diff --git a/RNN - Multilabel.ipynb b/RNN - Multilabel.ipynb
index 947c180..e671e5e 100644
--- a/RNN - Multilabel.ipynb	
+++ b/RNN - Multilabel.ipynb	
@@ -186,7 +186,7 @@
     "    output, state = tf.nn.dynamic_rnn(multi_layer_cell, x, dtype = tf.float32)\n",
     "    output_flattened = tf.reshape(output, [-1, n_hidden])\n",
     "    output_logits = tf.add(tf.matmul(output_flattened,weight),bias)\n",
-    "    output_all = tf.nn.sigmoid(output_logits)\n",
+    "    output_all = (output_logits)\n",
     "    output_reshaped = tf.reshape(output_all,[-1,n_steps,n_classes])\n",
     "    output_last = tf.gather(tf.transpose(output_reshaped,[1,0,2]), n_steps - 1)  \n",
     "    #output = tf.transpose(output, [1, 0, 2])\n",
@@ -226,8 +226,8 @@
    "outputs": [],
    "source": [
     "#all_steps_cost=tf.reduce_mean(-tf.reduce_mean((y_steps * tf.log(y_all))+(1 - y_steps) * tf.log(1 - y_all),reduction_indices=1))\n",
-    "all_steps_cost = -tf.reduce_mean((y_steps * tf.log(y_all))  + (1 - y_steps) * tf.log(1 - y_all))\n",
-    "last_step_cost = -tf.reduce_mean((y * tf.log(y_last)) + ((1 - y) * tf.log(1 - y_last)))\n",
+    "all_steps_cost = -tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(y_all, y))\n",
+    "last_step_cost = -tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(y_last, y))\n",
     "loss_function = (alpha * all_steps_cost) + ((1 - alpha) * last_step_cost)\n",
     "\n",
     "optimizer = tf.train.AdamOptimizer(learning_rate = learning_rate).minimize(loss_function)"