsigned char multiclass_data_5_pred[] = {0x33, 0x3a, 0x30, 0x2e, 0x31, 0x32, 0x37, 0x34, 0x39, 0x32, 0x2c, 0x33, 0x3a,
+    0x30, 0x2e, 0x34, 0x33, 0x38, 0x33, 0x33, 0x31, 0x2c, 0x32, 0x3a, 0x30, 0x2e, 0x34, 0x36, 0x31, 0x35, 0x32, 0x32,
+    0x2c, 0x32, 0x3a, 0x30, 0x2e, 0x36, 0x36, 0x38, 0x37, 0x34, 0x38, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x37, 0x39, 0x35,
+    0x35, 0x35, 0x32, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x37, 0x39, 0x35, 0x35, 0x35, 0x32, 0x2c, 0x31, 0x3a, 0x30, 0x2e,
+    0x38, 0x39, 0x39, 0x31, 0x36, 0x35, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x38, 0x39, 0x39, 0x31, 0x36, 0x35, 0x0a, 0x0a};
 unsigned int multiclass_data_5_pred_len = 89;
-unsigned char cb_data_5_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
-    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x55, 0x55, 0x55, 0x40, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x9a, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68, 0x5f,
-    0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20,
-    0x61, 0x62, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66,
-    0x20, 0x2d, 0x2d, 0x65, 0x70, 0x73, 0x69, 0x6c, 0x6f, 0x6e, 0x20, 0x30, 0x2e, 0x33, 0x20, 0x2d, 0x2d, 0x6c, 0x61,
-    0x6d, 0x62, 0x64, 0x61, 0x20, 0x31, 0x20, 0x2d, 0x2d, 0x70, 0x73, 0x69, 0x20, 0x31, 0x20, 0x2d, 0x2d, 0x63, 0x62,
-    0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x69, 0x70, 0x73, 0x20,
-    0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69,
-    0x6e, 0x65, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c,
-    0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x48, 0x31,
-    0x09, 0x85, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0xb2, 0x69, 0x01, 0x00, 0x87, 0xb4, 0x0d, 0x3d, 0xb7, 0x69, 0x01, 0x00, 0x08, 0xc2, 0x9d, 0x3d, 0x5c, 0xc5, 0x01,
-    0x00, 0xec, 0xa1, 0x6c, 0x3f, 0x03, 0x7e, 0x02, 0x00, 0xa3, 0x99, 0x78, 0xbe, 0x16, 0x15, 0x03, 0x00, 0x1a, 0xbd,
-    0xa5, 0xbc, 0x35, 0x1d, 0x03, 0x00, 0x1b, 0x69, 0x2e, 0xbc};
-unsigned int cb_data_5_model_len = 271;
-unsigned char cb_data_5_pred[] = {0x32, 0x3a, 0x30, 0x2e, 0x38, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x31, 0x2c, 0x30, 0x3a,
-    0x30, 0x2e, 0x31, 0x0a, 0x0a, 0x32, 0x3a, 0x30, 0x2e, 0x38, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x31, 0x2c, 0x30, 0x3a,
+unsigned char cb_data_5_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x77, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64,
+    0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66, 0x20,
+    0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x6d, 0x74, 0x72, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f,
+    0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x2d, 0x2d,
+    0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x65, 0x70, 0x73, 0x69, 0x6c, 0x6f,
+    0x6e, 0x20, 0x30, 0x2e, 0x33, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x31, 0x31, 0x39, 0x32, 0x30, 0x39, 0x32, 0x39,
+    0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x61, 0x62, 0x00, 0x04, 0x00, 0x00,
+    0x00, 0xf8, 0x33, 0x84, 0x29, 0xc8, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x58, 0x02, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0xb2, 0x69, 0x01, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0xb7, 0x69, 0x01, 0x00, 0x3c, 0x8e, 0x63, 0x3c,
+    0x5c, 0xc5, 0x01, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0x03, 0x7e, 0x02, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0x16, 0x15, 0x03,
+    0x00, 0x3c, 0x8e, 0x63, 0x3c, 0x35, 0x1d, 0x03, 0x00, 0xaa, 0xaa, 0x2a, 0x3e};
+unsigned int cb_data_5_model_len = 236;
+unsigned char cb_data_5_pred[] = {0x30, 0x3a, 0x30, 0x2e, 0x38, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x31, 0x2c, 0x32, 0x3a,
+    0x30, 0x2e, 0x31, 0x0a, 0x0a, 0x30, 0x3a, 0x30, 0x2e, 0x38, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x31, 0x2c, 0x32, 0x3a,
     0x30, 0x2e, 0x31, 0x0a, 0x0a};
 unsigned int cb_data_5_pred_len = 38;
-unsigned char cb_data_6_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
-    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x55, 0x55, 0x55, 0x40, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x97, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68, 0x5f,
-    0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20,
-    0x61, 0x62, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66,
-    0x20, 0x2d, 0x2d, 0x6c, 0x61, 0x6d, 0x62, 0x64, 0x61, 0x20, 0x2d, 0x32, 0x20, 0x2d, 0x2d, 0x70, 0x73, 0x69, 0x20,
-    0x31, 0x20, 0x2d, 0x2d, 0x73, 0x6f, 0x66, 0x74, 0x6d, 0x61, 0x78, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64,
-    0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x69, 0x70, 0x73, 0x20, 0x2d, 0x2d, 0x63,
-    0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20,
-    0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b,
-    0x20, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x4f, 0x24, 0x56, 0xa0, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xb2, 0x69, 0x01,
-    0x00, 0xb3, 0x32, 0x51, 0x3d, 0xb7, 0x69, 0x01, 0x00, 0x10, 0x4e, 0xd3, 0x3c, 0x5c, 0xc5, 0x01, 0x00, 0x64, 0x7a,
-    0x9e, 0x3e, 0x03, 0x7e, 0x02, 0x00, 0xac, 0x65, 0x84, 0xbb, 0x16, 0x15, 0x03, 0x00, 0x98, 0x88, 0xb0, 0xb9, 0x35,
-    0x1d, 0x03, 0x00, 0x94, 0x6a, 0xdb, 0x3a};
-unsigned int cb_data_6_model_len = 268;
-unsigned char cb_data_6_pred[] = {0x32, 0x3a, 0x30, 0x2e, 0x33, 0x33, 0x37, 0x36, 0x31, 0x34, 0x2c, 0x31, 0x3a, 0x30,
-    0x2e, 0x33, 0x33, 0x33, 0x33, 0x31, 0x35, 0x2c, 0x30, 0x3a, 0x30, 0x2e, 0x33, 0x32, 0x39, 0x30, 0x37, 0x31, 0x0a,
-    0x0a, 0x32, 0x3a, 0x30, 0x2e, 0x34, 0x31, 0x30, 0x39, 0x37, 0x31, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x33, 0x32, 0x37,
-    0x37, 0x31, 0x31, 0x2c, 0x30, 0x3a, 0x30, 0x2e, 0x32, 0x36, 0x31, 0x33, 0x31, 0x39, 0x0a, 0x0a};
-unsigned int cb_data_6_pred_len = 68;
-unsigned char cb_data_7_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
-    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x55, 0x55, 0x55, 0x40, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x94, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68, 0x5f,
-    0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20,
-    0x61, 0x62, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66,
-    0x20, 0x2d, 0x2d, 0x62, 0x61, 0x67, 0x20, 0x33, 0x20, 0x2d, 0x2d, 0x6c, 0x61, 0x6d, 0x62, 0x64, 0x61, 0x20, 0x31,
-    0x20, 0x2d, 0x2d, 0x70, 0x73, 0x69, 0x20, 0x31, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d,
-    0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x69, 0x70, 0x73, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61,
-    0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x2d, 0x2d, 0x63,
-    0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64,
-    0x65, 0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x83, 0x4b, 0x89, 0xbc, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x58, 0x54, 0x00, 0x00, 0x00, 0x4e,
-    0xaa, 0xbc, 0x59, 0x54, 0x00, 0x00, 0xe9, 0x6a, 0xae, 0xbc, 0x5a, 0x54, 0x00, 0x00, 0x2e, 0x68, 0xa5, 0xbc, 0xd4,
-    0x74, 0x00, 0x00, 0x9f, 0xa9, 0xf3, 0xbb, 0xd5, 0x74, 0x00, 0x00, 0xaf, 0xba, 0xf5, 0xbb, 0xd6, 0x74, 0x00, 0x00,
-    0xb7, 0x94, 0x29, 0xbc, 0xc8, 0xa6, 0x01, 0x00, 0x39, 0x8e, 0xdf, 0x3c, 0xc9, 0xa6, 0x01, 0x00, 0x9e, 0xb1, 0xdf,
-    0x3c, 0xca, 0xa6, 0x01, 0x00, 0xe6, 0x26, 0x0f, 0x3d, 0xdc, 0xa6, 0x01, 0x00, 0x8b, 0x17, 0xa0, 0x3d, 0xdd, 0xa6,
-    0x01, 0x00, 0xd2, 0xc3, 0xa2, 0x3d, 0xde, 0xa6, 0x01, 0x00, 0x96, 0x56, 0x9d, 0x3d, 0x0c, 0xf8, 0x01, 0x00, 0xbc,
-    0x6e, 0x7f, 0xbe, 0x0d, 0xf8, 0x01, 0x00, 0x9b, 0xd2, 0x82, 0xbe, 0x0e, 0xf8, 0x01, 0x00, 0x6c, 0x1a, 0x78, 0xbe,
-    0x70, 0x15, 0x03, 0x00, 0x4a, 0x2e, 0x70, 0x3f, 0x71, 0x15, 0x03, 0x00, 0x7b, 0x25, 0x74, 0x3f, 0x72, 0x15, 0x03,
-    0x00, 0x9c, 0x07, 0x6c, 0x3f};
-unsigned int cb_data_7_model_len = 361;
-unsigned char cb_data_7_pred[] = {0x32, 0x3a, 0x31, 0x2c, 0x31, 0x3a, 0x30, 0x2c, 0x30, 0x3a, 0x30, 0x0a, 0x0a, 0x32,
-    0x3a, 0x31, 0x2c, 0x31, 0x3a, 0x30, 0x2c, 0x30, 0x3a, 0x30, 0x0a, 0x0a};
+unsigned char cb_data_6_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x71, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64,
+    0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66, 0x20,
+    0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x6d, 0x74, 0x72, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f,
+    0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x2d, 0x2d,
+    0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c, 0x61, 0x6d, 0x62, 0x64, 0x61,
+    0x20, 0x2d, 0x32, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x61, 0x62, 0x20,
+    0x2d, 0x2d, 0x73, 0x6f, 0x66, 0x74, 0x6d, 0x61, 0x78, 0x00, 0x04, 0x00, 0x00, 0x00, 0x06, 0xe4, 0xbd, 0x5d, 0x0a,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1e, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xb2, 0x69, 0x01,
+    0x00, 0x0d, 0x57, 0x2a, 0x3e, 0xb7, 0x69, 0x01, 0x00, 0xbd, 0x1e, 0x63, 0x3c, 0x5c, 0xc5, 0x01, 0x00, 0x0d, 0x57,
+    0x2a, 0x3e, 0x03, 0x7e, 0x02, 0x00, 0x0d, 0x57, 0x2a, 0x3e, 0x16, 0x15, 0x03, 0x00, 0xbd, 0x1e, 0x63, 0x3c, 0x35,
+    0x1d, 0x03, 0x00, 0x0d, 0x57, 0x2a, 0x3e};
+unsigned int cb_data_6_model_len = 230;
+unsigned char cb_data_6_pred[] = {0x30, 0x3a, 0x30, 0x2e, 0x36, 0x36, 0x34, 0x37, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x32,
+    0x34, 0x34, 0x39, 0x39, 0x38, 0x2c, 0x32, 0x3a, 0x30, 0x2e, 0x30, 0x39, 0x30, 0x33, 0x30, 0x32, 0x0a, 0x0a, 0x30,
+    0x3a, 0x30, 0x2e, 0x38, 0x36, 0x36, 0x38, 0x31, 0x33, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x31, 0x31, 0x37, 0x33, 0x31,
+    0x2c, 0x32, 0x3a, 0x30, 0x2e, 0x30, 0x31, 0x35, 0x38, 0x37, 0x36, 0x0a, 0x0a};
+unsigned int cb_data_6_pred_len = 65;
+unsigned char cb_data_7_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x85, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x62, 0x61, 0x67, 0x20, 0x33,
+    0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c,
+    0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x6d,
+    0x74, 0x72, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74,
+    0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20,
+    0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x61, 0x62, 0x20, 0x2d, 0x2d, 0x72, 0x61,
+    0x6e, 0x64, 0x6f, 0x6d, 0x5f, 0x73, 0x65, 0x65, 0x64, 0x20, 0x37, 0x35, 0x38, 0x39, 0x30, 0x36, 0x30, 0x31, 0x32,
+    0x31, 0x34, 0x36, 0x33, 0x30, 0x32, 0x31, 0x32, 0x34, 0x30, 0x00, 0x04, 0x00, 0x00, 0x00, 0x6c, 0x85, 0x14, 0x21,
+    0xc5, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x4f, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x58, 0x54,
+    0x00, 0x00, 0xbe, 0x9c, 0x29, 0x3c, 0x59, 0x54, 0x00, 0x00, 0x94, 0xad, 0x2e, 0x3c, 0x5a, 0x54, 0x00, 0x00, 0x3c,
+    0x8e, 0x63, 0x3c, 0xd4, 0x74, 0x00, 0x00, 0xba, 0x19, 0x98, 0x3c, 0xd5, 0x74, 0x00, 0x00, 0xe5, 0x51, 0xbd, 0x3c,
+    0xd6, 0x74, 0x00, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0xc8, 0xa6, 0x01, 0x00, 0xc8, 0x50, 0x3c, 0x3d, 0xc9, 0xa6, 0x01,
+    0x00, 0xf4, 0x52, 0x56, 0x3d, 0xca, 0xa6, 0x01, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0xdc, 0xa6, 0x01, 0x00, 0x29, 0x7f,
+    0xea, 0x3c, 0xdd, 0xa6, 0x01, 0x00, 0xe5, 0x3e, 0xe4, 0x3c, 0xde, 0xa6, 0x01, 0x00, 0x3c, 0x8e, 0x63, 0x3c, 0x0c,
+    0xf8, 0x01, 0x00, 0x1f, 0x6b, 0xfe, 0x3d, 0x0d, 0xf8, 0x01, 0x00, 0x31, 0x02, 0x03, 0x3e, 0x0e, 0xf8, 0x01, 0x00,
+    0xaa, 0xaa, 0x2a, 0x3e, 0x70, 0x15, 0x03, 0x00, 0x5c, 0xdf, 0xaf, 0x3e, 0x71, 0x15, 0x03, 0x00, 0x28, 0x2f, 0xab,
+    0x3e, 0x72, 0x15, 0x03, 0x00, 0xaa, 0xaa, 0x2a, 0x3e};
+unsigned int cb_data_7_model_len = 346;
+unsigned char cb_data_7_pred[] = {0x30, 0x3a, 0x31, 0x2c, 0x31, 0x3a, 0x30, 0x2c, 0x32, 0x3a, 0x30, 0x0a, 0x0a, 0x30,
+    0x3a, 0x31, 0x2c, 0x31, 0x3a, 0x30, 0x2c, 0x32, 0x3a, 0x30, 0x0a, 0x0a};
 unsigned int cb_data_7_pred_len = 26;
-unsigned char cb_data_8_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
-    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x55, 0x55, 0x55, 0x40, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xa3, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68, 0x5f,
-    0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20,
-    0x61, 0x62, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66,
-    0x20, 0x2d, 0x2d, 0x62, 0x61, 0x67, 0x20, 0x35, 0x20, 0x2d, 0x2d, 0x65, 0x70, 0x73, 0x69, 0x6c, 0x6f, 0x6e, 0x20,
-    0x30, 0x2e, 0x32, 0x37, 0x20, 0x2d, 0x2d, 0x6c, 0x61, 0x6d, 0x62, 0x64, 0x61, 0x20, 0x31, 0x20, 0x2d, 0x2d, 0x70,
-    0x73, 0x69, 0x20, 0x31, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f,
-    0x74, 0x79, 0x70, 0x65, 0x20, 0x69, 0x70, 0x73, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64,
-    0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61,
-    0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c, 0x69, 0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x69,
-    0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0xb1, 0xf7, 0x2d, 0x17, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xb0, 0xa8, 0x00, 0x00, 0x71, 0x18, 0x9b, 0xbc, 0xb1, 0xa8,
-    0x00, 0x00, 0x7d, 0x0d, 0xb6, 0xbc, 0xb2, 0xa8, 0x00, 0x00, 0x54, 0x87, 0xbb, 0xbc, 0xb3, 0xa8, 0x00, 0x00, 0x42,
-    0x3d, 0xab, 0xbc, 0xb4, 0xa8, 0x00, 0x00, 0x43, 0xee, 0xa1, 0xbc, 0xa8, 0xe9, 0x00, 0x00, 0xb6, 0x76, 0x38, 0xbc,
-    0xa9, 0xe9, 0x00, 0x00, 0x20, 0xb2, 0x39, 0xbc, 0xaa, 0xe9, 0x00, 0x00, 0x5a, 0x46, 0x45, 0xbc, 0xab, 0xe9, 0x00,
-    0x00, 0x7e, 0x0e, 0x49, 0xbc, 0xac, 0xe9, 0x00, 0x00, 0xa1, 0x9a, 0x5a, 0xbc, 0xe0, 0x2a, 0x02, 0x00, 0x51, 0x4e,
-    0x63, 0x3f, 0xe1, 0x2a, 0x02, 0x00, 0x4d, 0xe9, 0x79, 0x3f, 0xe2, 0x2a, 0x02, 0x00, 0x14, 0xcd, 0x7e, 0x3f, 0xe3,
-    0x2a, 0x02, 0x00, 0x41, 0xca, 0x70, 0x3f, 0xe4, 0x2a, 0x02, 0x00, 0xe8, 0xb9, 0x68, 0x3f, 0x90, 0x4d, 0x03, 0x00,
-    0x6b, 0x97, 0x17, 0x3d, 0x91, 0x4d, 0x03, 0x00, 0xd1, 0x6f, 0x17, 0x3d, 0x92, 0x4d, 0x03, 0x00, 0x0f, 0xd3, 0x1d,
-    0x3d, 0x93, 0x4d, 0x03, 0x00, 0xed, 0xa8, 0x23, 0x3d, 0x94, 0x4d, 0x03, 0x00, 0x44, 0xa8, 0x2b, 0x3d, 0xb8, 0x4d,
-    0x03, 0x00, 0x8f, 0x87, 0x97, 0x3d, 0xb9, 0x4d, 0x03, 0x00, 0x2a, 0x98, 0xa6, 0x3d, 0xba, 0x4d, 0x03, 0x00, 0x7a,
-    0xdf, 0xa9, 0x3d, 0xbb, 0x4d, 0x03, 0x00, 0xc9, 0x85, 0xa0, 0x3d, 0xbc, 0x4d, 0x03, 0x00, 0xc9, 0x1d, 0x9b, 0x3d,
-    0x18, 0xf0, 0x03, 0x00, 0xda, 0xa7, 0x68, 0xbe, 0x19, 0xf0, 0x03, 0x00, 0xc3, 0x88, 0x88, 0xbe, 0x1a, 0xf0, 0x03,
-    0x00, 0x95, 0xa5, 0x8c, 0xbe, 0x1b, 0xf0, 0x03, 0x00, 0x5c, 0x6c, 0x80, 0xbe, 0x1c, 0xf0, 0x03, 0x00, 0xea, 0xf0,
-    0x72, 0xbe};
-unsigned int cb_data_8_model_len = 472;
-unsigned char cb_data_8_pred[] = {0x32, 0x3a, 0x30, 0x2e, 0x38, 0x32, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x30, 0x39, 0x2c,
-    0x30, 0x3a, 0x30, 0x2e, 0x30, 0x39, 0x0a, 0x0a, 0x32, 0x3a, 0x30, 0x2e, 0x38, 0x32, 0x2c, 0x31, 0x3a, 0x30, 0x2e,
-    0x30, 0x39, 0x2c, 0x30, 0x3a, 0x30, 0x2e, 0x30, 0x39, 0x0a, 0x0a};
+unsigned char cb_data_8_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3f, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xa2, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x62, 0x61, 0x67, 0x20, 0x35,
+    0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c,
+    0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x6d,
+    0x74, 0x72, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74,
+    0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20,
+    0x2d, 0x2d, 0x65, 0x70, 0x73, 0x69, 0x6c, 0x6f, 0x6e, 0x20, 0x30, 0x2e, 0x32, 0x37, 0x30, 0x30, 0x30, 0x30, 0x30,
+    0x31, 0x30, 0x37, 0x32, 0x38, 0x38, 0x33, 0x36, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69,
+    0x63, 0x20, 0x61, 0x62, 0x20, 0x2d, 0x2d, 0x72, 0x61, 0x6e, 0x64, 0x6f, 0x6d, 0x5f, 0x73, 0x65, 0x65, 0x64, 0x20,
+    0x31, 0x36, 0x37, 0x30, 0x39, 0x32, 0x38, 0x30, 0x35, 0x30, 0x37, 0x32, 0x38, 0x33, 0x33, 0x33, 0x35, 0x35, 0x36,
+    0x30, 0x00, 0x04, 0x00, 0x00, 0x00, 0x77, 0x5b, 0xb6, 0xc8, 0xd0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x70,
+    0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xb0, 0xa8, 0x00, 0x00, 0x3c, 0x8e, 0x63, 0x3c, 0xb1, 0xa8, 0x00,
+    0x00, 0xa8, 0xfd, 0x2c, 0x3c, 0xb2, 0xa8, 0x00, 0x00, 0x94, 0xad, 0x2e, 0x3c, 0xb3, 0xa8, 0x00, 0x00, 0x3c, 0x8e,
+    0x63, 0x3c, 0xb4, 0xa8, 0x00, 0x00, 0x3c, 0x8e, 0x63, 0x3c, 0xa8, 0xe9, 0x00, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0xa9,
+    0xe9, 0x00, 0x00, 0xfb, 0x83, 0xb3, 0x3c, 0xaa, 0xe9, 0x00, 0x00, 0xe5, 0x51, 0xbd, 0x3c, 0xab, 0xe9, 0x00, 0x00,
+    0xaa, 0xaa, 0x2a, 0x3e, 0xac, 0xe9, 0x00, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0xe0, 0x2a, 0x02, 0x00, 0xaa, 0xaa, 0x2a,
+    0x3e, 0xe1, 0x2a, 0x02, 0x00, 0xe0, 0x91, 0xac, 0x3e, 0xe2, 0x2a, 0x02, 0x00, 0x28, 0x2f, 0xab, 0x3e, 0xe3, 0x2a,
+    0x02, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0xe4, 0x2a, 0x02, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0x90, 0x4d, 0x03, 0x00, 0xaa,
+    0xaa, 0x2a, 0x3e, 0x91, 0x4d, 0x03, 0x00, 0x08, 0x2e, 0x4f, 0x3d, 0x92, 0x4d, 0x03, 0x00, 0xf4, 0x52, 0x56, 0x3d,
+    0x93, 0x4d, 0x03, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0x94, 0x4d, 0x03, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0xb8, 0x4d, 0x03,
+    0x00, 0x3c, 0x8e, 0x63, 0x3c, 0xb9, 0x4d, 0x03, 0x00, 0xd5, 0x17, 0xe6, 0x3c, 0xba, 0x4d, 0x03, 0x00, 0xe5, 0x3e,
+    0xe4, 0x3c, 0xbb, 0x4d, 0x03, 0x00, 0x3c, 0x8e, 0x63, 0x3c, 0xbc, 0x4d, 0x03, 0x00, 0x3c, 0x8e, 0x63, 0x3c, 0x18,
+    0xf0, 0x03, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0x19, 0xf0, 0x03, 0x00, 0x40, 0xbe, 0x01, 0x3e, 0x1a, 0xf0, 0x03, 0x00,
+    0x31, 0x02, 0x03, 0x3e, 0x1b, 0xf0, 0x03, 0x00, 0xaa, 0xaa, 0x2a, 0x3e, 0x1c, 0xf0, 0x03, 0x00, 0xaa, 0xaa, 0x2a,
+    0x3e};
+unsigned int cb_data_8_model_len = 471;
+unsigned char cb_data_8_pred[] = {0x30, 0x3a, 0x30, 0x2e, 0x38, 0x32, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x30, 0x39, 0x2c,
+    0x32, 0x3a, 0x30, 0x2e, 0x30, 0x39, 0x0a, 0x0a, 0x30, 0x3a, 0x30, 0x2e, 0x38, 0x32, 0x2c, 0x31, 0x3a, 0x30, 0x2e,
+    0x30, 0x39, 0x2c, 0x32, 0x3a, 0x30, 0x2e, 0x30, 0x39, 0x0a, 0x0a};
 unsigned int cb_data_8_pred_len = 44;
-unsigned char cb_data_9_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2e, 0x35, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
-    0x00, 0x6d, 0x00, 0x00, 0x00, 0x00, 0x55, 0x55, 0x55, 0x40, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x99, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x68, 0x61, 0x73, 0x68, 0x5f,
-    0x73, 0x65, 0x65, 0x64, 0x20, 0x30, 0x20, 0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20,
-    0x61, 0x62, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66,
-    0x20, 0x2d, 0x2d, 0x65, 0x70, 0x73, 0x69, 0x6c, 0x6f, 0x6e, 0x20, 0x30, 0x2e, 0x33, 0x20, 0x2d, 0x2d, 0x6c, 0x61,
-    0x6d, 0x62, 0x64, 0x61, 0x20, 0x31, 0x20, 0x2d, 0x2d, 0x70, 0x73, 0x69, 0x20, 0x31, 0x20, 0x2d, 0x2d, 0x63, 0x62,
-    0x5f, 0x61, 0x64, 0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x64, 0x72, 0x20, 0x2d,
-    0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69, 0x6e,
-    0x65, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x6c, 0x69,
-    0x6e, 0x6b, 0x20, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x69, 0x74, 0x79, 0x00, 0x04, 0x00, 0x00, 0x00, 0x2a, 0x6d, 0x36,
-    0x99, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06,
-    0xfc, 0x00, 0x00, 0x5d, 0x18, 0x2c, 0x38, 0x07, 0xfc, 0x00, 0x00, 0x9f, 0x19, 0xe6, 0x33, 0x2c, 0x2a, 0x02, 0x00,
-    0xe3, 0x3a, 0x8e, 0x36, 0x6a, 0x3a, 0x02, 0x00, 0xaf, 0x53, 0x0a, 0xb8, 0x64, 0xd3, 0x02, 0x00, 0x62, 0x48, 0xa4,
-    0x38, 0x6e, 0xd3, 0x02, 0x00, 0x0b, 0x9f, 0x2a, 0x3d, 0xb8, 0x8a, 0x03, 0x00, 0x44, 0xf5, 0xff, 0x3e, 0xb9, 0x8a,
-    0x03, 0x00, 0xfd, 0xff, 0x7f, 0x3f};
-unsigned int cb_data_9_model_len = 286;
+unsigned char cb_data_9_model[] = {0x06, 0x00, 0x00, 0x00, 0x39, 0x2e, 0x39, 0x2e, 0x30, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x00, 0x6d, 0x36, 0x70, 0xc2, 0xbf, 0x29, 0x5e, 0x62, 0x40, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x76, 0x00, 0x00, 0x00, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x61, 0x64,
+    0x66, 0x20, 0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x65, 0x78, 0x70, 0x6c, 0x6f, 0x72, 0x65, 0x5f, 0x61, 0x64, 0x66, 0x20,
+    0x2d, 0x2d, 0x63, 0x62, 0x5f, 0x74, 0x79, 0x70, 0x65, 0x20, 0x64, 0x72, 0x20, 0x2d, 0x2d, 0x63, 0x73, 0x6f, 0x61,
+    0x61, 0x5f, 0x6c, 0x64, 0x66, 0x20, 0x6d, 0x75, 0x6c, 0x74, 0x69, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x2d, 0x2d, 0x63,
+    0x73, 0x6f, 0x61, 0x61, 0x5f, 0x72, 0x61, 0x6e, 0x6b, 0x20, 0x2d, 0x2d, 0x65, 0x70, 0x73, 0x69, 0x6c, 0x6f, 0x6e,
+    0x20, 0x30, 0x2e, 0x33, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x31, 0x31, 0x39, 0x32, 0x30, 0x39, 0x32, 0x39, 0x20,
