Merge tag 'v0.15.0rc1' into v0.15.0rc1-dev

vllm/model_executor/models/granitemoehybrid.py

@@ -368,7 +368,7 @@ class GraniteMoeHybridModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -685,7 +685,7 @@ class GraniteMoeHybridForCausalLM(
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/granitemoeshared.py

@@ -183,7 +183,7 @@ class GraniteMoeSharedModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -295,7 +295,7 @@ class GraniteMoeSharedForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/grok1.py

@@ -491,7 +491,7 @@ class Grok1Model(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -705,7 +705,7 @@ class GrokBaseForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/hunyuan_v1.py

@@ -938,7 +938,7 @@ class HunyuanV1ModelBase(nn.Module, SupportsLoRA, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/hunyuan_vision.py

@@ -968,7 +968,7 @@ class HunYuanVLForConditionalGeneration(
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None,

vllm/model_executor/models/hyperclovax_vision.py

@@ -747,7 +747,7 @@ class HCXVisionForCausalLM(nn.Module, SupportsMultiModal, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/idefics3.py

@@ -559,7 +559,7 @@ class Idefics3Model(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -689,7 +689,7 @@ class Idefics3ForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsLo
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/interfaces.py

@@ -607,6 +607,8 @@ class SupportsPP(Protocol):
 
     def forward(
         self,
+        input_ids: Tensor | None,
+        positions: Tensor,
         *,
         intermediate_tensors: IntermediateTensors | None,
     ) -> IntermediateTensors | None:
@@ -635,6 +637,8 @@ class _SupportsPPType(Protocol):
 
     def forward(
         self,
+        input_ids: Tensor | None,
+        positions: Tensor,
         *,
         intermediate_tensors: IntermediateTensors | None,
     ) -> Tensor | IntermediateTensors: ...

vllm/model_executor/models/internlm2.py

@@ -284,7 +284,7 @@ class InternLM2Model(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -350,7 +350,7 @@ class InternLM2ForCausalLM(nn.Module, SupportsPP, SupportsLoRA):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -446,7 +446,7 @@ class InternLM2ForRewardModel(InternLM2ForCausalLM):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/internlm2_ve.py

@@ -101,7 +101,7 @@ class InternLM2VEModel(InternLM2Model):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/interns1.py

@@ -782,7 +782,7 @@ class InternS1ForConditionalGeneration(
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/internvl.py

@@ -1371,7 +1371,7 @@ class InternVLChatModel(nn.Module, SupportsMultiModal, SupportsPP, SupportsLoRA)
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/iquest_loopcoder.py

@@ -438,7 +438,7 @@ class IQuestLoopCoderModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -570,7 +570,7 @@ class IQuestLoopCoderForCausalLM(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/isaac.py

@@ -1450,7 +1450,7 @@ class IsaacForConditionalGeneration(
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/jais.py

@@ -280,7 +280,7 @@ class JAISModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         position_ids: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -344,7 +344,7 @@ class JAISLMHeadModel(nn.Module, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/jais2.py

@@ -483,7 +483,7 @@ class Jais2ForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/jamba.py

@@ -348,7 +348,7 @@ class JambaModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -516,7 +516,7 @@ class JambaForCausalLM(
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/jina_vl.py

@@ -125,7 +125,7 @@ class JinaVLForSequenceClassification(
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/kanana_v.py

@@ -732,7 +732,7 @@ class KananaVForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsPP)
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/keye.py

@@ -1438,7 +1438,7 @@ class BaseKeyeModule(nn.Module, SupportsMultiModal):
 
     def forward(
         self,
-        input_ids: torch.Tensor,
+        input_ids: torch.Tensor | None,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,