Drop 0.10.2 (#3284)

Drop v0.10.2 support, we support vLLM 0.11.0rc3 now.
- vLLM version: v0.11.0rc3
- vLLM main:
https://github.com/vllm-project/vllm/commit/releases/v0.11.0

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

vllm_ascend/models/qwen2_5_vl.py

@@ -42,8 +42,6 @@ from vllm.model_executor.models.qwen2_5_vl import (
 from vllm.model_executor.models.utils import maybe_prefix
 from vllm.multimodal import MULTIMODAL_REGISTRY
 
-from vllm_ascend.utils import vllm_version_is
-
 MIN_PAD_SIZE = 64  # min_size to pad weight
 MAX_PAD_SIZE = 128  # max_size to pad weight
 
@@ -498,20 +496,12 @@ class AscendQwen2_5_VLForConditionalGeneration(
         super().__init__(vllm_config=vllm_config, prefix=prefix)
         config: Qwen2_5_VLConfig = vllm_config.model_config.hf_config
         quant_config = vllm_config.quant_config
-        if vllm_version_is("0.10.2"):
-            self.visual = AscendQwen2_5_VisionTransformer(
-                vision_config=config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self._maybe_ignore_quant_config(quant_config),
-                prefix=maybe_prefix(prefix, "visual"),
-            )
-        else:
-            self.visual = AscendQwen2_5_VisionTransformer(
-                vision_config=config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=quant_config,
-                prefix=maybe_prefix(prefix, "visual"),
-            )
+        self.visual = AscendQwen2_5_VisionTransformer(
+            vision_config=config.vision_config,
+            norm_eps=getattr(config, "rms_norm_eps", 1e-6),
+            quant_config=quant_config,
+            prefix=maybe_prefix(prefix, "visual"),
+        )
 
     def _process_image_input(self, image_input) -> tuple[torch.Tensor, ...]:
 

vllm_ascend/models/qwen2_5_vl_without_padding.py

@@ -68,7 +68,6 @@ from vllm.model_executor.models.utils import WeightsMapper, maybe_prefix
 from vllm.multimodal import MULTIMODAL_REGISTRY
 
 from vllm_ascend.models.qwen2_5_vl import AscendQwen2_5_VisionRotaryEmbedding
-from vllm_ascend.utils import vllm_version_is
 
 
 class AscendQwen2_5_VisionAttention_Without_Padding(Qwen2_5_VisionAttention):
@@ -484,20 +483,12 @@ class AscendQwen2_5_VLForConditionalGeneration_Without_Padding(
         super().__init__(vllm_config=vllm_config, prefix=prefix)
         config: Qwen2_5_VLConfig = vllm_config.model_config.hf_config
         quant_config = vllm_config.quant_config
-        if vllm_version_is("0.10.2"):
-            self.visual = AscendQwen2_5_VisionTransformer_Without_Padding(
-                vision_config=config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self._maybe_ignore_quant_config(quant_config),
-                prefix=maybe_prefix(prefix, "visual"),
-            )
-        else:
-            self.visual = AscendQwen2_5_VisionTransformer_Without_Padding(
-                vision_config=config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=quant_config,
-                prefix=maybe_prefix(prefix, "visual"),
-            )
+        self.visual = AscendQwen2_5_VisionTransformer_Without_Padding(
+            vision_config=config.vision_config,
+            norm_eps=getattr(config, "rms_norm_eps", 1e-6),
+            quant_config=quant_config,
+            prefix=maybe_prefix(prefix, "visual"),
+        )
 
     def _process_image_input(self, image_input) -> tuple[torch.Tensor, ...]:
 
@@ -563,20 +554,12 @@ class AscendQwen3VLForConditionalGeneration(Qwen3VLForConditionalGeneration):
         super().__init__(vllm_config=vllm_config, prefix=prefix)
         config: Qwen3VLConfig = vllm_config.model_config.hf_config
         quant_config = vllm_config.quant_config
-        if vllm_version_is("0.10.2"):
-            self.visual = AscendQwen3_VisionTransformer(
-                config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self._maybe_ignore_quant_config(quant_config),
-                prefix=maybe_prefix(prefix, "visual"),
-                use_data_parallel=self.use_data_parallel)
-        else:
-            self.visual = AscendQwen3_VisionTransformer(
-                config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=quant_config,
-                prefix=maybe_prefix(prefix, "visual"),
-                use_data_parallel=self.use_data_parallel)
+        self.visual = AscendQwen3_VisionTransformer(
+            config.vision_config,
+            norm_eps=getattr(config, "rms_norm_eps", 1e-6),
+            quant_config=quant_config,
+            prefix=maybe_prefix(prefix, "visual"),
+            use_data_parallel=self.use_data_parallel)
 