+    0x2d, 0x2d, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x61, 0x62, 0x00, 0x04, 0x00, 0x00, 0x00,
+    0x0f, 0xd3, 0x3b, 0x51, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x06, 0xfc, 0x00, 0x00, 0x4e, 0x66, 0xa0, 0x3d, 0x07, 0xfc, 0x00, 0x00, 0x30, 0xa5, 0xa0, 0x3d, 0x2c,
+    0x2a, 0x02, 0x00, 0x92, 0xdd, 0xd5, 0x3b, 0x2d, 0x2a, 0x02, 0x00, 0x90, 0x31, 0xd6, 0x3b, 0x6a, 0x3a, 0x02, 0x00,
+    0x06, 0xa3, 0x94, 0xbc, 0x6b, 0x3a, 0x02, 0x00, 0xcb, 0x5a, 0x95, 0xbc, 0x64, 0xd3, 0x02, 0x00, 0x60, 0x13, 0x23,
+    0xbc, 0x65, 0xd3, 0x02, 0x00, 0x09, 0xc9, 0x1f, 0xbc, 0x6e, 0xd3, 0x02, 0x00, 0xf8, 0xc2, 0x14, 0x3d, 0x6f, 0xd3,
+    0x02, 0x00, 0x23, 0xab, 0x14, 0x3d, 0xb8, 0x8a, 0x03, 0x00, 0x66, 0x24, 0xdf, 0x3e, 0xb9, 0x8a, 0x03, 0x00, 0xa7,
+    0x00, 0xdf, 0x3e};
+unsigned int cb_data_9_model_len = 283;
 unsigned char cb_data_9_pred[] = {0x30, 0x3a, 0x30, 0x2e, 0x38, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x31, 0x2c, 0x32, 0x3a,
     0x30, 0x2e, 0x31, 0x0a, 0x0a, 0x30, 0x3a, 0x30, 0x2e, 0x38, 0x2c, 0x31, 0x3a, 0x30, 0x2e, 0x31, 0x2c, 0x32, 0x3a,
     0x30, 0x2e, 0x31, 0x0a, 0x0a};
 unsigned int cb_data_9_pred_len = 38;
-unsigned char cb_data_epsilon_0_skype_jb_model[] = {0x06, 0x00, 0x00, 0x00, 0x38, 0x2E, 0x34, 0x2E, 0x30, 0x00, 0x01,
-    0x00, 0x00, 0x00, 0x00, 0x6D, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x20, 0x41, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x9B, 0x00, 0x00, 0x00, 0x20, 0x2D, 0x2D, 0x71, 0x75,
-    0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x36, 0x38, 0x20, 0x2D, 0x2D, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61,
-    0x74, 0x69, 0x63, 0x20, 0x31, 0x38, 0x20, 0x2D, 0x2D, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20,
-    0x33, 0x38, 0x20, 0x2D, 0x2D, 0x71, 0x75, 0x61, 0x64, 0x72, 0x61, 0x74, 0x69, 0x63, 0x20, 0x34, 0x38, 0x20, 0x2D,
-    0x2D, 0x63, 0x62, 0x5F, 0x65, 0x78, 0x70, 0x6C, 0x6F, 0x72, 0x65, 0x5F, 0x61, 0x64, 0x66, 0x20, 0x2D, 0x2D, 0x65,
-    0x70, 0x73, 0x69, 0x6C, 0x6F, 0x6E, 0x20, 0x30, 0x2E, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x20, 0x2D, 0x2D, 0x63,
-    0x62, 0x5F, 0x61, 0x64, 0x66, 0x20, 0x2D, 0x2D, 0x63, 0x62, 0x5F, 0x74, 0x79, 0x70, 0x65, 0x20, 0x69, 0x70, 0x73,
-    0x20, 0x2D, 0x2D, 0x63, 0x73, 0x6F, 0x61, 0x61, 0x5F, 0x6C, 0x64, 0x66, 0x20, 0x6D, 0x75, 0x6C, 0x74, 0x69, 0x6C,
-    0x69, 0x6E, 0x65, 0x20, 0x2D, 0x2D, 0x63, 0x73, 0x6F, 0x61, 0x61, 0x5F, 0x72, 0x61, 0x6E, 0x6B, 0x00, 0x04, 0x00,
-    0x00, 0x00, 0x50, 0x5F, 0x61, 0x6D, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x1F, 0x6C, 0x90, 0x3C, 0x11, 0x00, 0x00, 0x00, 0xC9, 0x0B, 0xD1,
-    0x3C, 0x22, 0x00, 0x00, 0x00, 0x16, 0xDA, 0x91, 0x3B, 0x24, 0x00, 0x00, 0x00, 0x1B, 0xBC, 0x92, 0x3C, 0x30, 0x00,
-    0x00, 0x00, 0xFF, 0xC4, 0x63, 0x3B, 0x31, 0x00, 0x00, 0x00, 0x24, 0x2A, 0x91, 0xB9, 0x34, 0x00, 0x00, 0x00, 0x34,
-    0x71, 0x24, 0x3D, 0x40, 0x00, 0x00, 0x00, 0x08, 0x54, 0xF9, 0x3C, 0x42, 0x00, 0x00, 0x00, 0x0F, 0x4A, 0x55, 0x3C,
-    0x50, 0x00, 0x00, 0x00, 0x52, 0x60, 0x14, 0x3C, 0x51, 0x00, 0x00, 0x00, 0x71, 0x7B, 0x23, 0x3C, 0x52, 0x00, 0x00,
-    0x00, 0x4E, 0xD3, 0x10, 0x3C, 0x53, 0x00, 0x00, 0x00, 0x4A, 0x47, 0x17, 0x3C, 0x54, 0x00, 0x00, 0x00, 0x4E, 0x6F,
-    0x16, 0x3C, 0x55, 0x00, 0x00, 0x00, 0x24, 0x25, 0x1B, 0x3C, 0x56, 0x00, 0x00, 0x00, 0x52, 0xF4, 0x1A, 0x3C, 0x57,
-    0x00, 0x00, 0x00, 0x5E, 0x4A, 0x28, 0x3C, 0x58, 0x00, 0x00, 0x00, 0x3A, 0xFA, 0x24, 0x3C, 0x59, 0x00, 0x00, 0x00,
-    0x2E, 0x47, 0x2B, 0x3C, 0x07, 0x04, 0x00, 0x00, 0xEA, 0x0E, 0x1A, 0x3D, 0x17, 0x04, 0x00, 0x00, 0xF6, 0x83, 0x64,
-    0xBA, 0x27, 0x04, 0x00, 0x00, 0x08, 0xA3, 0xA2, 0xBA, 0x60, 0x19, 0x00, 0x00, 0xCF, 0xB0, 0x93, 0x3B, 0x61, 0x19,
-    0x00, 0x00, 0xD0, 0xE4, 0x8C, 0x3B, 0x62, 0x19, 0x00, 0x00, 0xE0, 0x96, 0x98, 0x3B, 0x63, 0x19, 0x00, 0x00, 0x85,
-    0xE5, 0x8A, 0x3B, 0x64, 0x19, 0x00, 0x00, 0x06, 0xC2, 0xA5, 0x3B, 0x65, 0x19, 0x00, 0x00, 0x70, 0x22, 0xB7, 0x3B,
-    0x66, 0x19, 0x00, 0x00, 0x6F, 0xA2, 0x90, 0x3B, 0x67, 0x19, 0x00, 0x00, 0xAF, 0x6D, 0xAF, 0x3B, 0x68, 0x19, 0x00,
-    0x00, 0x5C, 0x03, 0x9F, 0x3B, 0x69, 0x19, 0x00, 0x00, 0x7C, 0x12, 0xBA, 0x3B, 0x90, 0x1A, 0x00, 0x00, 0xCE, 0x61,
-    0xEB, 0x3B, 0x91, 0x1A, 0x00, 0x00, 0xDE, 0x5C, 0xEC, 0x3B, 0x92, 0x1A, 0x00, 0x00, 0xA8, 0x99, 0xE6, 0x3B, 0x93,
-    0x1A, 0x00, 0x00, 0x3B, 0x98, 0xDF, 0x3B, 0x94, 0x1A, 0x00, 0x00, 0x17, 0xA7, 0x00, 0x3C, 0x95, 0x1A, 0x00, 0x00,
-    0x54, 0xC4, 0x00, 0x3C, 0x96, 0x1A, 0x00, 0x00, 0x3D, 0xF9, 0x07, 0x3C, 0x97, 0x1A, 0x00, 0x00, 0x72, 0xBB, 0xF9,
-    0x3B, 0x9A, 0x1A, 0x00, 0x00, 0xC5, 0x53, 0x06, 0x3C, 0x9B, 0x1A, 0x00, 0x00, 0x0B, 0x26, 0xFA, 0x3B, 0xD0, 0x35,
-    0x00, 0x00, 0x01, 0xA2, 0x73, 0x3A, 0xD1, 0x35, 0x00, 0x00, 0x2E, 0x1B, 0x76, 0xB9, 0xD2, 0x35, 0x00, 0x00, 0x0A,
-    0xD5, 0xFB, 0x38, 0xD3, 0x35, 0x00, 0x00, 0x3F, 0xE2, 0x59, 0x3A, 0xD4, 0x35, 0x00, 0x00, 0x05, 0x35, 0xA8, 0xB9,
-    0xD5, 0x35, 0x00, 0x00, 0xED, 0x75, 0xC1, 0xB8, 0xD6, 0x35, 0x00, 0x00, 0xE9, 0x66, 0x84, 0xBA, 0xD7, 0x35, 0x00,
-    0x00, 0xA3, 0x93, 0x5B, 0xBA, 0xDE, 0x35, 0x00, 0x00, 0xD8, 0xEB, 0x28, 0x3A, 0xDF, 0x35, 0x00, 0x00, 0xF2, 0xA3,
-    0x28, 0x3A, 0xF4, 0x38, 0x00, 0x00, 0x07, 0xE5, 0xB3, 0x3B, 0xF5, 0x38, 0x00, 0x00, 0xF3, 0x63, 0xB1, 0x3B, 0xF8,
-    0x38, 0x00, 0x00, 0xC5, 0xF0, 0xC0, 0x3B, 0xF9, 0x38, 0x00, 0x00, 0x40, 0x85, 0xBE, 0x3B, 0xFA, 0x38, 0x00, 0x00,
-    0xEB, 0xCB, 0xAE, 0x3B, 0xFB, 0x38, 0x00, 0x00, 0x64, 0xBC, 0xC0, 0x3B, 0xFC, 0x38, 0x00, 0x00, 0xB8, 0x35, 0x94,
-    0x3B, 0xFD, 0x38, 0x00, 0x00, 0xB2, 0x87, 0x9D, 0x3B, 0xFE, 0x38, 0x00, 0x00, 0x86, 0x21, 0xA4, 0x3B, 0xFF, 0x38,
-    0x00, 0x00, 0x4F, 0x11, 0xA1, 0x3B, 0xC0, 0x4B, 0x00, 0x00, 0x9C, 0x97, 0x14, 0x3A, 0xC1, 0x4B, 0x00, 0x00, 0x8C,
-    0x68, 0x75, 0x3A, 0xC2, 0x4B, 0x00, 0x00, 0x31, 0xC2, 0x20, 0x3A, 0xC3, 0x4B, 0x00, 0x00, 0xEF, 0x20, 0xB5, 0x39,
-    0xC4, 0x4B, 0x00, 0x00, 0xC0, 0x6A, 0x4C, 0x39, 0xC5, 0x4B, 0x00, 0x00, 0x37, 0x8A, 0x7B, 0xB9, 0xC6, 0x4B, 0x00,
-    0x00, 0x0C, 0x50, 0xB6, 0x3A, 0xC7, 0x4B, 0x00, 0x00, 0x7F, 0x54, 0xE2, 0x39, 0xC8, 0x4B, 0x00, 0x00, 0x0F, 0x8B,
-    0x8A, 0x3A, 0xC9, 0x4B, 0x00, 0x00, 0xD3, 0x50, 0xB2, 0x3A, 0x70, 0x4D, 0x00, 0x00, 0x98, 0x6D, 0xA5, 0xB9, 0x71,
-    0x4D, 0x00, 0x00, 0x69, 0x76, 0xA5, 0xB9, 0x72, 0x4D, 0x00, 0x00, 0xD1, 0xBD, 0xA5, 0x39, 0x73, 0x4D, 0x00, 0x00,
-    0x88, 0x23, 0xA5, 0xB9, 0x74, 0x4D, 0x00, 0x00, 0xFD, 0x72, 0xA5, 0xB9, 0x75, 0x4D, 0x00, 0x00, 0x1E, 0x79, 0xA5,
-    0xB9, 0x76, 0x4D, 0x00, 0x00, 0x79, 0x77, 0xA5, 0xB9, 0x77, 0x4D, 0x00, 0x00, 0x6C, 0x77, 0xA5, 0xB9, 0x7A, 0x4D,
-    0x00, 0x00, 0x06, 0x78, 0xA5, 0xB9, 0x7B, 0x4D, 0x00, 0x00, 0xEF, 0x6D, 0xA5, 0xB9, 0x84, 0x51, 0x00, 0x00, 0x1A,
-    0x82, 0x57, 0x3C, 0x85, 0x51, 0x00, 0x00, 0x52, 0xA5, 0x54, 0x3C, 0x88, 0x51, 0x00, 0x00, 0x54, 0xD8, 0x4D, 0x3C,
-    0x89, 0x51, 0x00, 0x00, 0x9F, 0xE9, 0x3D, 0x3C, 0x8A, 0x51, 0x00, 0x00, 0xD7, 0xA9, 0x4A, 0x3C, 0x8B, 0x51, 0x00,
-    0x00, 0x90, 0x5C, 0x4D, 0x3C, 0x8C, 0x51, 0x00, 0x00, 0x7F, 0x1A, 0x4B, 0x3C, 0x8D, 0x51, 0x00, 0x00, 0x01, 0x70,
-    0x51, 0x3C, 0x8E, 0x51, 0x00, 0x00, 0x36, 0x65, 0x2E, 0x3C, 0x8F, 0x51, 0x00, 0x00, 0xAF, 0x53, 0x4F, 0x3C, 0x90,
-    0x64, 0x00, 0x00, 0xA4, 0x16, 0x13, 0x3C, 0x91, 0x64, 0x00, 0x00, 0x1C, 0x79, 0xE7, 0x3B, 0x92, 0x64, 0x00, 0x00,
-    0xCE, 0xB9, 0x07, 0x3C, 0x93, 0x64, 0x00, 0x00, 0xC9, 0x54, 0x0D, 0x3C, 0x94, 0x64, 0x00, 0x00, 0x94, 0x39, 0x0A,
-    0x3C, 0x95, 0x64, 0x00, 0x00, 0xAB, 0xD9, 0x14, 0x3C, 0x96, 0x64, 0x00, 0x00, 0xC1, 0x47, 0x0C, 0x3C, 0x97, 0x64,
-    0x00, 0x00, 0x45, 0x0E, 0x1C, 0x3C, 0x98, 0x64, 0x00, 0x00, 0xD0, 0xA3, 0x16, 0x3C, 0x99, 0x64, 0x00, 0x00, 0xE3,
-    0xBB, 0x17, 0x3C, 0xB0, 0x67, 0x00, 0x00, 0x2F, 0x90, 0x8C, 0x3B, 0xB1, 0x67, 0x00, 0x00, 0x77, 0xB4, 0x3C, 0x3B,
-    0xB2, 0x67, 0x00, 0x00, 0xD3, 0x3E, 0x8C, 0x3B, 0xB3, 0x67, 0x00, 0x00, 0x7A, 0xD7, 0x09, 0x3B, 0xB4, 0x67, 0x00,
-    0x00, 0xC6, 0x3C, 0x59, 0x3B, 0xB5, 0x67, 0x00, 0x00, 0x69, 0x4B, 0x0A, 0x3B, 0xB6, 0x67, 0x00, 0x00, 0x77, 0xDD,
-    0x48, 0x3B, 0xB7, 0x67, 0x00, 0x00, 0x78, 0xE8, 0xA1, 0x3A, 0xBE, 0x67, 0x00, 0x00, 0x81, 0x47, 0x87, 0x3B, 0xBF,
-    0x67, 0x00, 0x00, 0x97, 0x87, 0x8E, 0x3B, 0x5C, 0xC5, 0x01, 0x00, 0x15, 0x90, 0xFD, 0x3C, 0x50, 0x57, 0x02, 0x00,
-    0x9C, 0xC7, 0x2C, 0x3C, 0x51, 0x57, 0x02, 0x00, 0x5B, 0x0E, 0x14, 0x3C, 0x52, 0x57, 0x02, 0x00, 0xE4, 0x30, 0x24,
-    0x3C, 0x53, 0x57, 0x02, 0x00, 0x0E, 0x23, 0x0D, 0x3C, 0x54, 0x57, 0x02, 0x00, 0x15, 0xCC, 0x42, 0x3C, 0x55, 0x57,
-    0x02, 0x00, 0x8D, 0x14, 0x3F, 0x3C, 0x56, 0x57, 0x02, 0x00, 0x84, 0x3F, 0x44, 0x3C, 0x57, 0x57, 0x02, 0x00, 0xA2,
-    0x59, 0x0D, 0x3C, 0x5C, 0x57, 0x02, 0x00, 0xA6, 0x7B, 0x44, 0x3C, 0x5D, 0x57, 0x02, 0x00, 0xC9, 0xC8, 0x32, 0x3C,
-    0x60, 0x70, 0x02, 0x00, 0x56, 0x18, 0x0D, 0xBA, 0x61, 0x70, 0x02, 0x00, 0x00, 0x10, 0x39, 0x34, 0x62, 0x70, 0x02,
-    0x00, 0xE1, 0x7E, 0x9B, 0x39, 0x63, 0x70, 0x02, 0x00, 0x41, 0xFD, 0x0C, 0xBA, 0x64, 0x70, 0x02, 0x00, 0xF0, 0xBE,
-    0x0C, 0xBA, 0x65, 0x70, 0x02, 0x00, 0xC3, 0xC2, 0x0C, 0xBA, 0x66, 0x70, 0x02, 0x00, 0xCA, 0x22, 0x0D, 0xBA, 0x67,
-    0x70, 0x02, 0x00, 0x54, 0xC2, 0x0C, 0xBA, 0x6C, 0x70, 0x02, 0x00, 0x48, 0x2C, 0x0D, 0xBA, 0x6D, 0x70, 0x02, 0x00,
-    0xDC, 0x44, 0x0D, 0xBA, 0x30, 0x89, 0x02, 0x00, 0x39, 0x9F, 0x00, 0xB9, 0x31, 0x89, 0x02, 0x00, 0x23, 0x11, 0x03,
-    0xBB, 0x32, 0x89, 0x02, 0x00, 0x19, 0xA2, 0x02, 0xBB, 0x33, 0x89, 0x02, 0x00, 0xFF, 0xF0, 0xD6, 0xBA, 0x34, 0x89,
-    0x02, 0x00, 0x0C, 0x23, 0x03, 0xBB, 0x35, 0x89, 0x02, 0x00, 0x4C, 0xC7, 0x02, 0xBB, 0x36, 0x89, 0x02, 0x00, 0x36,
-    0xE6, 0x9D, 0x39, 0x37, 0x89, 0x02, 0x00, 0x38, 0xE8, 0x3F, 0xB9, 0x3C, 0x89, 0x02, 0x00, 0xDD, 0x62, 0xD9, 0x39,
-    0x3D, 0x89, 0x02, 0x00, 0x88, 0xE6, 0x03, 0xBB};
-unsigned int cb_data_epsilon_0_skype_jb_model_len = 1368;
-unsigned char cb_data_epsilon_0_skype_jb_pred[] = {0x00};
-unsigned int cb_data_epsilon_0_skype_jb_pred_len = 1;
\ No newline at end of file
diff --git a/vowpalwabbit/slim/test/data/cb_data_5.model b/vowpalwabbit/slim/test/data/cb_data_5.model
index b62b3d410dfff32300ee5529928b076d47707a9d..140b6f4740b8ceedcf1c0a690072fe6570d1369a 100644
GIT binary patch
literal 236
zcmZQ$U|_J+v(z(SU<9%lazO+GLxa5#gaIeZfh+}G-Q=YB#FR7$BekL+C%-5aAy86T
zkgAYdQUsDN&QDB?&jBjUEzK#(%*o74g^Ly?=4AspsRhNEIr(`C271N@U|?uyX=Gq&
zWC@fjEKN))N-W7tR!B@@U;$eF!?;EB1k^DROi;$AOh$%PtF-L418JMSWSf|y5I%Dq
R6IfhKlo>2;D$5M!0|3()IynFU

