[Bugfix][Qwen3-Next] Fix Qwen3-Next with the latest maintained vllm commit id

Signed-off-by: MengqingCao <[email protected]>

Author: MengqingCao

vllm_ascend/models/qwen3_next.py

@@ -51,6 +51,8 @@
 from vllm.transformers_utils.configs import Qwen3NextConfig
 from vllm.v1.attention.backends.gdn_attn import GDNAttentionMetadata
 
+from vllm_ascend.utils import vllm_version_is
+
 from vllm.model_executor.models.qwen3_next import (  # isort: skip
     Qwen3NextAttention, Qwen3NextDecoderLayer, Qwen3NextForCausalLM,
     Qwen3NextGatedDeltaNet, Qwen3NextModel, Qwen3NextSparseMoeBlock,
@@ -201,7 +203,13 @@ def _forward(
         spec_query_start_loc = attn_metadata.spec_query_start_loc
         non_spec_query_start_loc = attn_metadata.non_spec_query_start_loc
         spec_sequence_masks = attn_metadata.spec_sequence_masks
-        spec_token_masks = attn_metadata.spec_token_masks
+        if vllm_version_is("0.11.0"):
+            spec_token_masks = attn_metadata.spec_token_masks
+            if spec_token_masks is not None:
+                spec_token_masks = spec_token_masks[:num_actual_tokens]
+        else:
+            spec_token_indx = attn_metadata.spec_token_indx
+            non_spec_token_indx = attn_metadata.non_spec_token_indx
         spec_state_indices_tensor = attn_metadata.spec_state_indices_tensor  # noqa: E501
         non_spec_state_indices_tensor = attn_metadata.non_spec_state_indices_tensor  # noqa: E501
         self_kv_cache = self.kv_cache[forward_context.virtual_engine]
@@ -216,8 +224,6 @@ def _forward(
 
         # 1. Set up dimensions for reshapes later
         projected_states, _ = self.in_proj(hidden_states[:num_actual_tokens])
-        if spec_token_masks is not None:
-            spec_token_masks = spec_token_masks[:num_actual_tokens]
         projected_states_qkvz, projected_states_ba = torch.split(
             projected_states,
             [
@@ -293,10 +299,16 @@ def _forward(
                 g_non_spec = None
                 beta_non_spec = None
             else:
-                g_spec = g[:, spec_token_masks]
-                beta_spec = beta[:, spec_token_masks]
-                g_non_spec = g[:, ~spec_token_masks]
-                beta_non_spec = beta[:, ~spec_token_masks]
+                if vllm_version_is("0.11.0"):
+                    g_spec = g[:, spec_token_masks]
+                    beta_spec = beta[:, spec_token_masks]
+                    g_non_spec = g[:, ~spec_token_masks]
+                    beta_non_spec = beta[:, ~spec_token_masks]
+                else:
+                    g_spec = g.index_select(1, spec_token_indx)
+                    beta_spec = beta.index_select(1, spec_token_indx)
+                    g_non_spec = g.index_select(1, non_spec_token_indx)
+                    beta_non_spec = beta.index_select(1, non_spec_token_indx)
         else:
             g_spec = None
             beta_spec = None
@@ -404,8 +416,15 @@ def _forward(
                 dtype=core_attn_out_non_spec.dtype,
                 device=core_attn_out_non_spec.device,
             )
-            core_attn_out[:, spec_token_masks] = core_attn_out_spec
-            core_attn_out[:, ~spec_token_masks] = core_attn_out_non_spec
+            if vllm_version_is("0.11.0"):
+                core_attn_out[:, spec_token_masks] = core_attn_out_spec
+                core_attn_out[:, ~spec_token_masks] = core_attn_out_non_spec
+            else:
+                core_attn_out.index_copy_(1, spec_token_indx,
+                                          core_attn_out_spec)
+                core_attn_out.index_copy_(1, non_spec_token_indx,
+                                          core_attn_out_non_spec)
+
         elif spec_sequence_masks is not None:
             core_attn_out = core_attn_out_spec
         else: