diff --git a/CoRML Demonstration.ipynb b/CoRML Demonstration.ipynb
index 5c49b47..615a9fb 100644
--- a/CoRML Demonstration.ipynb	
+++ b/CoRML Demonstration.ipynb	
@@ -310,8 +310,8 @@
         "\n",
         "        self.lambda_ = config[\"lambda\"]                     # Weights for H and G in preference scores\n",
         "        self.rho = config[\"dual_step_length\"]               # Dual step length of ADMM\n",
-        "        self.theta = config[\"l2_regularization\"]            # L2-regularization for learning weight matrix G\n",
-        "        self.norm_di = 2 * config[\"item_degree_norm\"]       # Item degree norm for learning weight matrix G\n",
+        "        self.theta = config[\"l2_regularization\"]            # L2-regularization for learning weight matrix H\n",
+        "        self.norm_di = 2 * config[\"item_degree_norm\"]       # Item degree norm for learning weight matrix H\n",
         "        self.eps = np.power(10, config[\"global_scaling\"])   # Global scaling in approximated ranking weights (in logarithm scale)\n",
         "\n",
         "        self.sparse_approx = config[\"sparse_approx\"]        # Sparse approximation to reduce storage size of H\n",
@@ -359,7 +359,7 @@
         "        self.H_indices = []\n",
         "        self.H_values = []\n",
         "\n",
-        "        for ilist in tqdm(item_list, desc=\"Partition\", bar_format=\"{elapsed}\"):\n",
+        "        for ilist in tqdm(item_list, desc=\"Partition\", bar_format=\"{l_bar}{bar}| {n_fmt}/{total_fmt} [{elapsed}]\"):\n",
         "            H_triu = self.update_H_part(ilist)\n",
         "            H_triu = torch.where(H_triu >= 5e-4, H_triu, 0).to_sparse_coo()\n",
         "            self.H_indices.append(ilist[H_triu.indices()])\n",
diff --git a/CoRML/model.py b/CoRML/model.py
index 19e0970..39223d1 100644
--- a/CoRML/model.py
+++ b/CoRML/model.py
@@ -132,8 +132,8 @@ def __init__(self, config, dataset):
 
         self.lambda_ = config["lambda"]                     # Weights for H and G in preference scores
         self.rho = config["dual_step_length"]               # Dual step length of ADMM
-        self.theta = config["l2_regularization"]            # L2-regularization for learning weight matrix G
-        self.norm_di = 2 * config["item_degree_norm"]       # Item degree norm for learning weight matrix G
+        self.theta = config["l2_regularization"]            # L2-regularization for learning weight matrix H
+        self.norm_di = 2 * config["item_degree_norm"]       # Item degree norm for learning weight matrix H
         self.eps = np.power(10, config["global_scaling"])   # Global scaling in approximated ranking weights (in logarithm scale)
 
         self.sparse_approx = config["sparse_approx"]        # Sparse approximation to reduce storage size of H