literal 271
zcmZQ$U|_J&Gu1O-U<9%lazR9BXsCk_h=BxV0oe+=x*3VZ8S%xbsVNEuKwe>KVoFhB
zNoKM_ViJ&>oD`p0QIL~glp3Fyk_Hq=Ehx^+$<I?T&@%?|auRcsQW6ymflQz{m<H;A
zs01=fDhpB-G7E}91{LQgCdTIg&B!gyDap*q%u9uf7A59ogAB_AQkf~Kc_o=8l?*IE
z*LfInwjwzg!rYX}$k4up*LFLQ<~TIhHs&ZJ!<&UU_RMum42x%0>=P4ZW{}#ubdRYl
KGlO)d-W~wR=tnjH

diff --git a/vowpalwabbit/slim/test/data/cb_data_5.pred b/vowpalwabbit/slim/test/data/cb_data_5.pred
index a95a5317690..37724df1fb5 100644
--- a/vowpalwabbit/slim/test/data/cb_data_5.pred
+++ b/vowpalwabbit/slim/test/data/cb_data_5.pred
@@ -1,4 +1,4 @@
-2:0.8,1:0.1,0:0.1
+0:0.8,1:0.1,2:0.1
 
-2:0.8,1:0.1,0:0.1
+0:0.8,1:0.1,2:0.1
 
diff --git a/vowpalwabbit/slim/test/data/cb_data_6.model b/vowpalwabbit/slim/test/data/cb_data_6.model
index d5d865327217dc0b0b1d7f172f2a00dd9b7d3e7d..a3c9537945a183c4ca8f0809ba2204f88482ea19 100644
GIT binary patch
literal 230
zcmZQ$U|_J+v(z(SU<9%lazO+GLxa5#gaIcDfh+}G-Q=YB#FR7$BekL+C%-5aAy86T
zkgAYdQUsDN&QDB?&jBjUEzK#(%*o74g^Ly?=4AspIf=PRDTxZYMnHC9X<|xIVo7GQ
zLShn-Tb!R(lABn;zyh?C?aAI)E~qo)V9wc;$;iMPu4T6!Nbi+Pwuw0k;WO7UfyKo{
MnZe?wvdmyU0Ba64w*UYD

literal 268
zcmZQ$U|_J&Gu1O-U<9%lazR9BXsCk_h=ByA1KA3?x*3VZ8S%xbsVNEuKwe>KVoFhB
zNoKM_ViJ&>oD`p0QIL~glp3Fyk_HsWNz6@3NmS4^0<sH=GZhShba8%KNp4~VL^(tm
zP+v)9L8?M#K`}_UI6pBlJ_l$(ZfQ<QW=>{aDqOTEF)tg)0g7cSWTvF%m1LGwGOz%h
z=C2aA0Lirw=B7+WhRsHSw%dWUfZt`Cn4^pgDOL0AnCqAr)}*%V787M=n9;Fer>QJ6
J!<4MsRsbj}M<)OP

diff --git a/vowpalwabbit/slim/test/data/cb_data_6.pred b/vowpalwabbit/slim/test/data/cb_data_6.pred
index 7f05b1bbb53..3a9347e2e86 100644
--- a/vowpalwabbit/slim/test/data/cb_data_6.pred
+++ b/vowpalwabbit/slim/test/data/cb_data_6.pred
@@ -1,4 +1,4 @@
-2:0.337614,1:0.333315,0:0.329071
+0:0.6647,1:0.244998,2:0.090302
 
