sync v0.15.1 (models)

vllm/model_executor/models/afmoe.py

@@ -425,7 +425,7 @@ class AfmoeModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -675,7 +675,7 @@ class AfmoeForCausalLM(nn.Module, SupportsPP, SupportsLoRA):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -694,4 +694,4 @@ class AfmoeForCausalLM(nn.Module, SupportsPP, SupportsLoRA):
         return loader.load_weights(weights, mapper=self.hf_to_vllm_mapper)
 
     def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
-        return self.model.get_expert_mapping()
+        return self.model.get_expert_mapping()
\ No newline at end of file

vllm/model_executor/models/apertus.py

@@ -542,7 +542,7 @@ class ApertusForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -564,4 +564,4 @@ class ApertusForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
             self,
             skip_prefixes=(["lm_head."] if self.config.tie_word_embeddings else None),
         )
-        return loader.load_weights(weights)
+        return loader.load_weights(weights)
\ No newline at end of file

vllm/model_executor/models/arcee.py

@@ -394,7 +394,7 @@ class ArceeForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -425,4 +425,4 @@ class ArceeForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
         )
         # AutoWeightLoader handles weight name remapping, including fusing
         # separate q_proj, k_proj, v_proj into qkv_proj
-        return loader.load_weights(weights)
+        return loader.load_weights(weights)
\ No newline at end of file

vllm/model_executor/models/arctic.py

@@ -406,7 +406,7 @@ class ArcticModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -460,7 +460,7 @@ class ArcticForCausalLM(nn.Module, SupportsPP, SupportsQuant):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -591,4 +591,4 @@ class ArcticForCausalLM(nn.Module, SupportsPP, SupportsQuant):
                         )
                         weight_loader(param, loaded_weight)
             loaded_params.add(name)
-        return loaded_params
+        return loaded_params
\ No newline at end of file

vllm/model_executor/models/aria.py

@@ -629,7 +629,7 @@ class AriaForConditionalGeneration(nn.Module, SupportsMultiModal):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -656,4 +656,4 @@ class AriaForConditionalGeneration(nn.Module, SupportsMultiModal):
 
     def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]):
         loader = AutoWeightsLoader(self)
-        loader.load_weights(weights, mapper=self.hf_to_vllm_mapper)
+        loader.load_weights(weights, mapper=self.hf_to_vllm_mapper)
\ No newline at end of file

vllm/model_executor/models/audioflamingo3.py

@@ -609,7 +609,7 @@ class AudioFlamingo3ForConditionalGeneration(
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -634,4 +634,4 @@ class AudioFlamingo3ForConditionalGeneration(
 
     def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
         loader = AutoWeightsLoader(self)
-        return loader.load_weights(weights)
+        return loader.load_weights(weights)
\ No newline at end of file

vllm/model_executor/models/aya_vision.py

@@ -420,7 +420,7 @@ class AyaVisionForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsP
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -441,4 +441,4 @@ class AyaVisionForConditionalGeneration(nn.Module, SupportsMultiModal, SupportsP
         self,
         hidden_states: torch.Tensor,
     ) -> torch.Tensor | None:
-        return self.language_model.compute_logits(hidden_states)
+        return self.language_model.compute_logits(hidden_states)
\ No newline at end of file

vllm/model_executor/models/bagel.py

@@ -507,7 +507,7 @@ class BagelForConditionalGeneration(
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -581,4 +581,4 @@ class BagelForConditionalGeneration(
 
         # Skip vit_pos_embed.pos_embed as it's handled by PositionEmbedding module
         loader = AutoWeightsLoader(self, skip_prefixes=["vit_pos_embed.pos_embed"])
-        return loader.load_weights(filtered_weights, mapper=self.hf_to_vllm_mapper)
+        return loader.load_weights(filtered_weights, mapper=self.hf_to_vllm_mapper)
\ No newline at end of file

vllm/model_executor/models/baichuan.py

@@ -334,7 +334,7 @@ class BaiChuanModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -534,7 +534,7 @@ class BaiChuanBaseForCausalLM(nn.Module, SupportsLoRA, SupportsPP, SupportsQuant
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/bailing_moe.py

