Update the requirements, fix the notebook and improve the readme

.gitignore

@@ -1,6 +1,6 @@
-tmp_encrypted_prediction.npy
-tmp_encrypted_quantized_encoding.npy
-tmp_evaluation_key.npy
+tmp/
 .venv
 .fhe_keys
 *.pyc
+local_datasets/
+.vscode/

README.md

@@ -13,11 +13,7 @@ python_version: 3.9
 
 # Sentiment Analysis With FHE
 
-## Running the application on your machine
-
-In this directory, ie `sentiment-analysis-with-transformer`, you can do the following steps.
-
-### Do once
+## Set up the app locally
 
 - First, create a virtual env and activate it:
 
@@ -34,43 +30,36 @@ pip3 install -U pip wheel setuptools --ignore-installed
 pip3 install -r requirements.txt --ignore-installed
 ```
 
-- If not on Linux, or if you want to compile the FHE algorithms by yourself:
+- (optional) Compile the FHE algorithm:
 
 ```bash
 python3 compile.py
 ```
 
-Check it finish well (with a "Done!").
-
-### Do each time you relaunch the application
-
-- Then, in a terminal Tab 1:
-
-```bash
-source .venv/bin/activate
-uvicorn server:app
-```
+Check it finish well (with a "Done!"). Please note that the actual model initialization and training 
+can be found in the [SentimentClassification notebook](SentimentClassification.ipynb) (see below).
 
-Tab 1 will be for the Server side.
+### Launch the app locally
 
-- And, in another terminal Tab 2:
+- In a terminal:
 
 ```bash
 source .venv/bin/activate
 python3 app.py
 ```
 
-Tab 2 will be for the Client side.
-
-## Interacting with the application
-
-Open the given URL link (search for a line like `Running on local URL:  http://127.0.0.1:8888/` in your Terminal 2).
+## Interact with the application
 
-## Training a new model
+Open the given URL link (search for a line like `Running on local URL:  http://127.0.0.1:8888/` in the 
+terminal).
 
-The notebook SentimentClassification.ipynb provides a way to train a new model.
+## Train a new model
 
-Before running the notebook, you need to download the data.
+The notebook [SentimentClassification notebook](SentimentClassification.ipynb) provides a way to 
+train a new model. Be aware that the data needs to be downloaded beforehand using the 
+[download_data.sh](download_data.sh) file (which requires Kaggle CLI). 
+Alternatively, the dataset can be downloaded manually at 
+https://www.kaggle.com/datasets/crowdflower/twitter-airline-sentiment
 
 ```bash
 bash download_data.sh

SentimentClassification.ipynb

@@ -21,16 +21,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 31,
    "metadata": {},
    "outputs": [],
    "source": [
     "# Import the required packages\n",
     "import os\n",
     "import time\n",
+    "from pathlib import Path\n",
     "\n",
     "import numpy\n",
-    "import onnx\n",
     "import pandas as pd\n",
     "from sklearn.metrics import average_precision_score\n",
     "from sklearn.model_selection import GridSearchCV, train_test_split\n",
@@ -40,7 +40,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [
     {
@@ -76,7 +76,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -105,7 +105,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -123,7 +123,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -135,55 +135,55 @@
     "    \"n_bits\": [2, 3],\n",
     "    \"max_depth\": [1],\n",
     "    \"n_estimators\": [10, 30, 50],\n",
-    "    \"n_jobs\": [-1],\n",
+    "    # \"n_jobs\": [-1],\n",
     "}"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 21,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/html": [
-       "<style>#sk-container-id-3 {color: black;background-color: white;}#sk-container-id-3 pre{padding: 0;}#sk-container-id-3 div.sk-toggleable {background-color: white;}#sk-container-id-3 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-3 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-3 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-3 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-3 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-3 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-3 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-3 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-3 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-3 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-3 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-3 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-3 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-3 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-3 div.sk-item {position: relative;z-index: 1;}#sk-container-id-3 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-3 div.sk-item::before, #sk-container-id-3 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-3 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-3 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-3 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-3 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-3 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-3 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-3 div.sk-label-container {text-align: center;}#sk-container-id-3 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-3 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-3\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(), n_jobs=1,\n",
+       "<style>#sk-container-id-1 {color: black;background-color: white;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(n_jobs=1),\n",
        "             param_grid={&#x27;max_depth&#x27;: [1], &#x27;n_bits&#x27;: [2, 3],\n",
-       "                         &#x27;n_estimators&#x27;: [10, 30, 50], &#x27;n_jobs&#x27;: [-1]},\n",
-       "             scoring=&#x27;accuracy&#x27;)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-7\" type=\"checkbox\" ><label for=\"sk-estimator-id-7\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">GridSearchCV</label><div class=\"sk-toggleable__content\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(), n_jobs=1,\n",
+       "                         &#x27;n_estimators&#x27;: [10, 30, 50]},\n",
+       "             scoring=&#x27;accuracy&#x27;)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">GridSearchCV</label><div class=\"sk-toggleable__content\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(n_jobs=1),\n",
        "             param_grid={&#x27;max_depth&#x27;: [1], &#x27;n_bits&#x27;: [2, 3],\n",
-       "                         &#x27;n_estimators&#x27;: [10, 30, 50], &#x27;n_jobs&#x27;: [-1]},\n",
-       "             scoring=&#x27;accuracy&#x27;)</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-8\" type=\"checkbox\" ><label for=\"sk-estimator-id-8\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">estimator: XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier()</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-9\" type=\"checkbox\" ><label for=\"sk-estimator-id-9\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier()</pre></div></div></div></div></div></div></div></div></div></div>"
+       "                         &#x27;n_estimators&#x27;: [10, 30, 50]},\n",
+       "             scoring=&#x27;accuracy&#x27;)</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-2\" type=\"checkbox\" ><label for=\"sk-estimator-id-2\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">estimator: XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier(n_jobs=1)</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" ><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier(n_jobs=1)</pre></div></div></div></div></div></div></div></div></div></div>"
       ],
       "text/plain": [
-       "GridSearchCV(cv=3, estimator=XGBClassifier(), n_jobs=1,\n",
+       "GridSearchCV(cv=3, estimator=XGBClassifier(n_jobs=1),\n",
