[V1] Refactor model executable interface for all text-only language models (#10374)

Signed-off-by: Roger Wang <ywang@roblox.com>

vllm/model_executor/models/granite.py

@@ -409,6 +409,9 @@ class GraniteForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
         else:
             self.lm_head = PPMissingLayer()
 
+    def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
+        return self.model.get_input_embeddings(input_ids)
+
     def forward(
         self,
         input_ids: torch.Tensor,
@@ -416,9 +419,11 @@ class GraniteForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
         kv_caches: List[torch.Tensor],
         attn_metadata: AttentionMetadata,
         intermediate_tensors: Optional[IntermediateTensors] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
     ) -> Union[torch.Tensor, IntermediateTensors]:
         model_output = self.model(input_ids, positions, kv_caches,
-                                  attn_metadata, intermediate_tensors)
+                                  attn_metadata, intermediate_tensors,
+                                  inputs_embeds)
         return model_output
 
     def compute_logits(