Add rope_scaling to Aquila model (vllm-project#1457)

vllm/model_executor/models/aquila.py

@@ -25,7 +25,7 @@
 The input of the model is flattened to a 1D tensor of tokens. The model uses
 InputMetadata to extract the original 2D shape of the input.
 """
-from typing import List, Optional, Tuple
+from typing import Any, Dict, List, Optional, Tuple
 
 import torch
 from torch import nn
@@ -110,6 +110,7 @@ def __init__(
         num_kv_heads: int,
         rope_theta: float = 10000,
         max_position_embeddings: int = 8192,
+        rope_scaling: Optional[Dict[str, Any]] = None,
     ):
         super().__init__()
         self.hidden_size = hidden_size
@@ -148,6 +149,7 @@ def __init__(
             base=self.rope_theta,
             max_position=self.max_position_embeddings,
             num_kv_heads=self.num_kv_heads,
+            rope_scaling=rope_scaling,
         )
 
     def forward(
@@ -173,6 +175,7 @@ def __init__(self, config: AquilaConfig):
         super().__init__()
         self.hidden_size = config.hidden_size
         rope_theta = getattr(config, "rope_theta", 10000)
+        rope_scaling = getattr(config, "rope_scaling", None)
         max_position_embeddings = getattr(config, "max_position_embeddings",
                                           8192)
         self.self_attn = AquilaAttention(
@@ -181,6 +184,7 @@ def __init__(self, config: AquilaConfig):
             num_kv_heads=config.num_key_value_heads,
             rope_theta=rope_theta,
             max_position_embeddings=max_position_embeddings,
+            rope_scaling=rope_scaling,
         )
         self.mlp = AquilaMLP(
             hidden_size=self.hidden_size,