        "             param_grid={'max_depth': [1], 'n_bits': [2, 3],\n",
-       "                         'n_estimators': [10, 30, 50], 'n_jobs': [-1]},\n",
+       "                         'n_estimators': [10, 30, 50]},\n",
        "             scoring='accuracy')"
       ]
      },
-     "execution_count": 21,
+     "execution_count": 6,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
     "# Run the gridsearch\n",
-    "grid_search = GridSearchCV(model, parameters, cv=3, n_jobs=1, scoring=\"accuracy\")\n",
+    "grid_search = GridSearchCV(model, parameters, cv=3, scoring=\"accuracy\")\n",
     "grid_search.fit(X_train, y_train)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Best score: 0.6842744383727991\n",
-      "Best parameters: {'max_depth': 1, 'n_bits': 3, 'n_estimators': 50, 'n_jobs': -1}\n"
+      "Best score: 0.705980570734669\n",
+      "Best parameters: {'max_depth': 1, 'n_bits': 3, 'n_estimators': 50}\n"
      ]
     }
    ],
@@ -200,17 +200,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Accuracy: 0.6810\n",
-      "Average precision score for positive class: 0.5615\n",
-      "Average precision score for negative class: 0.8349\n",
-      "Average precision score for neutral class: 0.3820\n"
+      "Accuracy: 0.7117\n",
+      "Average precision score for positive class: 0.6404\n",
+      "Average precision score for negative class: 0.8719\n",
+      "Average precision score for neutral class: 0.4349\n"
      ]
     }
    ],
@@ -238,7 +238,36 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 48,
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array([2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,\n",
+       "       2, 2, 2, 2, 2, 2])"
+      ]
+     },
+     "execution_count": 9,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "y_pred_test_tfidf[y_pred_test_tfidf == 2]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
    "metadata": {},
    "outputs": [
     {
@@ -246,18 +275,18 @@
      "output_type": "stream",
      "text": [
       "5 most positive tweets (class 2):\n",
-      "@united sent a DM just now. Thanks I am incredibly happy the fast response I got via Twitter than via customer care. Thank you\n",
-      "@JetBlue Great Thank you, lets hope so! Could you please notify me if flight 2302 leaves JFK? Thank you again\n",
-      "@AmericanAir Great, thanks. Followed.\n",
-      "@SouthwestAir I continue to be amazed by the amazing customer service.  Thank you SWA!\n",
-      "@JetBlue Awesome thanks! Thanks for the quick response. You guys ROCK! :)\n",
+      "@JetBlue do bags still fly free or have you started charging? thanks!\n",
+      "@SouthwestAir Is there a way to receive a refund on a trip that was Cancelled Flight online instead of calling? Your phone lines are super busy.\n",
+      "@JetBlue bag is supposedly here in Boston\n",
+      "@AmericanAir Cancelled Flights my flight, doesn't send an email, text or call. Now I'm stranded in Louisville.\n",
+      "@SouthwestAir I need to Cancelled Flight one leg of a flight, but can't seem to do this online. Been on hold on the phone for 10 minutes. Any help?\n",
       "----------------------------------------------------------------------------------------------------\n",
       "5 most negative tweets (class 0):\n",
-      "@USAirways been on hold 2 hours for a Cancelled Flighted flight. I understand the delay. I don't understand you auto-reFlight Booking Problems me on TUESDAY. HELP!\n",
-      "@SouthwestAir 2 hours on hold for customer service never us SW again\n",
-      "@SouthwestAir  placed on hold for total of two hours today after flight was Cancelled Flightled. Online option not available. What to do?\n",
-      "@southwestair I've been on hold for 2 hours to reschedule my Cancelled Flightled flight for the morning. What gives? I need help NOW\n",
-      "@USAirways Customer service is dead. Last wk, flts delayed/Cancelled Flighted. Bags lost 4 days. Last nt, flt delayed/Cancelled Flighted.  No meal voucher?\n"
+      "@AmericanAir - keeping AA up in the Air! My crew chief cousin Alex Espinosa in DFW! http://t.co/0HXLNvZknP\n",
+      "@JetBlue  Called JB 3 times!Everytime, Auto Vmsg:\"your wait time should not be longer than 9 mins\" waited longer than 18 mins and no answer!\n",
+      "@SouthwestAir can you outline the policies for both scenarios?\n",
+      "@united is not a company that values it's customer &amp; after reading tweets to them I'm not the only one who feels that way #lostmybusiness\n",
+      "@JetBlue how about free wifi on flt 1254 out of PBI to make up for 2.5 hr delay? Treat us right.\n"
      ]
     }
    ],
@@ -265,26 +294,26 @@
     "# Let's see what are the top predictions based on the probabilities in y_pred_test\n",
     "print(\"5 most positive tweets (class 2):\")\n",
     "for i in range(5):\n",
-    "    print(text_X_test.iloc[y_proba_test_tfidf[:, 2].argsort()[-1 - i]])\n",
+    "    print(text_X_test.iloc[y_pred_test_tfidf[y_pred_test_tfidf==2].argsort()[-1 - i]])\n",
     "\n",
     "print(\"-\" * 100)\n",
     "\n",
     "print(\"5 most negative tweets (class 0):\")\n",
     "for i in range(5):\n",
-    "    print(text_X_test.iloc[y_proba_test_tfidf[:, 0].argsort()[-1 - i]])"
+    "    print(text_X_test.iloc[y_pred_test_tfidf[y_pred_test_tfidf==0].argsort()[-1 - i]])"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 56,
+   "execution_count": 11,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Compilation time: 11.5009 seconds\n",
-      "FHE inference time: 48.6880 seconds\n"
+      "Compilation time: 5.4779 seconds\n",
+      "FHE inference time: 1.1039 seconds\n"
      ]
     }
    ],
@@ -303,22 +332,22 @@
     "\n",
     "# Now let's predict with FHE over a single tweet and print the time it takes\n",
     "start = time.perf_counter()\n",
-    "decrypted_proba = best_model.predict_proba(X_tested_tweet, execute_in_fhe=True)\n",
+    "decrypted_proba = best_model.predict_proba(X_tested_tweet, fhe=\"execute\")\n",
     "end = time.perf_counter()\n",
     "print(f\"FHE inference time: {end - start:.4f} seconds\")"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 57,
+   "execution_count": 12,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Probabilities from the FHE inference: [[0.50224707 0.25647676 0.24127617]]\n",
-      "Probabilities from the clear model: [[0.50224707 0.25647676 0.24127617]]\n"
+      "Probabilities from the FHE inference: [[0.30244059 0.17506451 0.5224949 ]]\n",
+      "Probabilities from the clear model: [[0.30244059 0.17506451 0.5224949 ]]\n"
      ]
     }
    ],
@@ -354,7 +383,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 13,
    "metadata": {},
    "outputs": [
     {
@@ -385,14 +414,19 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
+   "execution_count": 14,
    "metadata": {},
    "outputs": [
     {
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 30/30 [00:33<00:00,  1.10s/it]\n"
+      "huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...\n",
+      "To disable this warning, you can either:\n",
+      "\t- Avoid using `tokenizers` before the fork if possible\n",
+      "\t- Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)\n",
+      "  0%|          | 0/30 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked.\n",
+      "100%|██████████| 30/30 [00:20<00:00,  1.46it/s]\n"
      ]
     }
    ],
@@ -421,7 +455,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 12,
+   "execution_count": 15,
    "metadata": {},
    "outputs": [
     {
@@ -429,9 +463,9 @@
      "output_type": "stream",
      "text": [
       "Predictions for the first 3 tweets:\n",
-      " [[-2.3807464  -0.61802083  2.9900746 ]\n",
-      " [ 2.0166504   0.4938078  -2.8006463 ]\n",
-      " [ 2.3892698   0.1344364  -2.6873822 ]]\n"
+      " [[-2.3807454  -0.61802197  2.9900734 ]\n",
+      " [ 2.0166504   0.49380752 -2.8006463 ]\n",
+      " [ 2.3892734   0.13443531 -2.6873832 ]]\n"
      ]
     }
    ],
@@ -442,7 +476,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
+   "execution_count": 16,
    "metadata": {},
    "outputs": [
     {
@@ -488,15 +522,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 14,
+   "execution_count": 17,
    "metadata": {},
    "outputs": [
     {
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 13176/13176 [07:20<00:00, 29.91it/s]\n",
-      "100%|██████████| 1464/1464 [00:47<00:00, 30.75it/s]\n"
+      "100%|██████████| 13176/13176 [08:10<00:00, 26.88it/s]\n",
+      "100%|██████████| 1464/1464 [00:54<00:00, 26.90it/s]\n"
      ]
     }
    ],
@@ -542,28 +576,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 15,
+   "execution_count": 18,
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/html": [
-       "<style>#sk-container-id-2 {color: black;background-color: white;}#sk-container-id-2 pre{padding: 0;}#sk-container-id-2 div.sk-toggleable {background-color: white;}#sk-container-id-2 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-2 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-2 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-2 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-2 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-2 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-2 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-2 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-2 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-2 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-2 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-2 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-2 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-2 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-2 div.sk-item {position: relative;z-index: 1;}#sk-container-id-2 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-2 div.sk-item::before, #sk-container-id-2 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-2 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-2 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-2 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-2 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-2 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-2 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-2 div.sk-label-container {text-align: center;}#sk-container-id-2 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-2 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-2\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(), n_jobs=1,\n",
