diff --git a/notebooks/Predict_XAI.ipynb b/notebooks/Predict_XAI.ipynb
index 517c7a3019ca40fd5cd0c7ec2d2f35338a460c37..694cd1bc35a809837a3780254ab77bd325dff079 100644
--- a/notebooks/Predict_XAI.ipynb
+++ b/notebooks/Predict_XAI.ipynb
@@ -755,7 +755,7 @@
       "source": [
         "## 4. Load model and predict\n",
         "\n",
-        "### 4.1 BERT"
+        "### 4.1 Load BERT model"
       ]
     },
     {
@@ -770,6 +770,17 @@
         "model_path = path + \"models/model_\" + model_name + \"_s10000.pt\""
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "encoder_filename = \"models/label_encoder.pkl\"\n",
+        "with open(path + encoder_filename, 'rb') as file:\n",
+        "      encoder = pickle.load(file)"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": 16,
@@ -784,10 +795,25 @@
         }
       ],
       "source": [
-        "print('Loading Bert Tokenizer...')\n",
         "tokenizer = BertTokenizer.from_pretrained(model_name)"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "model = BertForSequenceClassification.from_pretrained(model_path).to(gpu_name) #.to(\"cuda\")"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "### 4.2 Prepare datasets"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": 17,
@@ -802,23 +828,35 @@
         }
       ],
       "source": [
-        "data_loader = generate_dataloader(tokenizer, data_LGE)"
+        "# LGE\n",
+        "data_loader_LGE = generate_dataloader(tokenizer, df_LGE.content.values)"
       ]
     },
     {
       "cell_type": "code",
-      "execution_count": 18,
+      "execution_count": null,
       "metadata": {},
       "outputs": [],
       "source": [
-        "model = BertForSequenceClassification.from_pretrained(model_path).to(gpu_name) #.to(\"cuda\")"
+        "# LGE parallel\n",
+        "data_loader_LGE_par = generate_dataloader(tokenizer, df_LGE_par.content.values)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "# EDdA\n",
+        "data_loader_EDdA = generate_dataloader(tokenizer, df_EDdA.content.values)"
       ]
     },
     {
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "### 4.2 Predict"
+        "### 4.3 Predict"
       ]
     },
     {
@@ -833,36 +871,28 @@
       },
       "outputs": [],
       "source": [
-        "pred = predict(model, data_loader, device)"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": 22,
-      "metadata": {},
-      "outputs": [],
-      "source": [
-        "encoder_filename = \"models/label_encoder.pkl\"\n",
-        "with open(path + encoder_filename, 'rb') as file:\n",
-        "      encoder = pickle.load(file)"
+        "pred_LGE = predict(model, data_loader_LGE, device)\n",
+        "df_LGE['class_pred'] = list(encoder.inverse_transform(pred_LGE))"
       ]
     },
     {
       "cell_type": "code",
-      "execution_count": 23,
+      "execution_count": null,
       "metadata": {},
       "outputs": [],
       "source": [
-        "p2 = list(encoder.inverse_transform(pred))"
+        "pred_LGE_par = predict(model, data_loader_LGE_par, device)\n",
+        "df_LGE_par['class_pred'] = list(encoder.inverse_transform(pred_LGE_par))"
       ]
     },
     {
       "cell_type": "code",
-      "execution_count": 24,
+      "execution_count": 22,
       "metadata": {},
       "outputs": [],
       "source": [
-        "df_LGE['domain'] = p2"
+        "pred_EDdA = predict(model, data_loader_EDdA, device)\n",
+        "df_EDdA['class_pred'] = list(encoder.inverse_transform(pred_EDdA))"
       ]
     },
     {
@@ -1569,7 +1599,7 @@
         }
       ],
       "source": [
-        "df_LGE.head(50)"
+        "df_LGE.head()"
       ]
     },
     {
@@ -1766,7 +1796,7 @@
         }
       ],
       "source": [
-        "content = \"Instrument de musique\" #df_LGE.content[2][:512]\n",
+        "content = \"Instrument de musique\" #df_LGE.content[2]\n",
         "word_attributions = cls_explainer(content if len(content) < 512 else content[:512])\n",
         "word_attributions"
       ]