Add typo checker in pre-commit (#6179)

Co-authored-by: Brayden Zhong <b8zhong@uwaterloo.ca>

python/sglang/srt/managers/data_parallel_controller.py

@@ -210,7 +210,7 @@ class DataParallelController:
                     )
                     # compute zmq ports for this dp rank
                     rank_port_args = PortArgs.init_new(server_args, dp_rank)
-                    # Data parallelism resues the tensor parallelism group,
+                    # Data parallelism reuses the tensor parallelism group,
                     # so all dp ranks should use the same nccl port.
                     rank_port_args.nccl_port = port_args.nccl_port
 

python/sglang/srt/managers/io_struct.py

@@ -12,7 +12,7 @@
 # limitations under the License.
 # ==============================================================================
 """
-The definition of objects transfered between different
+The definition of objects transferred between different
 processes (TokenizerManager, DetokenizerManager, Controller).
 """
 

python/sglang/srt/managers/mm_utils.py

@@ -51,7 +51,7 @@ class MultiModalityDataPaddingPatternTokenPairs(MultiModalityDataPaddingPattern)
         self, input_ids: List[int], mm_inputs: MultimodalInputs
     ) -> List[int]:
         """
-        This function will replace the data-tokens inbetween with pad_values accordingly
+        This function will replace the data-tokens in between with pad_values accordingly
         """
         pad_values = [item.pad_value for item in mm_inputs.mm_items]
         data_token_pairs = self.data_token_id_pairs

python/sglang/srt/managers/schedule_batch.py

@@ -879,7 +879,7 @@ class ScheduleBatch(ScheduleBatchDisaggregationDecodeMixin):
             error_msg = (
                 f"{phase_str} out of memory. Try to lower your batch size.\n"
                 f"Try to allocate {num_tokens} tokens.\n"
-                f"Avaliable tokens: {self.token_to_kv_pool_allocator.available_size() + self.tree_cache.evictable_size()}\n"
+                f"Available tokens: {self.token_to_kv_pool_allocator.available_size() + self.tree_cache.evictable_size()}\n"
             )
             logger.error(error_msg)
             if self.tree_cache is not None:
@@ -920,7 +920,7 @@ class ScheduleBatch(ScheduleBatchDisaggregationDecodeMixin):
             error_msg = (
                 f"Prefill out of memory. Try to lower your batch size.\n"
                 f"Try to allocate {extend_num_tokens} tokens.\n"
-                f"Avaliable tokens: {self.token_to_kv_pool_allocator.available_size() + self.tree_cache.evictable_size()}\n"
+                f"Available tokens: {self.token_to_kv_pool_allocator.available_size() + self.tree_cache.evictable_size()}\n"
                 f"{self.token_to_kv_pool_allocator.available_size()=}\n"
                 f"{self.tree_cache.evictable_size()=}\n"
             )
@@ -955,7 +955,7 @@ class ScheduleBatch(ScheduleBatchDisaggregationDecodeMixin):
             error_msg = (
                 f"Decode out of memory. Try to lower your batch size.\n"
                 f"Try to allocate {len(seq_lens)} tokens.\n"
-                f"Avaliable tokens: {self.token_to_kv_pool_allocator.available_size() + self.tree_cache.evictable_size()}\n"
+                f"Available tokens: {self.token_to_kv_pool_allocator.available_size() + self.tree_cache.evictable_size()}\n"
                 f"{self.token_to_kv_pool_allocator.available_size()=}\n"
                 f"{self.tree_cache.evictable_size()=}\n"
             )

python/sglang/srt/managers/scheduler.py

@@ -1325,7 +1325,7 @@ class Scheduler(
             return None
 
         running_bs = len(self.running_batch.reqs)
-        # Igore the check if self.chunked_req is not None.
+        # Ignore the check if self.chunked_req is not None.
         # In the non-PP case, when self.chunked_req is not None, num_allocatable_reqs should always be greater than 0,
         # as the space for the chunked request has just been released.
         # In PP case, a chunked req can start in one microbatch and end in another microbatch, so the max_running_requests per microbatch should not be strict.

python/sglang/srt/managers/tokenizer_manager.py

@@ -1273,7 +1273,7 @@ class TokenizerManager:
             self.model_update_result.set_result(recv_obj)
         else:  # self.server_args.dp_size > 1
             self.model_update_tmp.append(recv_obj)
-            # set future if the all results are recevied
+            # set future if the all results are received
             if len(self.model_update_tmp) == self.server_args.dp_size:
                 self.model_update_result.set_result(self.model_update_tmp)