+       "<style>#sk-container-id-2 {color: black;background-color: white;}#sk-container-id-2 pre{padding: 0;}#sk-container-id-2 div.sk-toggleable {background-color: white;}#sk-container-id-2 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-2 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-2 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-2 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-2 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-2 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-2 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-2 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-2 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-2 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-2 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-2 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-2 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-2 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-2 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-2 div.sk-item {position: relative;z-index: 1;}#sk-container-id-2 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-2 div.sk-item::before, #sk-container-id-2 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-2 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-2 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-2 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-2 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-2 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-2 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-2 div.sk-label-container {text-align: center;}#sk-container-id-2 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-2 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-2\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(n_jobs=1), n_jobs=1,\n",
        "             param_grid={&#x27;max_depth&#x27;: [1], &#x27;n_bits&#x27;: [2, 3],\n",
-       "                         &#x27;n_estimators&#x27;: [10, 30, 50], &#x27;n_jobs&#x27;: [-1]},\n",
-       "             scoring=&#x27;accuracy&#x27;)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" ><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">GridSearchCV</label><div class=\"sk-toggleable__content\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(), n_jobs=1,\n",
+       "                         &#x27;n_estimators&#x27;: [10, 30, 50]},\n",
+       "             scoring=&#x27;accuracy&#x27;)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" ><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">GridSearchCV</label><div class=\"sk-toggleable__content\"><pre>GridSearchCV(cv=3, estimator=XGBClassifier(n_jobs=1), n_jobs=1,\n",
        "             param_grid={&#x27;max_depth&#x27;: [1], &#x27;n_bits&#x27;: [2, 3],\n",
-       "                         &#x27;n_estimators&#x27;: [10, 30, 50], &#x27;n_jobs&#x27;: [-1]},\n",
-       "             scoring=&#x27;accuracy&#x27;)</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-5\" type=\"checkbox\" ><label for=\"sk-estimator-id-5\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">estimator: XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier()</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-6\" type=\"checkbox\" ><label for=\"sk-estimator-id-6\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier()</pre></div></div></div></div></div></div></div></div></div></div>"
+       "                         &#x27;n_estimators&#x27;: [10, 30, 50]},\n",
+       "             scoring=&#x27;accuracy&#x27;)</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-5\" type=\"checkbox\" ><label for=\"sk-estimator-id-5\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">estimator: XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier(n_jobs=1)</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-6\" type=\"checkbox\" ><label for=\"sk-estimator-id-6\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier(n_jobs=1)</pre></div></div></div></div></div></div></div></div></div></div>"
       ],
       "text/plain": [
-       "GridSearchCV(cv=3, estimator=XGBClassifier(), n_jobs=1,\n",
+       "GridSearchCV(cv=3, estimator=XGBClassifier(n_jobs=1), n_jobs=1,\n",
        "             param_grid={'max_depth': [1], 'n_bits': [2, 3],\n",
-       "                         'n_estimators': [10, 30, 50], 'n_jobs': [-1]},\n",
+       "                         'n_estimators': [10, 30, 50]},\n",
        "             scoring='accuracy')"
       ]
      },
-     "execution_count": 15,
+     "execution_count": 18,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -576,15 +610,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
+   "execution_count": 19,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Best score: 0.8378111718275654\n",
-      "Best parameters: {'max_depth': 1, 'n_bits': 3, 'n_estimators': 50, 'n_jobs': -1}\n"
+      "Best score: 0.8381147540983607\n",
+      "Best parameters: {'max_depth': 1, 'n_bits': 3, 'n_estimators': 50}\n"
      ]
     }
    ],
@@ -601,17 +635,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": 20,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Accuracy: 0.8504\n",
-      "Average precision score for positive class: 0.8917\n",
-      "Average precision score for negative class: 0.9597\n",
-      "Average precision score for neutral class: 0.7341\n"
+      "Accuracy: 0.8463\n",
+      "Average precision score for positive class: 0.8959\n",
+      "Average precision score for negative class: 0.9647\n",
+      "Average precision score for neutral class: 0.7449\n"
      ]
     }
    ],
@@ -648,7 +682,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 21,
    "metadata": {},
    "outputs": [
     {
@@ -656,18 +690,18 @@
      "output_type": "stream",
      "text": [
       "5 most positive tweets (class 2):\n",
+      "@united I think this is the best first class I have ever gotten!!  Denver to LAX and it's wonderful!!!\n",
+      "@AmericanAir Flight 236 was great. Fantastic cabin crew. A+ landing. #thankyou #JFK http://t.co/dRW08djHAI\n",
+      "@SouthwestAir Jason (108639) at Gate #3 in SAN made my afternoon!!! #southwestairlines #stellarservice #thanks!\n",
       "@SouthwestAir love them! Always get the best deals!\n",
-      "@AmericanAir THANK YOU FOR ALL THE HELP!  :P You guys are the best.  #americanairlines #americanair\n",
-      "@SouthwestAir - Great flight from Phoenix to Dallas tonight!Great service and ON TIME! Makes @timieyancey very happy! http://t.co/TkVCMhbPim\n",
-      "@AmericanAir AA2416 on time and awesome flight. Great job American!\n",
-      "@SouthwestAir AMAZING c/s today by SW thank you SO very much. This is the reason we fly you #southwest\n",
+      "@AmericanAir simply amazing. Smiles for miles.Thank u for my upgrade tomorrow for ORD.We are spending a lot of time together next few weeks!\n",
       "----------------------------------------------------------------------------------------------------\n",
       "5 most negative tweets (class 0):\n",
-      "@AmericanAir This entire process took sooooo long that no decent seats are left.  #customerservice\n",
+      "@united first you lost all my bags, now you Cancelled Flight my flight home. 30 min wait to talk to somebody #poorservice #notgoodenough\n",
       "@USAirways Not only did u lose the flight plan! Now ur flight crew is FAA timed out! Thx for havin us sit on the tarmac for an hr! #Pathetic\n",
-      "@United site errored out at last step of changing award. Now can't even pull up reservation. 60 minute wait time.  Thanks @United!\n",
-      "@united OKC ticket agent Roger McLarren(sp?) LESS than helpful with our Intl group travel problems Can't find a supervisor for help.\n",
-      "@AmericanAir the dinner and called me \"hon\". Not the service I would expect from 1st class.  #disappointed\n"
+      "@AmericanAir Phone just disconnects if you stay on the line. Need to checkout of hotel in 2 hrs &amp; have no place to go. Can't keep calling.\n",
+      "@VirginAmerica I have lots of flights to book and your site it not working!!!! I've been on the phone waiting for over 10 minutes..........\n",
+      "@united 3 hour delay plus a jetway that won't move. This biz traveler is never flying u again!\n"
      ]
     }
    ],
@@ -689,7 +723,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 22,
    "metadata": {},
    "outputs": [
     {
@@ -697,16 +731,16 @@
      "output_type": "stream",
      "text": [
       "5 most positive (predicted) tweets that are actually negative (ground truth class 0):\n",
-      "@USAirways as far as being delayed goes… Looks like tailwinds are going to make up for it. Good news!\n",
       "@united thanks for the link, now finally arrived in Brussels, 9 h after schedule...\n",
+      "@USAirways as far as being delayed goes… Looks like tailwinds are going to make up for it. Good news!\n",
+      "@united thanks for having changed me. Managed to arrive with only 8 hours of delay and exhausted\n",
       "@USAirways your saving grace was our flight attendant Dallas who was amazing. wish he would transfer to Delta where I would see him again\n",
       "@AmericanAir that luggage you forgot...#mia.....he just won an oscar😄💝💝💝\n",
-      "@united thanks for having changed me. Managed to arrive with only 8 hours of delay and exhausted\n",
       "----------------------------------------------------------------------------------------------------\n",
       "5 most negative (predicted) tweets that are actually positive (ground truth class 2):\n",
       "@united thanks for updating me about the 1+ hour delay the exact second I got to ATL. 🙅🙅🙅\n",
-      "@JetBlue you don't remember our date Monday night back to NYC? #heartbroken\n",
       "@SouthwestAir save mile to visit family in 2015 and this will impact how many times I can see my mother.  I planned and you change the rules\n",
+      "@JetBlue you don't remember our date Monday night back to NYC? #heartbroken\n",
       "@SouthwestAir hot stewardess flipped me off\n",
       "@SouthwestAir - We left iPad in a seat pocket.  Filed lost item report. Received it exactly 1 week Late Flightr.  Is that a record?  #unbelievable\n"
      ]
@@ -750,28 +784,35 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
+   "execution_count": 23,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Compilation time: 12.6855 seconds\n"
+      "Compilation time: 5.9232 seconds\n"
      ]
     },
     {
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 1/1 [00:00<00:00, 36.43it/s]\n"
+      "100%|██████████| 1/1 [00:00<00:00, 17.83it/s]"
      ]
     },
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "FHE inference time: 53.0192 seconds\n"
+      "FHE inference time: 0.8374 seconds\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n"
      ]
     }
    ],
@@ -791,7 +832,7 @@
     "\n",
     "# Now let's predict with FHE over a single tweet and print the time it takes\n",
