Update `pre-commit` hooks (#12475)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

vllm/v1/stats/common.py

@@ -311,8 +311,8 @@ class RequestStats:
             return []
         latency_s_lst = []
         for i in range(1, len(self.output_token_ts_s_lst)):
-            assert (self.output_token_ts_s_lst[i] >=
-                    self.output_token_ts_s_lst[i - 1])
+            assert (self.output_token_ts_s_lst[i]
+                    >= self.output_token_ts_s_lst[i - 1])
             latency_s = (self.output_token_ts_s_lst[i] -
                          self.output_token_ts_s_lst[i - 1])
             latency_s_lst.append(latency_s)

vllm/v1/worker/gpu_model_runner.py

@@ -205,7 +205,7 @@ class GPUModelRunner:
 
     def _update_states(self, scheduler_output: "SchedulerOutput") -> None:
         # Remove stopped requests from the cached states.
-        # Keep the states of the pre-empted requests.
+        # Keep the states of the preempted requests.
         for req_id in scheduler_output.finished_req_ids:
             self.requests.pop(req_id, None)
             self.encoder_cache.pop(req_id, None)

vllm/worker/hpu_worker.py

@@ -173,13 +173,13 @@ class HPUWorker(LocalOrDistributedWorkerBase):
                 cpu_fallback_ctx as cpu_fallback_local_metric:
                 output = LocalOrDistributedWorkerBase.execute_model(
                     self, execute_model_req)
-            if (log_graph_compilation and gc_local_metric.stats()[0][1] > 0
-                ) or log_graph_compilation_all:
+            if (log_graph_compilation and gc_local_metric.stats()[0][1]
+                    > 0) or log_graph_compilation_all:
                 msg = ("VLLM_HPU_STEP_GRAPH_COMPILATION: "
                        f"{gc_local_metric.stats()}, {input_stats}")
                 logger.warning(msg)
-            if (log_cpu_fallbacks and cpu_fallback_local_metric.stats()[0][1] >
-                    0) or log_cpu_fallbacks_all:
+            if (log_cpu_fallbacks and cpu_fallback_local_metric.stats()[0][1]
+                    > 0) or log_cpu_fallbacks_all:
                 msg = ("VLLM_HPU_STEP_CPU_FALLBACK: "
                        f"{cpu_fallback_local_metric.stats()}, {input_stats}")
                 logger.warning(msg)

vllm/worker/tpu_model_runner.py

@@ -316,8 +316,8 @@ class TPUModelRunner(ModelRunnerBase[ModelInputForTPU]):
                     logger.info("batch_size: %d, seq_len: %d", batch_size,
                                 seq_len)
                     num_tokens = batch_size * seq_len
-                    if (num_tokens >=
-                            self.scheduler_config.max_num_batched_tokens):
+                    if (num_tokens
+                            >= self.scheduler_config.max_num_batched_tokens):
                         break
                     seq_len = seq_len * 2
             end = time.time()