[hotfix] fix inference typo (#5438)

Author: hugo-syn

colossalai/legacy/inference/async_manager.py

@@ -55,14 +55,14 @@ def _step(self):
                 self.stats_tool.count_prompt_tokens(new_batch)
                 self.running_batch = new_batch
                 has_new_finished, outputs = self._prefill_batch(self.running_batch)
-                self._filter_runing_batch()
+                self._filter_running_batch()
                 self.has_wait_tokens = 0
 
         else:
             if self.has_wait_tokens < self.max_wait_tokens:
                 self.stats_tool.count_output_tokens(self.running_batch)
                 has_new_finished, outputs = self._decode_batch(self.running_batch)
-                self._filter_runing_batch()
+                self._filter_running_batch()
                 self.has_wait_tokens += 1
 
             else:
@@ -78,7 +78,7 @@ def _step(self):
                 else:
                     self.stats_tool.count_output_tokens(self.running_batch)
                     has_new_finished, outputs = self._decode_batch(self.running_batch)
-                    self._filter_runing_batch()
+                    self._filter_running_batch()
                     self.has_wait_tokens += 1
 
         if has_new_finished:

colossalai/legacy/inference/manager.py

@@ -131,14 +131,14 @@ def _step(self):
                 self.stats_tool.count_prompt_tokens(new_batch)
                 self.running_batch = new_batch
                 yield from self._prefill_batch(self.running_batch)
-                self._filter_runing_batch()
+                self._filter_running_batch()
                 self.has_wait_tokens = 0
             return
 
         if self.has_wait_tokens < self.max_wait_tokens:
             self.stats_tool.count_output_tokens(self.running_batch)
             yield from self._decode_batch(self.running_batch)
-            self._filter_runing_batch()
+            self._filter_running_batch()
             self.has_wait_tokens += 1
             return
         else:
@@ -154,7 +154,7 @@ def _step(self):
             else:
                 self.stats_tool.count_output_tokens(self.running_batch)
                 yield from self._decode_batch(self.running_batch)
-                self._filter_runing_batch()
+                self._filter_running_batch()
                 self.has_wait_tokens += 1
 
         return
@@ -243,7 +243,7 @@ def _handle_finish_req(self, batch: Batch, has_new_finished_req):
                 self._filter_batch(batch)
             yield from self._output_process(finished_reqs)
 
-    def _filter_runing_batch(self):
+    def _filter_running_batch(self):
         if self.running_batch is not None and self.running_batch.is_clear():
             self.running_batch = None