     "start = time.perf_counter()\n",
-    "decrypted_proba = best_model.predict_proba(X_tested_tweet, execute_in_fhe=True)\n",
+    "decrypted_proba = best_model.predict_proba(X_tested_tweet, fhe=\"execute\")\n",
     "end = time.perf_counter()\n",
     "fhe_exec_time = end - start\n",
     "print(f\"FHE inference time: {fhe_exec_time:.4f} seconds\")"
@@ -799,15 +840,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 40,
+   "execution_count": 24,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Probabilities from the FHE inference: [[0.08434131 0.05571389 0.8599448 ]]\n",
-      "Probabilities from the clear model: [[0.08434131 0.05571389 0.8599448 ]]\n"
+      "Probabilities from the FHE inference: [[0.05162184 0.04558276 0.90279541]]\n",
+      "Probabilities from the clear model: [[0.05162184 0.04558276 0.90279541]]\n"
      ]
     }
    ],
@@ -818,34 +859,38 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 40,
    "metadata": {},
    "outputs": [],
    "source": [
+    "DEPLOYMENT_DIR = Path(\"deployment\")\n",
+    "DEPLOYMENT_DIR.mkdir(exist_ok=True)\n",
+    "\n",
     "# Let's export the final model such that we can reuse it in a client/server environment\n",
     "\n",
-    "# Export the model to ONNX\n",
-    "onnx.save(best_model._onnx_model_, \"server_model.onnx\")  # pylint: disable=protected-access\n",
+    "# Serialize the model (for development only)\n",
+    "with (DEPLOYMENT_DIR / \"serialized_model\").open(\"w\") as file:\n",
+    "    best_model.dump(file)\n",
     "\n",
-    "# Export some data to be used for compilation\n",
+    "# Export some data to be used for compilation \n",
     "X_train_numpy = X_train_transformer[:100]\n",
     "\n",
     "# Merge the two arrays in a pandas dataframe\n",
     "X_test_numpy_df = pd.DataFrame(X_train_numpy)\n",
     "\n",
     "# to csv\n",
-    "X_test_numpy_df.to_csv(\"samples_for_compilation.csv\")\n",
+    "X_test_numpy_df.to_csv(DEPLOYMENT_DIR / \"samples_for_compilation.csv\")\n",
     "\n",
     "# Let's save the model to be pushed to a server later\n",
     "from concrete.ml.deployment import FHEModelDev\n",
     "\n",
-    "fhe_api = FHEModelDev(\"sentiment_fhe_model\", best_model)\n",
-    "fhe_api.save()"
+    "fhe_api = FHEModelDev(DEPLOYMENT_DIR / \"sentiment_fhe_model\", best_model)\n",
+    "fhe_api.save(via_mlir=True)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
+   "execution_count": null,
    "metadata": {},
    "outputs": [
     {
@@ -885,24 +930,24 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>TF-IDF + XGBoost</th>\n",
-       "      <td>0.681011</td>\n",
-       "      <td>0.561521</td>\n",
-       "      <td>0.834914</td>\n",
-       "      <td>0.382002</td>\n",
+       "      <td>0.711749</td>\n",
+       "      <td>0.640422</td>\n",
+       "      <td>0.871891</td>\n",
+       "      <td>0.43486</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Transformer Only</th>\n",
        "      <td>0.805328</td>\n",
        "      <td>0.854827</td>\n",
        "      <td>0.954804</td>\n",
-       "      <td>0.680110</td>\n",
+       "      <td>0.68011</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Transformer + XGBoost</th>\n",
-       "      <td>0.850410</td>\n",
-       "      <td>0.891691</td>\n",
-       "      <td>0.959747</td>\n",
-       "      <td>0.734144</td>\n",
+       "      <td>0.846311</td>\n",
+       "      <td>0.895930</td>\n",
+       "      <td>0.964674</td>\n",
+       "      <td>0.74489</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -911,24 +956,24 @@
       "text/plain": [
        "                       Accuracy  Average Precision (positive)  \\\n",
        "Model                                                           \n",
-       "TF-IDF + XGBoost       0.681011                      0.561521   \n",
+       "TF-IDF + XGBoost       0.711749                      0.640422   \n",
        "Transformer Only       0.805328                      0.854827   \n",
-       "Transformer + XGBoost  0.850410                      0.891691   \n",
+       "Transformer + XGBoost  0.846311                      0.895930   \n",
        "\n",
        "                       Average Precision (negative)  \\\n",
        "Model                                                 \n",
-       "TF-IDF + XGBoost                           0.834914   \n",
+       "TF-IDF + XGBoost                           0.871891   \n",
        "Transformer Only                           0.954804   \n",
-       "Transformer + XGBoost                      0.959747   \n",
+       "Transformer + XGBoost                      0.964674   \n",
        "\n",
        "                       Average Precision (neutral)  \n",
        "Model                                               \n",
-       "TF-IDF + XGBoost                          0.382002  \n",
-       "Transformer Only                          0.680110  \n",
-       "Transformer + XGBoost                     0.734144  "
+       "TF-IDF + XGBoost                           0.43486  \n",
+       "Transformer Only                           0.68011  \n",
+       "Transformer + XGBoost                      0.74489  "
       ]
      },
-     "execution_count": 26,
+     "execution_count": 33,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -991,7 +1036,15 @@
    "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": "3.10.11"
   }
  },

