refactor: replace the argument d_inner with d_ffn;

PyPOTS_Classification.ipynb

@@ -116,7 +116,7 @@
     "    n_classes=physionet2012_dataset[\"n_classes\"],\n",
     "    n_layers=2,\n",
     "    d_model=physionet2012_dataset[\"n_features\"] * 4,\n",
-    "    d_inner=256,\n",
+    "    d_ffn=256,\n",
     "    n_heads=2,\n",
     "    dropout=0.3,\n",
     "    batch_size=32,\n",

PyPOTS_Imputation.ipynb

@@ -111,7 +111,7 @@
     "    n_features=physionet2012_dataset['n_features'],\n",
     "    n_layers=2,\n",
     "    d_model=256,\n",
-    "    d_inner=128,\n",
+    "    d_ffn=128,\n",
     "    n_heads=4,\n",
     "    d_k=64,\n",
     "    d_v=64,\n",
@@ -245,7 +245,7 @@
     "    n_features=physionet2012_dataset['n_features'],\n",
     "    n_layers=6,\n",
     "    d_model=512,\n",
-    "    d_inner=256,\n",
+    "    d_ffn=256,\n",
     "    n_heads=4,\n",
     "    d_k=128,\n",
     "    d_v=128,\n",
@@ -530,7 +530,7 @@
     "    num_workers=0,\n",
     "    # just leave it to default as None, PyPOTS will automatically assign the best device for you.\n",
     "    # Set it as 'cpu' if you don't have CUDA devices. You can also set it to 'cuda:0' or 'cuda:1' if you have multiple CUDA devices, even parallelly on ['cuda:0', 'cuda:1']\n",
-    "    device='cuda:1',\n",
+    "    device=None,\n",
     "    # set the path for saving tensorboard and trained model files \n",
     "    saving_path=\"tutorial_results/imputation/csdi\",\n",
     "    # only save the best model after training finished.\n",
@@ -1047,10 +1047,8 @@
    "source": [
     "from pypots.optim import Adam\n",
     "from pypots.imputation import MRNN\n",
-    "from pypots.utils.metrics import cal_mae\n",
     "\n",
     "# initialize the model\n",
-    "# initialize the model\n",
     "mrnn = MRNN(\n",
     "    n_steps=physionet2012_dataset['n_steps'],\n",
     "    n_features=physionet2012_dataset['n_features'],\n",