diff --git a/tensorflow_recommenders/layers/factorized_top_k.py b/tensorflow_recommenders/layers/factorized_top_k.py
index b6a7f451..0316c3cf 100644
--- a/tensorflow_recommenders/layers/factorized_top_k.py
+++ b/tensorflow_recommenders/layers/factorized_top_k.py
@@ -466,7 +466,7 @@ def top_k(state: Tuple[tf.Tensor, tf.Tensor],
     def enumerate_rows(batch: tf.Tensor) -> Tuple[tf.Tensor, tf.Tensor]:
       """Enumerates rows in each batch using a total element counter."""
 
-      starting_counter = self._counter.read_value()
+      starting_counter = self._counter.value
       end_counter = self._counter.assign_add(tf.shape(batch)[0])
 
       return tf.range(starting_counter, end_counter), batch
diff --git a/tensorflow_recommenders/metrics/factorized_top_k.py b/tensorflow_recommenders/metrics/factorized_top_k.py
index b9bfd7bd..8d025a9e 100644
--- a/tensorflow_recommenders/metrics/factorized_top_k.py
+++ b/tensorflow_recommenders/metrics/factorized_top_k.py
@@ -177,7 +177,8 @@ def update_state(
             tf.reduce_sum(ids_match[:, :k], axis=1, keepdims=True),
             0.0, 1.0
         )
-        update_ops.append(metric.update_state(match_found, sample_weight))
+        metric.update_state(match_found, sample_weight)
+        update_ops.append(metric.result())
     else:
       # Score-based evaluation.
       y_pred = tf.concat([positive_scores, top_k_predictions], axis=1)
@@ -189,7 +190,8 @@ def update_state(
             predictions=y_pred,
             k=k
         )
-        update_ops.append(metric.update_state(top_k_accuracy, sample_weight))
+        metric.update_state(top_k_accuracy, sample_weight)
+        update_ops.append(metric.result())
 
     return tf.group(update_ops)
   
\ No newline at end of file