@@ -440,7 +440,7 @@ class BailingMoeModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         position_ids: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -611,7 +611,7 @@ class BailingMoeForCausalLM(nn.Module, SupportsPP, SupportsLoRA):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -640,4 +640,4 @@ class BailingMoeForCausalLM(nn.Module, SupportsPP, SupportsLoRA):
 
 
 class BailingMoeV2ForCausalLM(BailingMoeForCausalLM):
-    pass
+    pass
\ No newline at end of file

vllm/model_executor/models/bamba.py

@@ -311,7 +311,7 @@ class BambaModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -493,7 +493,7 @@ class BambaForCausalLM(
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -514,4 +514,4 @@ class BambaForCausalLM(
 
     def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
         loader = AutoWeightsLoader(self)
-        return loader.load_weights(weights)
+        return loader.load_weights(weights)
\ No newline at end of file

vllm/model_executor/models/bert_with_rope.py

@@ -475,7 +475,7 @@ class BertWithRope(nn.Module, SupportsQuant):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -726,4 +726,4 @@ class GteNewForSequenceClassification(nn.Module, SupportsCrossEncoding):
             positions=positions,
             inputs_embeds=inputs_embeds,
             intermediate_tensors=intermediate_tensors,
-        )
+        )
\ No newline at end of file

vllm/model_executor/models/blip2.py

@@ -641,7 +641,7 @@ class Blip2ForConditionalGeneration(
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -727,4 +727,4 @@ class Blip2ForConditionalGeneration(
             "the number of tokens per image."
         )
         num_images = num_vision_tokens / self._vision_tokens_per_image
-        return num_images * self.config.num_query_tokens
+        return num_images * self.config.num_query_tokens
\ No newline at end of file

vllm/model_executor/models/bloom.py

@@ -294,7 +294,7 @@ class BloomModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         position_ids: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -412,7 +412,7 @@ class BloomForCausalLM(nn.Module, SupportsPP, SupportsQuant):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/chameleon.py

@@ -994,7 +994,7 @@ class ChameleonForConditionalGeneration(
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -1100,4 +1100,4 @@ class ChameleonForConditionalGeneration(
                 weight_loader = getattr(param, "weight_loader", default_weight_loader)
                 weight_loader(param, loaded_weight)
             loaded_params.add(name)
-        return loaded_params
+        return loaded_params
\ No newline at end of file

vllm/model_executor/models/chatglm.py

@@ -381,7 +381,7 @@ class ChatGLMModel(nn.Module, SupportsQuant):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -554,7 +554,7 @@ class ChatGLMForCausalLM(ChatGLMBaseModel, SupportsLoRA, SupportsPP, SupportsQua
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,

vllm/model_executor/models/cohere2_vision.py

@@ -446,7 +446,7 @@ class Cohere2VisionForConditionalGeneration(nn.Module, SupportsMultiModal, Suppo
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -467,4 +467,4 @@ class Cohere2VisionForConditionalGeneration(nn.Module, SupportsMultiModal, Suppo
         self,
         hidden_states: torch.Tensor,
     ) -> torch.Tensor | None:
-        return self.language_model.compute_logits(hidden_states)
+        return self.language_model.compute_logits(hidden_states)
\ No newline at end of file

vllm/model_executor/models/commandr.py

@@ -312,7 +312,7 @@ class CohereModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -438,7 +438,7 @@ class CohereForCausalLM(nn.Module, SupportsLoRA, SupportsPP, SupportsQuant):
     @torch.no_grad()
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -466,4 +466,4 @@ class CohereForCausalLM(nn.Module, SupportsLoRA, SupportsPP, SupportsQuant):
         loader = AutoWeightsLoader(
             self, skip_prefixes=["lm_head", "rotary_emb.inv_freq"]
         )
-        return loader.load_weights(weights)
+        return loader.load_weights(weights)
\ No newline at end of file

vllm/model_executor/models/dbrx.py

@@ -361,7 +361,7 @@ class DbrxModel(nn.Module):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         position_ids: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None,
         inputs_embeds: torch.Tensor | None = None,
@@ -462,7 +462,7 @@ class DbrxForCausalLM(nn.Module, SupportsPP):
 
     def forward(
         self,
-        input_ids: torch.Tensor | None,
+        input_ids: torch.Tensor,
         positions: torch.Tensor,
         intermediate_tensors: IntermediateTensors | None = None,
         inputs_embeds: torch.Tensor | None = None,
@@ -481,4 +481,4 @@ class DbrxForCausalLM(nn.Module, SupportsPP):
 
     def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
         loader = AutoWeightsLoader(self)
-        return loader.load_weights(weights)
+        return loader.load_weights(weights)
\ No newline at end of file