diff --git a/thunder/benchmarks/benchmark_litgpt.py b/thunder/benchmarks/benchmark_litgpt.py
index f68dd1f0fa..5bf61b4d6c 100644
--- a/thunder/benchmarks/benchmark_litgpt.py
+++ b/thunder/benchmarks/benchmark_litgpt.py
@@ -286,7 +286,7 @@ def train(self):
             iter_times = []
             if i == self.warmup_iter:  # warmup
                 t0 = iter_t0
-                iter_times = [] #reset the iter time list
+                iter_times = []  # reset the iter time list
 
             for step_idx in range(self.gradient_accumulation_steps):
                 input_ids, targets = next(self.train_data_iter)
@@ -328,9 +328,7 @@ def train(self):
             t1 = time.perf_counter()
             if global_rank in [0, None]:
                 iter_time = (t1 - iter_t0) * 1000
-                print(
-                    f"iter {i}: loss {loss_item:.4f}, iter time: {iter_time:.2f}ms, t: {input_ids.size(1)}"
-                )
+                print(f"iter {i}: loss {loss_item:.4f}, iter time: {iter_time:.2f}ms, t: {input_ids.size(1)}")
                 iter_times.append(iter_time)
 
             # if global_rank in [0, None] and i >=warmup_iter:
@@ -348,7 +346,7 @@ def train(self):
 
         if global_rank in [0, None]:
             self.perf_metrics["average_iter_time"] = ((t1 - t0) * 1000) / (self.max_iters - self.warmup_iter)
-            self.perf_metrics["median_iter_time"] = np.median(iter_times) #To avoid outliers
+            self.perf_metrics["median_iter_time"] = np.median(iter_times)  # To avoid outliers
 
     def add_perf_metrics(self):
         # tokens_per_sec = total number of benchmarked iterations x global BS x block_size / total elapsed time (s)