 
 @MULTIMODAL_REGISTRY.register_processor(Qwen3VLMultiModalProcessor,
@@ -613,19 +596,10 @@ class AscendQwen3VLMoeForConditionalGeneration(
         multimodal_config = vllm_config.model_config.multimodal_config
         self.multimodal_config = multimodal_config
         self.use_data_parallel = multimodal_config.mm_encoder_tp_mode == "data"
-        if vllm_version_is("0.10.2"):
-            self.visual = AscendQwen3_VisionTransformer(
-                config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=self._maybe_ignore_quant_config(quant_config),
-                prefix=maybe_prefix(prefix, "visual"),
-                use_data_parallel=self.use_data_parallel,
-            )
-        else:
-            self.visual = AscendQwen3_VisionTransformer(
-                config.vision_config,
-                norm_eps=getattr(config, "rms_norm_eps", 1e-6),
-                quant_config=quant_config,
-                prefix=maybe_prefix(prefix, "visual"),
-                use_data_parallel=self.use_data_parallel,
-            )
+        self.visual = AscendQwen3_VisionTransformer(
+            config.vision_config,
+            norm_eps=getattr(config, "rms_norm_eps", 1e-6),
+            quant_config=quant_config,
+            prefix=maybe_prefix(prefix, "visual"),
+            use_data_parallel=self.use_data_parallel,
+        )

vllm_ascend/models/qwen2_vl.py

@@ -40,8 +40,6 @@ from vllm.model_executor.models.qwen2_vl import (
 from vllm.model_executor.models.utils import maybe_prefix
 from vllm.multimodal import MULTIMODAL_REGISTRY
 
-from vllm_ascend.utils import vllm_version_is
-
 MIN_PAD_SIZE = 64  # min_size to pad weight
 MAX_PAD_SIZE = 128  # max_size to pad weight
 
@@ -345,18 +343,9 @@ class AscendQwen2VLForConditionalGeneration(Qwen2VLForConditionalGeneration):
 
     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__(vllm_config=vllm_config, prefix=prefix)
-        if vllm_version_is("0.10.2"):
-            self.visual = AscendQwen2VisionTransformer(
-                self.config.vision_config,
-                norm_eps=getattr(self.config, "rms_norm_eps", 1e-6),
-                quant_config=self._maybe_ignore_quant_config(
-                    vllm_config.quant_config),
-                prefix=maybe_prefix(prefix, "visual"),
-            )
-        else:
-            self.visual = AscendQwen2VisionTransformer(
-                self.config.vision_config,
-                norm_eps=getattr(self.config, "rms_norm_eps", 1e-6),
-                quant_config=vllm_config.quant_config,
-                prefix=maybe_prefix(prefix, "visual"),
-            )
+        self.visual = AscendQwen2VisionTransformer(
+            self.config.vision_config,
+            norm_eps=getattr(self.config, "rms_norm_eps", 1e-6),
+            quant_config=vllm_config.quant_config,
+            prefix=maybe_prefix(prefix, "visual"),
+        )

vllm_ascend/models/qwen3_moe.py

@@ -47,7 +47,6 @@ from vllm.model_executor.models.utils import (
     make_empty_intermediate_tensors_factory, make_layers, maybe_prefix)
 
 from vllm_ascend.ops.fused_moe import AscendFusedMoE
-from vllm_ascend.utils import vllm_version_is
 
 
 class CustomSparseMoeBlock(Qwen3MoeSparseMoeBlock):
@@ -170,14 +169,8 @@ class CustomQwen3MoeDecoderLayer(Qwen3MoeDecoderLayer):
                                                 quant_config=quant_config,
                                                 prefix=f"{prefix}.mlp")
             else:
-                if vllm_version_is("0.10.2"):
-                    self.mlp = Qwen3MoeSparseMoeBlock(
-                        config=config,
-                        quant_config=quant_config,
-                        prefix=f"{prefix}.mlp")
-                else:
-                    self.mlp = Qwen3MoeSparseMoeBlock(vllm_config=vllm_config,
-                                                      prefix=f"{prefix}.mlp")
+                self.mlp = Qwen3MoeSparseMoeBlock(vllm_config=vllm_config,
+                                                  prefix=f"{prefix}.mlp")
         else:
             self.mlp = Qwen3MoeMLP(hidden_size=config.hidden_size,
                                    intermediate_size=config.intermediate_size,