app.py

@@ -26,6 +26,7 @@ time.sleep(5)
 # (encrypted data is too large to display in the browser)
 ENCRYPTED_DATA_BROWSER_LIMIT = 500
 N_USER_KEY_STORED = 20
+FHE_MODEL_PATH = "deployment/sentiment_fhe_model"
 
 print("Loading the transformer model...")
 
@@ -60,7 +61,7 @@ def keygen():
 
     # Let's create a user_id
     user_id = numpy.random.randint(0, 2**32)
-    fhe_api = FHEModelClient("sentiment_fhe_model/deployment", f".fhe_keys/{user_id}")
+    fhe_api = FHEModelClient(FHE_MODEL_PATH, f".fhe_keys/{user_id}")
     fhe_api.load()
 
 
@@ -79,7 +80,7 @@ def encode_quantize_encrypt(text, user_id):
     if not user_id:
         raise gr.Error("You need to generate FHE keys first.")
 
-    fhe_api = FHEModelClient("sentiment_fhe_model/deployment", f".fhe_keys/{user_id}")
+    fhe_api = FHEModelClient(FHE_MODEL_PATH, f".fhe_keys/{user_id}")
     fhe_api.load()
     encodings = transformer_vectorizer.transform([text])
     quantized_encodings = fhe_api.model.quantize_input(encodings).astype(numpy.uint8)
@@ -143,7 +144,7 @@ def decrypt_prediction(user_id):
     # Read encrypted_prediction from the file
     encrypted_prediction = numpy.load(encoded_data_path).tobytes()
 
-    fhe_api = FHEModelClient("sentiment_fhe_model/deployment", f".fhe_keys/{user_id}")
+    fhe_api = FHEModelClient(FHE_MODEL_PATH, f".fhe_keys/{user_id}")
     fhe_api.load()
 
     # We need to retrieve the private key that matches the client specs (see issue #18)

compile.py

@@ -1,7 +1,8 @@
 import onnx
 import pandas as pd
 from concrete.ml.deployment import FHEModelDev, FHEModelClient
-from concrete.ml.onnx.convert import get_equivalent_numpy_forward
+from concrete.ml.common.serialization.loaders import load
+from concrete.ml.onnx.convert import get_equivalent_numpy_forward_from_onnx_tree
 import json
 import os
 import shutil
@@ -10,48 +11,29 @@ from pathlib import Path
 
 script_dir = Path(__file__).parent
 
+DEPLOYMENT_DIR = script_dir / "deployment"
+
 print("Compiling the model...")
 
-# Load the onnx model
-model_onnx = onnx.load(Path.joinpath(script_dir, "sentiment_fhe_model/server_model.onnx"))
+with (DEPLOYMENT_DIR / "serialized_model").open("r") as file:
+    model = load(file)
 
 # Load the data from the csv file to be used for compilation
-data = pd.read_csv(
-    Path.joinpath(script_dir, "sentiment_fhe_model/samples_for_compilation.csv"), index_col=0
-).values
-
-# Convert the onnx model to a numpy model
-_tensor_tree_predict = get_equivalent_numpy_forward(model_onnx)
-
-model = FHEModelClient(
-    Path.joinpath(script_dir, "sentiment_fhe_model/deployment"), ".fhe_keys"
-).model
-
-# Assign the numpy model and compile the model
-model._tensor_tree_predict = _tensor_tree_predict
+data = pd.read_csv(DEPLOYMENT_DIR / "samples_for_compilation.csv", index_col=0).values
 
 # Compile the model
 model.compile(data)
 
-# Load the serialized_processing.json file
-with open(
-    Path.joinpath(script_dir, "sentiment_fhe_model/deployment/serialized_processing.json"), "r"
-) as f:
-    serialized_processing = json.load(f)
+dev_model_path = DEPLOYMENT_DIR / "sentiment_fhe_model"
 
 # Delete the deployment folder if it exist
-if Path.joinpath(script_dir, "sentiment_fhe_model/deployment").exists():
-    shutil.rmtree(Path.joinpath(script_dir, "sentiment_fhe_model/deployment"))
+if dev_model_path.is_dir():
+    shutil.rmtree(dev_model_path)
 
 fhe_api = FHEModelDev(
-    model=model, path_dir=Path.joinpath(script_dir, "sentiment_fhe_model/deployment")
+    model=model, path_dir=dev_model_path
 )
 fhe_api.save(via_mlir=True)
 
-# Write the serialized_processing.json file to the deployment folder
-with open(
-    Path.joinpath(script_dir, "sentiment_fhe_model/deployment/serialized_processing.json"), "w"
-) as f:
-    json.dump(serialized_processing, f)
 
 print("Done!")

deployment/sentiment_fhe_model/client.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:972f0c7d83f12e3a43e8f923fc422cdb443b9f64bb6f74c1abf912836ba27e60
+size 3887326

deployment/sentiment_fhe_model/server.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:216d2a78d7ec47ec2a478d5f32ed34cee8a9c45700325e5d8de4e087b7ed8dfc
+size 3004

deployment/sentiment_fhe_model/versions.json

@@ -0,0 +1 @@
+{"concrete-python": "2.5", "concrete-ml": "1.4.0", "python": "3.10.11"}

requirements.txt

@@ -1,5 +1,5 @@
-concrete-ml==1.1.0
+concrete-ml==1.4.0
 gradio==3.40.1
 pandas==1.4.3
-transformers==4.32.0 
+transformers==4.36.0 
 jupyter==1.0.0

server.py

@@ -9,7 +9,7 @@ from pathlib import Path
 current_dir = Path(__file__).parent
 
 # Load the model
-fhe_model = FHEModelServer(Path.joinpath(current_dir, "sentiment_fhe_model/deployment"))
+fhe_model = FHEModelServer("deployment/sentiment_fhe_model")
 
 class PredictRequest(BaseModel):
     evaluation_key: str