误删函数恢复

vllm/v1/worker/gpu_model_runner.py

@@ -2073,11 +2073,11 @@ class GPUModelRunner(
         )
         return common_prefix_len if use_cascade else 0
 
-    def _calc_mrope_positions(self, scheduler_output: "SchedulerOutput"):
-        mrope_pos_ptr = 0
+    def _calc_xdrope_positions(self, scheduler_output: "SchedulerOutput"):
+        xdrope_pos_ptr = 0
         for index, req_id in enumerate(self.input_batch.req_ids):
             req = self.requests[req_id]
-            assert req.mrope_positions is not None
+            assert req.xdrope_positions is not None
 
             num_computed_tokens = self.input_batch.num_computed_tokens_cpu[index]
             num_scheduled_tokens = scheduler_output.num_scheduled_tokens[req_id]
@@ -2095,33 +2095,31 @@ class GPUModelRunner(
             assert num_scheduled_tokens == prompt_part_len + completion_part_len
 
             if prompt_part_len > 0:
-                # prompt's mrope_positions are pre-computed
-                dst_start = mrope_pos_ptr
-                dst_end = mrope_pos_ptr + prompt_part_len
+                # prompt's xdrope_positions are pre-computed
+                dst_start = xdrope_pos_ptr
+                dst_end = xdrope_pos_ptr + prompt_part_len
                 src_start = num_computed_tokens
                 src_end = num_computed_tokens + prompt_part_len
 
-                self.mrope_positions.cpu[:, dst_start:dst_end] = req.mrope_positions[
+                self.xdrope_positions.cpu[:, dst_start:dst_end] = req.xdrope_positions[
                     :, src_start:src_end
                 ]
-                mrope_pos_ptr += prompt_part_len
+                xdrope_pos_ptr += prompt_part_len
 
             if completion_part_len > 0:
-                # compute completion's mrope_positions on-the-fly
-                dst_start = mrope_pos_ptr
-                dst_end = mrope_pos_ptr + completion_part_len
+                # compute completion's xdrope_positions on-the-fly
+                dst_start = xdrope_pos_ptr
+                dst_end = xdrope_pos_ptr + completion_part_len
 
-                assert req.mrope_position_delta is not None
-                MRotaryEmbedding.get_next_input_positions_tensor(
-                    out=self.mrope_positions.np,
+                XDRotaryEmbedding.get_next_input_positions_tensor(
+                    out=self.xdrope_positions.np,
                     out_offset=dst_start,
-                    mrope_position_delta=req.mrope_position_delta,
                     context_len=num_computed_tokens + prompt_part_len,
                     num_new_tokens=completion_part_len,
                 )
 
-                mrope_pos_ptr += completion_part_len
-
+                xdrope_pos_ptr += completion_part_len
+                
     def _calc_mrope_positions(self, scheduler_output: "SchedulerOutput"):
         mrope_pos_ptr = 0
         if self.use_1d_mrope: