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

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

vllm/model_executor/models/olmo.py

@@ -248,6 +248,9 @@ class OlmoModel(nn.Module):
             make_empty_intermediate_tensors_factory(["hidden_states"],
                                                     config.hidden_size))
 
+    def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
+        return self.embed_tokens(input_ids)
+
     def forward(
         self,
         input_ids: torch.Tensor,
@@ -255,17 +258,16 @@ class OlmoModel(nn.Module):
         kv_caches: List[torch.Tensor],
         attn_metadata: AttentionMetadata,
         intermediate_tensors: Optional[IntermediateTensors],
+        inputs_embeds: Optional[torch.Tensor] = None,
     ) -> Union[torch.Tensor, IntermediateTensors]:
         """
         :param input_ids: A tensor of shape `(batch_size, seq_len)`.
         """
         if get_pp_group().is_first_rank:
-            # Get embeddings of input.
-            # shape: (batch_size, seq_len, d_model)
-            inputs_embeds = self.embed_tokens(input_ids)
-
-            # embed positions
-            hidden_states = inputs_embeds
+            if inputs_embeds is not None:
+                hidden_states = inputs_embeds
+            else:
+                hidden_states = self.get_input_embeddings(input_ids)
         else:
             assert intermediate_tensors is not None
             hidden_states = intermediate_tensors["hidden_states"]
@@ -315,6 +317,9 @@ class OlmoForCausalLM(nn.Module, SupportsPP):
         self.make_empty_intermediate_tensors = (
             self.model.make_empty_intermediate_tensors)
 
+    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,
@@ -322,6 +327,7 @@ class OlmoForCausalLM(nn.Module, SupportsPP):
         kv_caches: List[torch.Tensor],
         attn_metadata: AttentionMetadata,
         intermediate_tensors: Optional[IntermediateTensors] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
     ) -> Union[torch.Tensor, IntermediateTensors]:
         hidden_states = self.model(
             input_ids=input_ids,
@@ -329,6 +335,7 @@ class OlmoForCausalLM(nn.Module, SupportsPP):
             kv_caches=kv_caches,
             attn_metadata=attn_metadata,
             intermediate_tensors=intermediate_tensors,
+            inputs_embeds=inputs_embeds,
         )
         return hidden_states