-2:0.410971,1:0.327711,0:0.261319
+0:0.866813,1:0.11731,2:0.015876
 
diff --git a/vowpalwabbit/slim/test/data/cb_data_7.model b/vowpalwabbit/slim/test/data/cb_data_7.model
index ffb0fa232617f3f86155d6f459d0c4f989b381a7..3cfd196c90d8d562f74eece93bee7b3af6e1a835 100644
GIT binary patch
literal 346
zcmZQ$U|_J+v(z(SU<9%lazO+GLxa5#gaId8fh+}G-K4~H1!EwcoD`p!k_KU<Rutsq
z7o{QuN-7Id6>>|8K+?teiHY$!K&82*IVG7nnR%&j(W1n>Y#^twG%=+pu_QBDAu$Qa
z1&XHR=f)SOrlu&En_5^Jm>C!v8Jd_G8yFcHnHVsz03DLkDx!E4>NI~QC?g_-fnnbq
zO`Av{J!P$)O%#x}=}We`Qo_KnOLB(IRUrK|aIeiZAiZjpmfeYEj0`6NY-~>g=`TTH
zwx@tJNc;|v)~tVJa~DWIwR>W74@iT|;rYSHAfNrumKR7HGBMln0cnu>0#Rm$nEUJP
Q3W2nS{%X4-APo`+0Mh(m<NyEw

literal 361
zcmZQ$U|_J&Gu1O-U<9%lazR9BXsCk_h=ByA0NDz<x*3VZ8S%xbsVNEuKwe>KVoFhB
zNoKM_ViJ&>oD`p0QIL~glp3Fyk_Hq=N=#QU2GTi+xk)LB3Wh*tL2)LS2C9Xq0WwM|
z3sMy_3yMKT6z3-<#^(Ta=a%M_Waeb%rNTvv67#Y_hGhb&%#_r;lFX7y1{R>Zn!P*s
zAUPGnj0geJeyjFG0_m4o>-IzeX}yf4d#;o)Fw9^1dG}Qyy?)o%-PeHh_9>csPAp?&
zu<W~Ua}r3;+j!sR6p((V#&3HENOy}bu)PbUFCAWFdk;uY3!7`p^MjFLPhR~#ULZaD
jQqw*@Ae|#sv9CasnZZl1z`hVjSF4uT7Xj%x>^b%TvhZ#<

diff --git a/vowpalwabbit/slim/test/data/cb_data_7.pred b/vowpalwabbit/slim/test/data/cb_data_7.pred
index dd30d21e58d..57b426532cc 100644
--- a/vowpalwabbit/slim/test/data/cb_data_7.pred
+++ b/vowpalwabbit/slim/test/data/cb_data_7.pred
@@ -1,4 +1,4 @@
-2:1,1:0,0:0
+0:1,1:0,2:0
 
-2:1,1:0,0:0
+0:1,1:0,2:0
 
diff --git a/vowpalwabbit/slim/test/data/cb_data_8.model b/vowpalwabbit/slim/test/data/cb_data_8.model
index 7cbdf184d7f035095f5652b4986229d76861f48b..c2a367b22466e4adcc7373f232eb8c9d4faf684c 100644
GIT binary patch
literal 471
zcmZQ$U|_J+v(z(SU<9%lazO+GLxa5#gaIcP0a*&Vx=D%Y3Z_6hIVnCdB@MzzttiOJ
zFG@uSlvEa^D&&?FfuxJ`6BFZefJ$>qb4oIEGV@a5qD6^$*+5QeL2+hIex8DXo{_l$
z2pAfe8(CNwn*rqtOA}Ly5=%0Z6%vzxT%eAW{M`8B)YKFOLo;&&OCt*dQ;=$7V`EcO
zGXn+|pexIxx1G2Eb$0<1l(At21A|Rpvdu;yz2dKq%_bl{Wv!mgW~lfUD81q(1H-CS
zT6Qae^zY`)HmiX2)4;tptD)j+p!5SRCa}7PK>ESNHFl4Hw1)m_yT?%RCs2BVFEd!(
zL?F$f=Wja+NPh_mvz-hTp8}<K0BMl>b^__E;?Hb$0qLi9Pi%HW#rHsIi4PF-B!RTU
TK1Mq!AZ^IRY$pvBmx0m%k0p$k

literal 472
zcmYk$IY<LR6b9hYh@A=|#1j_ml!PdVS`6zA77BuCL<p1YY8E%iCcBCe3^tM!f;SP<
zD556D!i(TBY-6QhA$B&Ng$jzDt$!56Ddzj%d%R)t2_a4P2D^jgAQ4~oayp$Y#o6%h
zY#<9Zo0sWcmoCea;6N1|Wm1S4ibr5>lpeQB9@o@BNOm!42n{zI78;<e>53Zg3l4jI
zBTAL|+!7P&5Ha`>tG_D<Mp%;tMbk0%=mExDDjxHVs)nK}e)+E(V*ZirV+v3t*>5OD
znB?J``(HNezn}lcy(A&Q(gfd!lLZ-m0Q;k7{1CR$G(UpvAL~5%N=Rm`iKn1&(9Cz>
zKvySE!$@H_--YX|1N^=wm-O{|#0S{(8Wta6$yG#rf~}XD_zYi8yy9Gsg^a|lbRJ#@
ztaJeu-Bi&<_@1n$ORytZOON4fG)_-o&B_)%g%fuvdIm41*C@|gm7t}c7IM4oHOt_6
WEM}I&<;@wh0uHKEW+i<444L2a2bcx`

diff --git a/vowpalwabbit/slim/test/data/cb_data_8.pred b/vowpalwabbit/slim/test/data/cb_data_8.pred
index cad4def2485..c04ca26f912 100644
--- a/vowpalwabbit/slim/test/data/cb_data_8.pred
+++ b/vowpalwabbit/slim/test/data/cb_data_8.pred
@@ -1,4 +1,4 @@
-2:0.82,1:0.09,0:0.09
+0:0.82,1:0.09,2:0.09
 
-2:0.82,1:0.09,0:0.09
+0:0.82,1:0.09,2:0.09
 
diff --git a/vowpalwabbit/slim/test/data/cb_data_9.model b/vowpalwabbit/slim/test/data/cb_data_9.model
index 483fe59757312e3ca9501ca40ab79dd096b117a8..885283780f1e90f06ba84cbb2c6182e2dea32926 100644
GIT binary patch
literal 283
zcmZQ$U|_J+v(z(SU<9%la?J`3?bnP;au5RXkU$xbt)Q!$oD`p!k_KU<Rutsq7o{Qu
zN-7Id6;g^ovc>s{iSao=mAR!kC7C&yd8u&GqQtyxASbn;I5Q_dPr*RX*Z>R+4K0lf
zER8IIa)qUdDMg7TnaK)?NenDNoB1zW2O>EJ!eslyz~Gm*z?L0I8!TO5tE0ulFzN19
zYh55c!SI@OmK75N+u|vEvVrvJsHuBWE;BJC2rKVN1=5@+<@e+P=^uwgZ1aJ%@@f&=
T9bL=}X)5>ab^__;4EOB-Jh@G{

literal 286
zcmZQ$U|_J&Gu1O-U<9%lazR9BXsCk_h=BxV0@(_>x*3VZ8S%xbsVNEuKwe>KVoFhB
zNoKM_ViJ&>oD`p0QIL~glp3Fyk_Hq=Ehx^+$<I?T&@%?|auRcsQW6ymflQz{m<H;A
zs01=fDhpB-Qi?$46z3-<#^(Ty$Sut&$;`>jONEOTCFW&=%*q5(nJKAxC7C6a3@kv$
zY2})M+=vR;{xC4aO6XXy1L^sa&y01nm>3>g^_gW^F)^$U=Gu{RnTa9EV~Ir`kmjDR
VWxJz`nZf1jf4iMP`tSdGdjR1%O<({3

diff --git a/vowpalwabbit/slim/test/data/generate-data.sh b/vowpalwabbit/slim/test/data/generate-data.sh
old mode 100644
new mode 100755
index ad4adbd996b..e8b1b65c462
--- a/vowpalwabbit/slim/test/data/generate-data.sh
+++ b/vowpalwabbit/slim/test/data/generate-data.sh
@@ -4,96 +4,101 @@ VW=${1:-vw}
 
 echo > $DATA_H
 
-$VW --quiet -d regression_data_1.txt -f regression_data_1.model -c -k --passes 100 --holdout_off
+$VW --quiet -d regression_data_1.txt -f regression_data_1.model -c -k --passes 100 --holdout_off --predict_only_model
 $VW --quiet -d regression_data_1.txt -i regression_data_1.model -t -p regression_data_1.pred
 
 xxd -i regression_data_1.model >> $DATA_H
 xxd -i regression_data_1.pred  >> $DATA_H
 
-$VW --quiet -d regression_data_1.txt -f regression_data_no_constant.model -c -k --passes 100 --holdout_off --noconstant
+$VW --quiet -d regression_data_1.txt -f regression_data_no_constant.model -c -k --passes 100 --holdout_off --noconstant --predict_only_model
 $VW --quiet -d regression_data_1.txt -i regression_data_no_constant.model -t -p regression_data_no_constant.pred
 
 xxd -i regression_data_no_constant.model >> $DATA_H
 xxd -i regression_data_no_constant.pred  >> $DATA_H
 
-$VW --quiet -d regression_data_2.txt -f regression_data_ignore_linear.model -c -k --passes 100 --holdout_off --ignore_linear a --ignore_linear b
+$VW --quiet -d regression_data_2.txt -f regression_data_ignore_linear.model -c -k --passes 100 --holdout_off --ignore_linear a --ignore_linear b --predict_only_model
 $VW --quiet -d regression_data_2.txt -i regression_data_ignore_linear.model -t -p regression_data_ignore_linear.pred
 
 xxd -i regression_data_ignore_linear.model >> $DATA_H
 xxd -i regression_data_ignore_linear.pred  >> $DATA_H
 
-$VW --quiet -d regression_data_2.txt -f regression_data_2.model -c -k --passes 100 --holdout_off
+$VW --quiet -d regression_data_2.txt -f regression_data_2.model -c -k --passes 100 --holdout_off --predict_only_model
 $VW --quiet -d regression_data_2.txt -i regression_data_2.model -t -p regression_data_2.pred
 
 xxd -i regression_data_2.model >> $DATA_H
 xxd -i regression_data_2.pred  >> $DATA_H
 
 # testing interactions order 2
-$VW --quiet -d regression_data_3.txt -f regression_data_3.model -c -k --passes 100 --holdout_off -q ab
+$VW --quiet -d regression_data_3.txt -f regression_data_3.model -c -k --passes 100 --holdout_off -q ab --predict_only_model
 $VW --quiet -d regression_data_3.txt -i regression_data_3.model -t -p regression_data_3.pred
 
 xxd -i regression_data_3.model >> $DATA_H
 xxd -i regression_data_3.pred  >> $DATA_H
 
 # testing interactions order 3
-$VW --quiet -d regression_data_4.txt -f regression_data_4.model -c -k --passes 2 --holdout_off --interactions abc
+$VW --quiet -d regression_data_4.txt -f regression_data_4.model -c -k --passes 2 --holdout_off --interactions abc --predict_only_model
 $VW --quiet -d regression_data_4.txt -i regression_data_4.model -t -p regression_data_4.pred
 
 xxd -i regression_data_4.model >> $DATA_H
 xxd -i regression_data_4.pred  >> $DATA_H
 
 # testing interactions order 4
-$VW --quiet -d regression_data_4.txt -f regression_data_5.model -c -k --passes 2 --holdout_off --interactions abcd
+$VW --quiet -d regression_data_4.txt -f regression_data_5.model -c -k --passes 2 --holdout_off --interactions abcd --predict_only_model
 $VW --quiet -d regression_data_4.txt -i regression_data_5.model -t -p regression_data_5.pred
 
 xxd -i regression_data_5.model >> $DATA_H
 xxd -i regression_data_5.pred  >> $DATA_H
 
 # testing large models
-$VW --quiet -d regression_data_3.txt -f regression_data_6.model -c -k --passes 2 --holdout_off -q ab -b 33 --sparse_weights
+$VW --quiet -d regression_data_3.txt -f regression_data_6.model -c -k --passes 2 --holdout_off -q ab -b 33 --sparse_weights --predict_only_model
 $VW --quiet -d regression_data_3.txt -i regression_data_6.model -t -p regression_data_6.pred --sparse_weights
 
 xxd -i regression_data_6.model >> $DATA_H
 xxd -i regression_data_6.pred  >> $DATA_H
 
 # testing feature hashing
-$VW --quiet -d regression_data_7.txt -f regression_data_7.model -c -k --passes 2 --holdout_off
+$VW --quiet -d regression_data_7.txt -f regression_data_7.model -c -k --passes 2 --holdout_off --predict_only_model
 $VW --quiet -d regression_data_7.txt -i regression_data_7.model -t -p regression_data_7.pred
 
 xxd -i regression_data_7.model >> $DATA_H
 xxd -i regression_data_7.pred  >> $DATA_H
 
 # multi-class classification
-$VW --quiet -d multiclass_data_4.txt --csoaa_ldf m --csoaa_rank -q ab -k -c --holdout_off --passes 100 -f multiclass_data_4.model
+$VW --quiet -d multiclass_data_4.txt --csoaa_ldf m --csoaa_rank -q ab -k -c --holdout_off --passes 100 -f multiclass_data_4.model --predict_only_model
 $VW --quiet -d multiclass_data_4.txt -i multiclass_data_4.model -t -p multiclass_data_4.pred
-# manual creation of (sort by action id) multiclass_data_4.pred.manual
 
 xxd -i multiclass_data_4.model >> $DATA_H
 xxd -i multiclass_data_4.pred >> $DATA_H
 
