Fix config for Falcon (#1164)

vllm/config.py

@@ -135,7 +135,8 @@ class ModelConfig:
         # FIXME(woosuk): This may not be true for all models.
         return self.hf_config.hidden_size // self.hf_config.num_attention_heads
 
-    def get_num_heads(self, parallel_config: "ParallelConfig") -> int:
+    def get_num_kv_heads(self, parallel_config: "ParallelConfig") -> int:
+        """Returns the number of KV heads per GPU worker."""
         # For GPTBigCode & Falcon:
         # Note: for falcon, when new_decoder_architecture is True, the
         # multi_query flag is ignored and we use n_head_kv for the number of
@@ -147,11 +148,15 @@ class ModelConfig:
         if not new_decoder_arch_falcon and getattr(self.hf_config,
                                                    "multi_query", False):
             # Multi-query attention, only one KV head.
+            # Currently, tensor parallelism is not supported in this case.
             return 1
         # For Falcon:
         if getattr(self.hf_config, "n_head_kv", None) is not None:
             return (self.hf_config.n_head_kv //
                     parallel_config.tensor_parallel_size)
+        if getattr(self.hf_config, "num_kv_heads", None) is not None:
+            return (self.hf_config.num_kv_heads //
+                    parallel_config.tensor_parallel_size)
         # For LLaMA-2:
         if getattr(self.hf_config, "num_key_value_heads", None) is not None:
             return (self.hf_config.num_key_value_heads //

vllm/worker/cache_engine.py

@@ -33,7 +33,7 @@ class CacheEngine:
 
         self.head_size = model_config.get_head_size()
         self.num_layers = model_config.get_num_layers(parallel_config)
-        self.num_heads = model_config.get_num_heads(parallel_config)
+        self.num_heads = model_config.get_num_kv_heads(parallel_config)
         self.dtype = model_config.dtype
 
         self.block_size = cache_config.block_size
@@ -146,7 +146,7 @@ class CacheEngine:
         parallel_config: ParallelConfig,
     ) -> int:
         head_size = model_config.get_head_size()
-        num_heads = model_config.get_num_heads(parallel_config)
+        num_heads = model_config.get_num_kv_heads(parallel_config)
         num_layers = model_config.get_num_layers(parallel_config)
 
         key_cache_block = block_size * num_heads * head_size