+$VW --quiet -d multiclass_data_5.txt --csoaa_ldf m --csoaa_rank -k -c --holdout_off --passes 100 -f multiclass_data_5.model --predict_only_model
+$VW --quiet -d multiclass_data_5.txt -i multiclass_data_5.model -t -p multiclass_data_5.pred
+
+xxd -i multiclass_data_5.model >> $DATA_H
+xxd -i multiclass_data_5.pred >> $DATA_H
+
 # epsilon greedy
-$VW --quiet -d cb_data_5.txt --cb_explore_adf --epsilon 0.3 -q ab -k -c --holdout_off --passes 100 -f cb_data_5.model
+$VW --quiet -d cb_data_5.txt --cb_explore_adf --epsilon 0.3 -q ab -k -c --holdout_off --passes 100 -f cb_data_5.model --predict_only_model
 $VW --quiet -d cb_data_5.txt -i cb_data_5.model -p cb_data_5.pred
 
 xxd -i cb_data_5.model >> $DATA_H
 xxd -i cb_data_5.pred >> $DATA_H
 
 # softmax
-$VW --quiet -d cb_data_5.txt --cb_explore_adf --softmax --lambda -2 -q ab -k -c --holdout_off --passes 5 -f cb_data_6.model
+$VW --quiet -d cb_data_5.txt --cb_explore_adf --softmax --lambda -2 -q ab -k -c --holdout_off --passes 5 -f cb_data_6.model --predict_only_model
 $VW --quiet -d cb_data_5.txt -i cb_data_6.model -p cb_data_6.pred
 
 xxd -i cb_data_6.model >> $DATA_H
 xxd -i cb_data_6.pred >> $DATA_H
 
 # bag
-$VW --quiet -d cb_data_5.txt --cb_explore_adf --bag 3 -q ab -k -c --holdout_off --passes 100 -f cb_data_7.model
+$VW --quiet -d cb_data_5.txt --cb_explore_adf --bag 3 -q ab -k -c --holdout_off --passes 100 -f cb_data_7.model --predict_only_model
 $VW --quiet -d cb_data_5.txt -i cb_data_7.model -p cb_data_7.pred
 
 xxd -i cb_data_7.model >> $DATA_H
 xxd -i cb_data_7.pred >> $DATA_H
 
 # bag + epsilon greedy
-$VW --quiet -d cb_data_5.txt --cb_explore_adf --bag 5 --epsilon 0.27 -q ab -k -c --holdout_off --passes 100 -f cb_data_8.model
+$VW --quiet -d cb_data_5.txt --cb_explore_adf --bag 5 --epsilon 0.27 -q ab -k -c --holdout_off --passes 100 -f cb_data_8.model --predict_only_model
 $VW --quiet -d cb_data_5.txt -i cb_data_8.model -p cb_data_8.pred
 
 xxd -i cb_data_8.model >> $DATA_H
@@ -101,7 +106,7 @@ xxd -i cb_data_8.pred >> $DATA_H
 
 # epsilon greedy + dr
 # TODO: results are confusing (for both dr & mtr) as they don't match ips in a simple example
-$VW --quiet -d cb_data_5.txt --cb_explore_adf --epsilon 0.3 --cb_type dr -q ab -k -c --holdout_off --passes 1000 -f cb_data_9.model
+$VW --quiet -d cb_data_5.txt --cb_explore_adf --epsilon 0.3 --cb_type dr -q ab -k -c --holdout_off --passes 1000 -f cb_data_9.model --predict_only_model
 $VW --quiet -d cb_data_5.txt -i cb_data_9.model -p cb_data_9.pred
 
 xxd -i cb_data_9.model >> $DATA_H
diff --git a/vowpalwabbit/slim/test/data/multiclass_data_4.model b/vowpalwabbit/slim/test/data/multiclass_data_4.model
index 34351571b78d93cf980d79048c905b8f794ded6e..b994afe6658b862c4ca77134b9edfe2c51d22f64 100644
GIT binary patch
delta 94
zcmZ3-IDwI!je&u|QqOWCyCRD=kU7y)&aJRCF{LQ6Br{ndF^Pc%B=}3%i(ykHBg3P(
m$9CI+^o%LuwlPN`eC9eP21NtmePW`_44MuudrW1S!F&LADH!zt

delta 124
zcmbQhxQ>yXje&u|LeF#}yP~2ekg1@nn~_+Y5nr5|nxbF;<Q0}CrW7TXWF{*lCQa0l
zbI-}l%T~xtNzE(CEU9E*0ctrOZ2FU7Qzj$BteD4k+kv#lbaC65qYyrG9TUSnJ>h*~
QqRb2ujxKvlWtqWz09bt_6aWAK

diff --git a/vowpalwabbit/slim/test/data/multiclass_data_4.pred b/vowpalwabbit/slim/test/data/multiclass_data_4.pred
index 151607bbf2d..60ff4928422 100644
--- a/vowpalwabbit/slim/test/data/multiclass_data_4.pred
+++ b/vowpalwabbit/slim/test/data/multiclass_data_4.pred
@@ -1,2 +1,2 @@
-2:0.0386223,1:0.46983,0:0.901038
+3:0.038605,2:0.46982,1:0.901035
 
diff --git a/vowpalwabbit/slim/test/data/multiclass_data_5.pred b/vowpalwabbit/slim/test/data/multiclass_data_5.pred
index 9d4795324c6..5d98d4b54fb 100644
--- a/vowpalwabbit/slim/test/data/multiclass_data_5.pred
+++ b/vowpalwabbit/slim/test/data/multiclass_data_5.pred
@@ -1,2 +1,2 @@
-0:0.551784,3:0.551784,4:0.560359,7:0.560359,1:0.575381,5:0.592531,2:0.598978,6:0.624703
+3:0.127492,3:0.438331,2:0.461522,2:0.668748,1:0.795552,1:0.795552,1:0.899165,1:0.899165
 
diff --git a/vowpalwabbit/slim/test/data/regression_data_1.model b/vowpalwabbit/slim/test/data/regression_data_1.model
index 3a2f6112659748adee6e38c7ff7128967ab92b32..e60c4fb0010c097b1d173f19786e93ad4d8f4a4e 100644
GIT binary patch
delta 44
vcmYdsV`pPvV6fD)oXD;y#|UIHumG`V-^$}4HiP$+>U}Xs85wL|&9w&rppgl;

delta 74
zcmZ=uVP|7tV6f0Loye|eBM)RM=;~%97H7m4r>3ST7yx-WnR(d?nJKAxC7C6a3@kt;
ZUc4)6ffNHn%arPUF-I90LSN0b2LOAe6O{k}

diff --git a/vowpalwabbit/slim/test/data/regression_data_1.pred b/vowpalwabbit/slim/test/data/regression_data_1.pred
index 961a1add22d..8f23c2b166d 100644
--- a/vowpalwabbit/slim/test/data/regression_data_1.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_1.pred
@@ -1,2 +1,2 @@
-0.988030
-0.005295
+0.988028
+0.005296
diff --git a/vowpalwabbit/slim/test/data/regression_data_2.model b/vowpalwabbit/slim/test/data/regression_data_2.model
index d730bd5c5a401a5e8d5ba0e1c81f8a14e09fb264..449d484133357ad44d177177ccca1098846f69e9 100644
GIT binary patch
delta 60
zcmXR3V`pPvV6fD)oXD=I#|UIHumG`V-^$}4HiOUmEq0qS8G*EkUCdENhBU=#cJbGk
K7&>%L>;(XD%nyzL

delta 90
zcmWG5VP|7tV6f0Loye}}BM)RM=;~%97H7m4r>3ST7yx-WnR(d?nJKAxC7C6a3@kt;
oUc4)6ffNJ7`FC6FHf1sbX%oAcql^p*O4ID(uQ4%P)jhEn005gBX8-^I

diff --git a/vowpalwabbit/slim/test/data/regression_data_3.model b/vowpalwabbit/slim/test/data/regression_data_3.model
index 31245ca1032a1ddb5f333768cec0975eed02b5a2..83e1c26faa4257cf225c01996c3151e89f992101 100644
GIT binary patch
delta 76
zcmZo>Ol4<dU|_J+vz*ATXe$6@D(LDKmL{eYC6;6+D<mc{umHu5F3mDw*p$i0uxgT>
Weauk^&05FAAnBsN&sdfjOalO))DqwT

delta 106
zcmYdoW@Kk$U|_J&Go8q;7_A3nD(LEFBo=4H7pJDCC>Q{Fg{6rpMTsSu$qI={KyFTE
xUbaGJN@`w7W=SOj3s8sil*={@n=%<0{!O&Ak2wmVS?ib>emm*!GnQor(*Wr8AA0}*

diff --git a/vowpalwabbit/slim/test/data/regression_data_3.pred b/vowpalwabbit/slim/test/data/regression_data_3.pred
index d6d02d56323..e6e7359d503 100644
--- a/vowpalwabbit/slim/test/data/regression_data_3.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_3.pred
@@ -1,2 +1,2 @@
-0.804214
-0.119599
+0.804218
+0.119585
diff --git a/vowpalwabbit/slim/test/data/regression_data_4.model b/vowpalwabbit/slim/test/data/regression_data_4.model
index db5bea2f15255807b29b4166b507458727c17908..57c7fe2fae26cbd3493f6b07da94eff251673c15 100644
GIT binary patch
delta 105
zcmbQw*vQDv#=yW}sb@KnT`^Jw$W+kP&CDxFElNx-$;{6yR!B@rW?%t|yW6ff$e^vr
zz@T(wu1$#uBZJublY7g7w0N+y?WRmd2DbFowlPN`G;19bgWa_S)=Afx7_Myhw*>%E
Cg&e#9

delta 135
zcmZo<oX^P4#=yW}p=UagU9sE<$W+kP%}6ZHh%Zh}O;Ioa@-p*EQi~FkOEUBGiWL%*
zl7T`wnR(d?nJKAxC7C6a3@kuhs?mWD7_=1`7}}1_wJ8x{WH4WSa&I|^4sf>Jl*!1j
cDS5SR%uxu<TF1mtadCll(lsUqh8_O40O5KlDF6Tf

diff --git a/vowpalwabbit/slim/test/data/regression_data_4.pred b/vowpalwabbit/slim/test/data/regression_data_4.pred
index 3c459e7518d..9259ea0dbb1 100644
--- a/vowpalwabbit/slim/test/data/regression_data_4.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_4.pred
@@ -1,2 +1,2 @@
-0.635620
-0.117813
+0.635637
+0.117808
diff --git a/vowpalwabbit/slim/test/data/regression_data_5.model b/vowpalwabbit/slim/test/data/regression_data_5.model
index 193591c1cbd752fa04f26a41f6e8c33c2f020c49..3e8c18fcf9974e7d0c89857fc84770adb8ab5e56 100644
GIT binary patch
delta 114
zcmZ3%*u}`s#=yW}sb@KnT`@@%$W+kP&CDxFElNx-$;{6yR!B@rPGMjHN(kR{U&2r#
z!pI=+KV@$@kak{RZhKS-%-?eiNN>3M#d=dFBSY8Od$uu0Av9|p6T_0K3Y(;BObov*
GgKYuVl_LoN

delta 145
zcmeBTT*1iB#=yW}p=UagU9sL6$W+kP%}6ZHh%Zh}O;Ioa@-p*EQi~FkOEUBGiWL%*
zl2d?UIhlFc3YjUXc_o=8l?*IEZQSyk%Na^U7#Vc^r|c~U((UuiZI3E}`FoB5=^IzR
jSZ~T?WT-uV&o<^Lgl4T{Vu-A)ut~bc#IW2Xn9&vhw5Kem

diff --git a/vowpalwabbit/slim/test/data/regression_data_5.pred b/vowpalwabbit/slim/test/data/regression_data_5.pred
index 04ee51f85e8..1aad0fa2718 100644
--- a/vowpalwabbit/slim/test/data/regression_data_5.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_5.pred
@@ -1,2 +1,2 @@
-0.761247
+0.761264
 0.040148
diff --git a/vowpalwabbit/slim/test/data/regression_data_6.model b/vowpalwabbit/slim/test/data/regression_data_6.model
index 566fcac1776f979fb0f1c2c3c93c853fea370241..43c6a9a5dd5a2c30153e8ba1fa327ba07a0bfd40 100644
GIT binary patch
delta 92
zcmbQtSjx`Ez`$UsXE~8w(N_S-RM6EeEKN))N-W7tR!B@@U;&Di@BQw^U@U9>7070I
g@o9o}%+ZY?Cc{p}8@8-<_ojl_An{F^sx~lj072CnNdN!<

delta 122
zcmXS&%*f8hz`$UkXF8ExF<TGFRM6GUNG#5XFHTKOQ7{1V3QH4HiV{mQlNAz^fZUwS
zyljQcl+?VE%#un57NCyVbC2C)FqXCc3S=`V+?ik<b95t!$<QTv!<Mz~-c%49B)%zA
I)dnUG0Q>MF;Q#;t

diff --git a/vowpalwabbit/slim/test/data/regression_data_6.pred b/vowpalwabbit/slim/test/data/regression_data_6.pred
index a2204c96afa..44277405900 100644
--- a/vowpalwabbit/slim/test/data/regression_data_6.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_6.pred
@@ -1,2 +1,2 @@
-0.228718
-0.246401
+0.228760
+0.246454
diff --git a/vowpalwabbit/slim/test/data/regression_data_7.model b/vowpalwabbit/slim/test/data/regression_data_7.model
index 0f2dae856a3b99aa87f82ab5331f9bc8ca16adf9..5584d8794698cf8c9267bf4aec48b4629c6b0b0b 100644
GIT binary patch
delta 60
zcmXR3V`pPvV6fD)oXD=I#|UIHumG`V-^$|*9XA*lPQ0|Si#f^&rY{9CF`UQ~v3pU$
K%<$k~=UxD=l@faZ

delta 90
zcmWG5VP|7tV6f0Loye}}BM)RM=;~%97H7m4r>3ST7yx-WnR(d?nJKAxC7C6a3@kt;
nUc4)689Ht-Fx-A&V;6Ij5lmkSVq(~oC1UrYf|=px;m*APC!icG

diff --git a/vowpalwabbit/slim/test/data/regression_data_7.pred b/vowpalwabbit/slim/test/data/regression_data_7.pred
index f7f8dd0188d..c6638c10565 100644
--- a/vowpalwabbit/slim/test/data/regression_data_7.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_7.pred
@@ -1,2 +1,2 @@
-0.658841
-0.099891
+0.658879
+0.099914
diff --git a/vowpalwabbit/slim/test/data/regression_data_ignore_linear.model b/vowpalwabbit/slim/test/data/regression_data_ignore_linear.model
index 68d808987f7bea25af5a5de93f6547fbb7604a27..b891f0a195f248827a05f242a6f3e098071b245e 100644
GIT binary patch
delta 72
zcmbQhn8?n?z`$UsXE~8w(NYA+RM6GUOwY?NN{!FS%u7uyQb<f<U;&DYFI83pQVa~5
YQ~2#;jxsW=x&FW|{u&d5>{HQw06PH^L;wH)

literal 144
zcmZQ$U|_J&Gu1O-U<9%lazO+GLxa5#gaId=fh+}G-HgQIjQHZz)D#5+ATKjLFTW@?
zJ|{CTHL*w`5kn*iC<2tsR>(|A%`3?)sbpXQ8uMl@(`>N0%O~^O#~fv3*l^>4UHmmB
KhEq>P_W=OFBP56b

diff --git a/vowpalwabbit/slim/test/data/regression_data_ignore_linear.pred b/vowpalwabbit/slim/test/data/regression_data_ignore_linear.pred
index f225669ada0..5e715fea88b 100644
--- a/vowpalwabbit/slim/test/data/regression_data_ignore_linear.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_ignore_linear.pred
@@ -1,2 +1,2 @@
 1.000000
-0.000000
+0
diff --git a/vowpalwabbit/slim/test/data/regression_data_no_constant.model b/vowpalwabbit/slim/test/data/regression_data_no_constant.model
index 1175c06f5446d8f44e010003c1889858e6ec4514..71ed9b947f1a4133feba34b8bc228320650d4e81 100644
GIT binary patch
delta 49
zcma#?W@lqyV6fD)oXD=I%m-vD=<4R>C+FuCmn7zuFt7kcK037i22u<R&*XV+0rZax
A1ONa4

delta 66
zcmeZ{VP|7tV6f0Loye|eA`fIL=;~%97H7m4r>3ST7yx-WnR(d?nJKAxC7C6a3@kt;
QUc4)6ffNHnj~uTp0OkD-@Bjb+

diff --git a/vowpalwabbit/slim/test/data/regression_data_no_constant.pred b/vowpalwabbit/slim/test/data/regression_data_no_constant.pred
index 89b02b9f031..ec553b2b5e9 100644
--- a/vowpalwabbit/slim/test/data/regression_data_no_constant.pred
+++ b/vowpalwabbit/slim/test/data/regression_data_no_constant.pred
@@ -1,2 +1,2 @@
-0.034453
-0.172265
+0.034454
+0.172271
diff --git a/vowpalwabbit/slim/test/ut_util.cc b/vowpalwabbit/slim/test/ut_util.cc
index 431a38d4649..65e8be040b2 100644
--- a/vowpalwabbit/slim/test/ut_util.cc
+++ b/vowpalwabbit/slim/test/ut_util.cc
@@ -8,6 +8,7 @@
 
 #include <stdlib.h>
 
+#include <algorithm>
 #include <array>
 #include <fstream>
 #include <set>
@@ -85,7 +86,6 @@ test_data get_test_data(const char* model_filename)
   TEST_DATA(model_filename, regression_data_ignore_linear);
   TEST_DATA(model_filename, multiclass_data_4);
   TEST_DATA(model_filename, multiclass_data_5);
-  TEST_DATA(model_filename, cb_data_epsilon_0_skype_jb);
   TEST_DATA(model_filename, cb_data_5);
   TEST_DATA(model_filename, cb_data_6);
   TEST_DATA(model_filename, cb_data_7);
@@ -214,7 +214,7 @@ void run_predict_in_memory(
   // compare output
   std::vector<float> preds_expected = read_floats(td.pred, td.pred_len);
 
-  EXPECT_THAT(preds, Pointwise(FloatNear(1e-5f), preds_expected));
+  EXPECT_THAT(preds, Pointwise(FloatNear(1e-4f), preds_expected));
 }
 
 enum class predict_param_weight_type
@@ -343,11 +343,11 @@ TEST_P(invalid_model_test, Run)
 }
 
 invalid_model_param invalid_model_param[] = {
-    {"regression_data_1", regression_data_1_model, regression_data_1_model_len, {84, 92},
+    {"regression_data_1", regression_data_1_model, regression_data_1_model_len, {54, 62},
         predict_param_weight_type::SPARSE},  // 2 weights
-    {"regression_data_1", regression_data_1_model, regression_data_1_model_len, {84, 92},
+    {"regression_data_1", regression_data_1_model, regression_data_1_model_len, {54, 62},
         predict_param_weight_type::DENSE},  // 2 weights
-    {"regression_data_6", regression_data_6_model, regression_data_6_model_len, {99, 111, 123, 135},
+    {"regression_data_6", regression_data_6_model, regression_data_6_model_len, {69, 81, 93, 105},
         predict_param_weight_type::SPARSE}  // 4 weights
 };
 
@@ -384,7 +384,9 @@ TEST(VowpalWabbitSlim, MulticlassData4)
   std::vector<float> preds_expected = {0.901038f, 0.46983f, 0.0386223f};
 
   // compare output
-  EXPECT_THAT(out_scores, Pointwise(FloatNear(1e-5f), preds_expected));
+  std::sort(out_scores.begin(), out_scores.end());
+  std::sort(preds_expected.begin(), preds_expected.end());
+  EXPECT_THAT(out_scores, Pointwise(FloatNear(1e-4f), preds_expected));
 }
 
 TEST(VowpalWabbitSlim, MulticlassData5)
@@ -429,51 +431,13 @@ TEST(VowpalWabbitSlim, MulticlassData5)
 
   ASSERT_EQ(S_VW_PREDICT_OK, vw.predict(shared, ex, 8, out_scores));
 
-  // 0:0.551784,3:0.551784,4:0.560359,7:0.560359,1:0.575381,5:0.592531,2:0.598978,6:0.624703
-  std::vector<float> preds_expected = {
-      0.551784f, 0.575380862f, 0.598977983f, 0.5517838f, 0.560358882f, 0.592531085f, 0.624703348f, 0.560358882f};
+  // 3:0.127492,3:0.438331,2:0.461522,2:0.668748,1:0.795552,1:0.795552,1:0.899165,1:0.899165
+  std::vector<float> preds_expected = {0.127492, 0.438331, 0.461522, 0.668748, 0.795552, 0.795552, 0.899165, 0.899165};
 
   // compare output
-  EXPECT_THAT(out_scores, Pointwise(FloatNear(1e-5f), preds_expected));
-}
-
-void cb_data_epsilon_0_skype_jb_test_runner(int call_type, int modality, int network_type, int platform,
-    const std::vector<int>& ranking_expected, const std::vector<float>& pdf_expected)
-{
-  // load model.
-  vw_predict<VW::sparse_parameters> vw;
-  test_data td = get_test_data("cb_data_epsilon_0_skype_jb");
-  ASSERT_EQ(0, vw.load((const char*)td.model, td.model_len));
-
-  // we have loaded the model and can push the features
-  VW::example_predict features;
-  vw_slim::example_predict_builder bOa(&features, "64");  // NOLINT
-  bOa.push_feature(static_cast<int>(call_type), 1.f);
-  vw_slim::example_predict_builder bOb(&features, "16");  // NOLINT
-  bOb.push_feature(static_cast<int>(modality), 1.f);
-  vw_slim::example_predict_builder bOc(&features, "32");  // NOLINT
-  bOc.push_feature(static_cast<int>(network_type), 1.f);
-  vw_slim::example_predict_builder bOd(&features, "48");  // NOLINT
-  bOd.push_feature(static_cast<int>(platform), 1.f);
-
-  // push actions
-  const int min_delay_actions = 10;
-  VW::example_predict actions[min_delay_actions];
-  for (int i = 0; i < min_delay_actions; i++)
-  {
-    vw_slim::example_predict_builder bOe(&actions[i], "80");  // NOLINT
-    bOe.push_feature(i, 1.f);
-  }
-
-  // predict CB value
-  std::vector<float> pdfs;
-  std::vector<int> rankings;
-  int result = vw.predict("eid", features, actions, min_delay_actions, pdfs, rankings);
-
-  // compare output with expected.
-  EXPECT_EQ(result, 0);
-  EXPECT_THAT(pdfs, Pointwise(FloatNear(1e-5f), pdf_expected));
-  EXPECT_THAT(rankings, Pointwise(Eq(), ranking_expected));
+  std::sort(out_scores.begin(), out_scores.end());
+  std::sort(preds_expected.begin(), preds_expected.end());
+  EXPECT_THAT(out_scores, Pointwise(FloatNear(1e-4f), preds_expected));
 }
 
 TEST(VowpalWabbitSlim, InteractionNumBitsBug)
@@ -522,52 +486,6 @@ TEST(VowpalWabbitSlim, InteractionNumBitsBug)
   EXPECT_EQ(rankings[0], 3);
 }
 
-TEST(VowpalWabbitSlim, CbDataEpsilon0SkypeJb)
-{
-  // Since the model is epsilon=0, the first entry should always be 0.
-  std::vector<float> pdf_expected = {1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0};
-
-  // {0, 1, 2, 0} => 1
-  std::vector<int> ranking_expected = {1, 0, 2, 3, 4, 5, 7, 6, 8, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(0, 1, 2, 0, ranking_expected, pdf_expected);
-
-  // {0, 1, 4, 0} => 1
-  ranking_expected = {1, 0, 2, 3, 4, 5, 6, 8, 7, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(0, 1, 4, 0, ranking_expected, pdf_expected);
-
-  // {0, 0, 2, 0} => 1
-  ranking_expected = {1, 2, 0, 3, 4, 5, 6, 7, 8, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(0, 0, 2, 0, ranking_expected, pdf_expected);
-
-  // {0, 0, 4, 0} => 1
-  ranking_expected = {1, 3, 2, 0, 4, 6, 5, 8, 7, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(0, 0, 4, 0, ranking_expected, pdf_expected);
-
-  // {2, 0, 4, 0} => 3
-  ranking_expected = {3, 1, 0, 2, 5, 4, 7, 6, 8, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(2, 0, 4, 0, ranking_expected, pdf_expected);
-
-  // {0, 1, 999, 0} => 2
-  ranking_expected = {2, 4, 6, 1, 0, 5, 3, 7, 8, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(0, 1, 999, 0, ranking_expected, pdf_expected);
-
-  // {0, 0, 999, 0} => 2
-  ranking_expected = {2, 4, 6, 1, 3, 5, 0, 7, 8, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(0, 0, 999, 0, ranking_expected, pdf_expected);
-
-  // {2, 0, 999, 0} => 2
-  ranking_expected = {2, 5, 4, 3, 6, 1, 0, 7, 8, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(2, 0, 999, 0, ranking_expected, pdf_expected);
-
-  // {999, 0, 4, 0} => 1
-  ranking_expected = {0, 1, 4, 2, 6, 3, 8, 7, 5, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(999, 0, 4, 0, ranking_expected, pdf_expected);
-
-  // {999, 0, 2, 0} => 2
-  ranking_expected = {0, 1, 4, 2, 7, 3, 6, 8, 5, 9};
-  cb_data_epsilon_0_skype_jb_test_runner(999, 0, 2, 0, ranking_expected, pdf_expected);
-}
-
 void clear(VW::example_predict& ex)
 {
   for (auto ns : ex.indices) { ex.feature_space[ns].clear(); }
@@ -670,38 +588,38 @@ TEST_P(cb_predict_test, CBRunPredict)
   //}
 #endif
 
-  EXPECT_THAT(histogram, Pointwise(FloatNear(1e-2f), GetParam().ranking_pdf_expected));
+  EXPECT_THAT(histogram, Pointwise(FloatNear(2e-2f), GetParam().ranking_pdf_expected));
 }
 
 cb_predict_param cb_predict_params[] = {
-    {"CB Epsilon Greedy", "cb_data_5", 10000, {0.1f, 0.1f, 0.8f},
+    {"CB Epsilon Greedy", "cb_data_5", 10000, {0.8f, 0.1f, 0.1f},
         {
-            // see top action 2 w / 0.8
-            0.1f, 0.09f, 0.80f,  // slot 0
-                                 // most of the time we should see action 2
-                                 // in 10% we should see the top action 0 swapped from top-slot to here
-            0.0f, 0.90f, 0.09f,  // slot 1
-                                 // most of the time we should see action 1
-                                 // in 20% we should see the top action 2 swapped from top-slot to here
-            0.89f, 0.0f, 0.10f,  // slot 2
+            0.8f, 0.1f, 0.1f,  // slot 0
+                               // most of the time we should see action 0
+            0.1f, 0.9f, 0.0f,  // slot 1
+                               // most of the time we should see action 1
+                               // in 10% we should see the top action 0 swapped from top-slot to here
+            0.1f, 0.0f, 0.9f,  // slot 2
+                               // most of the time we should see action 2
+                               // in 10% we should see the top action 0 swapped from top-slot to here
         }},
-    {"CB Softmax", "cb_data_6", 1000, {0.329f, 0.333f, 0.337f},
+    {"CB Softmax", "cb_data_6", 1000, {0.664f, 0.245f, 0.090f},
         {
-            0.328f, 0.354f, 0.316f,  // slot 0
-            0.000f, 0.644f, 0.354f,  // slot 1
-            0.671f, 0.000f, 0.328f,  // slot 2
+            0.664f, 0.245f, 0.090f,  // slot 0
+            0.245f, 0.755f, 0.000f,  // slot 1
+            0.090f, 0.000f, 0.910f,  // slot 2
         }},
-    {"CB Bag", "cb_data_7", 5, {0.0f, 0.0f, 1.0f},
+    {"CB Bag", "cb_data_7", 5, {1.0f, 0.0f, 0.0f},
         {
-            0.0f, 0.0f, 1.0f,  // slot 0
+            1.0f, 0.0f, 0.0f,  // slot 0
             0.0f, 1.0f, 0.0f,  // slot 1
-            1.0f, 0.0f, 0.0f,  // slot 2
+            0.0f, 0.0f, 1.0f,  // slot 2
         }},
-    {"CB Bag Epsilon Greedy", "cb_data_8", 10000, {0.09f, 0.09f, 0.82f},
+    {"CB Bag Epsilon Greedy", "cb_data_8", 10000, {0.82f, 0.09f, 0.09f},
         {
-            0.091f, 0.086f, 0.82f,  // slot 0
-            0.00f, 0.91f, 0.086f,   // slot 1
-            0.90f, 0.00f, 0.09f,    // slot 2
+            0.82f, 0.09f, 0.09f,  // slot 0
+            0.09f, 0.91f, 0.00f,  // slot 1
+            0.09f, 0.00f, 0.91f,  // slot 2
         }},
 };
 

From f204897f2f020f3102a8080e3cef0b0531a2ee0a Mon Sep 17 00:00:00 2001
From: Etienne Kintzler <etienne.kintzler@gmail.com>
Date: Thu, 23 May 2024 17:30:50 +0200
Subject: [PATCH 3/3] docs: Simplification of the DFtoVW tutorial (#4693)

* first version of the simplified tutorial

* fix typo + rm dedicated section for df creation

* rename title

* use black linting

* use default kernel

---------

Co-authored-by: Griffin Bassman <griffinbassman@gmail.com>
---
 .../python_simplified_dftovw_tuto.ipynb       | 307 ++++++++++++++++++
 1 file changed, 307 insertions(+)
 create mode 100644 python/docs/source/tutorials/python_simplified_dftovw_tuto.ipynb

diff --git a/python/docs/source/tutorials/python_simplified_dftovw_tuto.ipynb b/python/docs/source/tutorials/python_simplified_dftovw_tuto.ipynb
new file mode 100644
index 00000000000..85f9c1efc99
--- /dev/null
+++ b/python/docs/source/tutorials/python_simplified_dftovw_tuto.ipynb
@@ -0,0 +1,307 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "51f41eaf-f24f-44fc-8178-3270efa46ec4",
+   "metadata": {},
+   "source": [
+    "# Simple pandas to vowpalwabbit conversion tutorial"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "b9a21a43-39ad-4213-9c7f-814bbafd8a54",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "from vowpalwabbit.dftovw import DFtoVW\n",
+    "from vowpalwabbit import Workspace"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "fc831353-b5aa-4bb0-a928-c47b340397a5",
+   "metadata": {},
+   "source": [
+    "### Building simple examples using `DftoVW.from_column_names`"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c60089f1-ce41-49ee-a3a9-74f0fb2cb34f",
+   "metadata": {},
+   "source": [
+    "Let's create the following pandas dataframe:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "a31118c2-b315-4129-b28a-2ea37d2dae50",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df = pd.DataFrame(\n",
+    "    [\n",
+    "        {\n",
+    "            \"income\": 0,\n",
+    "            \"age\": 27,\n",
+    "            \"marital-status\": \"Separated\",\n",
+    "            \"education\": \"HS-grad\",\n",
+    "            \"occupation\": \"Handlers-cleaners\",\n",
+    "            \"hours-per-week\": 25,\n",
+    "        },\n",
+    "        {\n",
+    "            \"income\": 1,\n",
+    "            \"age\": 34,\n",
+    "            \"marital-status\": \"Married-civ-spouse\",\n",
+    "            \"education\": \"Bachelors\",\n",
+    "            \"occupation\": \"Prof-specialty\",\n",
+    "            \"hours-per-week\": 40,\n",
+    "        },\n",
+    "        {\n",
+    "            \"income\": 0,\n",
+    "            \"age\": 44,\n",
+    "            \"marital-status\": \"Never-married\",\n",
+    "            \"education\": \"Assoc-voc\",\n",
+    "            \"occupation\": \"Priv-house-serv\",\n",
+    "            \"hours-per-week\": 25,\n",
+    "        },\n",
+    "        {\n",
+    "            \"income\": 1,\n",
+    "            \"age\": 38,\n",
+    "            \"marital-status\": \"Married-civ-spouse\",\n",
+    "            \"education\": \"Bachelors\",\n",
+    "            \"occupation\": \"Prof-specialty\",\n",
+    "            \"hours-per-week\": 60,\n",
+    "        },\n",
+    "        {\n",
+    "            \"income\": 0,\n",
+    "            \"age\": 34,\n",
+    "            \"marital-status\": \"Married-civ-spouse\",\n",
+    "            \"education\": \"HS-grad\",\n",
+    "            \"occupation\": \"Other-service\",\n",
+    "            \"hours-per-week\": 36,\n",
+    "        },\n",
+    "    ]\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "473e5c72-ab6c-4d72-a466-7352ec604393",
+   "metadata": {},
+   "source": [
+    "The user builds the examples using the class method `DftoVW.from_column_names`. The method is called using the dataframe object (`df`) and its various column names. The conversion to vowpal wabbit examples is then performed by calling the `convert_df` method:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "2be83f6c-ecaa-45cb-bb3f-2f47827d6016",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['0 | age:27 marital-status=Separated education=HS-grad occupation=Handlers-cleaners hours-per-week:25',\n",
+       " '1 | age:34 marital-status=Married-civ-spouse education=Bachelors occupation=Prof-specialty hours-per-week:40',\n",
+       " '0 | age:44 marital-status=Never-married education=Assoc-voc occupation=Priv-house-serv hours-per-week:25',\n",
+       " '1 | age:38 marital-status=Married-civ-spouse education=Bachelors occupation=Prof-specialty hours-per-week:60',\n",
+       " '0 | age:34 marital-status=Married-civ-spouse education=HS-grad occupation=Other-service hours-per-week:36']"
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "converter = DFtoVW.from_column_names(\n",
+    "    df=df,\n",
+    "    y=\"income\",\n",
+    "    x=[\"age\", \"marital-status\", \"education\", \"occupation\", \"hours-per-week\"],\n",
+    ")\n",
+    "examples = converter.convert_df()\n",
+    "examples"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "6109f95e-cd17-485b-947d-8c2c33a5843a",
+   "metadata": {},
+   "source": [
+    "Note that the vowpal wabbit format for categorical features is `feature_name=feature_value` whereas for numerical features the format is `feature_name:feature_value`. When using `DFtoVW` class, the appropriate format will be inferred from the dataframe columns types.\n",
+    "\n",
+    "We then train the model on these examples:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "c0269980-78b3-4123-84eb-27e0fba929b4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model = Workspace(P=1, enable_logging=True)\n",
+    "\n",
+    "for ex in examples:\n",
+    "    model.learn(ex)\n",
+    "model.finish()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "50470ca2-f33d-495e-a3f9-46ae1a618e6d",
+   "metadata": {},
+   "source": [
+    "### Building more complex examples"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "30a526a6-7f8f-48e4-8dca-f9058a0d87fb",
+   "metadata": {},
+   "source": [
+    "The class method `DFtoVW.from_column_names` represents a quick and simple way to build the examples, but if the user needs more control over the way the examples are created, she or he can either use the class `Feature` or the class `Namespace` for building features, and any of the label class available (see below) based on the nature of the task. \n",
+    "\n",
+    "- When using `Namespace` class (see https://github.com/VowpalWabbit/vowpal_wabbit/wiki/Namespaces for the meaning) the user specifies the name of the namespace with the `name` field, and will pass one or a list of `Feature` object to the `features` field.\n",
+    "\n",
+    "- The `Feature` class has a `value` field, which is the name of the column. The user can also rename the feature using the `rename_feature` field or choose to enforce a specific type (`\"numerical\"` or `\"categorical\"`) using `as_type` field.\n",
+    "\n",
+    "Regarding the labels, multiple classes are available:\n",
+    "- `SimpleLabel` for regression\n",
+    "- `MulticlassLabel` and `Multilabel` for classification\n",
+    "- `ContextualbanditLabel`.\n",
+    "\n",
+    "In the following examples we'll build 2 namespaces based on socio-demographic features and the job features."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "90a69d90-a0a6-42d4-8867-5d1b0e73f4ec",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['0 |ns_sociodemo age:27 marital-status=Separated education=HS-grad |ns_job occupation=Handlers-cleaners hours-per-week:25',\n",
+       " '1 |ns_sociodemo age:34 marital-status=Married-civ-spouse education=Bachelors |ns_job occupation=Prof-specialty hours-per-week:40',\n",
+       " '0 |ns_sociodemo age:44 marital-status=Never-married education=Assoc-voc |ns_job occupation=Priv-house-serv hours-per-week:25',\n",
+       " '1 |ns_sociodemo age:38 marital-status=Married-civ-spouse education=Bachelors |ns_job occupation=Prof-specialty hours-per-week:60',\n",
+       " '0 |ns_sociodemo age:34 marital-status=Married-civ-spouse education=HS-grad |ns_job occupation=Other-service hours-per-week:36']"
+      ]
+     },
+     "execution_count": 5,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "from vowpalwabbit.dftovw import SimpleLabel, Namespace, Feature\n",
+    "\n",
+    "ns_sociodemo = Namespace(\n",
+    "    features=[Feature(col) for col in [\"age\", \"marital-status\", \"education\"]],\n",
+    "    name=\"ns_sociodemo\",\n",
+    ")\n",
+    "ns_job = Namespace(\n",
+    "    features=[Feature(col) for col in [\"occupation\", \"hours-per-week\"]], name=\"ns_job\"\n",
+    ")\n",
+    "label = SimpleLabel(\"income\")\n",
+    "\n",
+    "converter_advanced = DFtoVW(df=df, namespaces=[ns_sociodemo, ns_job], label=label)\n",
+    "examples_advanced = converter_advanced.convert_df()\n",
+    "examples_advanced[:5]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "071326d7-f969-4db1-a73e-3cee225921f4",
+   "metadata": {},
+   "source": [
+    "We train the model by also including interactions between the variables of the 2 namespaces:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "f0ed661f-d9a0-4ebb-93b8-f5747347c7b4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model_advanced = Workspace(\n",
+    "    # arg_str=\"--interactions ns_sociodemo:ns_job\", P=1, enable_logging=True\n",
+    "    arg_str=\"--redefine a:=ns_job b:=ns_sociodemo -q ab \",\n",
+    "    P=1,\n",
+    "    enable_logging=True,\n",
+    ")\n",
+    "\n",
+    "for ex in examples_advanced:\n",
+    "    model_advanced.learn(ex)\n",
+    "\n",
+    "model_advanced.finish()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "5bb2208e-9d0e-44ef-8d91-faccedf41ac0",
+   "metadata": {},
+   "source": [
+    "Finally, we can get the estimated weights associated to each namespace and feature:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "06aabeab-2365-4f86-bf60-7043b0e59190",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "[('ns_job', 'occupation', 0.0),\n",
+       " ('ns_job', 'hours-per-week', 0.0019117757910862565),\n",
+       " ('ns_sociodemo', 'age', 0.001858704723417759),\n",
+       " ('ns_sociodemo', 'marital-status', 0.0),\n",
+       " ('ns_sociodemo', 'education', 0.0)]"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "[\n",
+    "    (ns.name, feature.name, model_advanced.get_weight_from_name(feature.name, ns.name))\n",
+    "    for ns in [ns_job, ns_sociodemo]\n",
+    "    for feature in ns.features\n",
+    "]"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.10.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}