update bench run error

benchmarks/benchmark_throughput.py

@@ -103,9 +103,11 @@ def run_vllm(
         "prompt_token_ids": batch
     } for batch in dummy_prompt_token_ids.tolist()]
     
+    use_beam_search = False
+    
     print("Warming up...")
     for _ in tqdm(range(num_iters_warmup), desc="Warmup iterations"):
-        if not args.use_beam_search:
+        if not use_beam_search:
             llm.generate(dummy_prompts, sampling_params=sampling_params, use_tqdm=False)
         else:
             llm.beam_search(
@@ -117,8 +119,6 @@ def run_vllm(
                 ),
             )
 
-    use_beam_search = False
-
     outputs = None
     if not use_beam_search:
         if args.profile:
@@ -790,4 +790,4 @@ if __name__ == "__main__":
     if args.tokenizer is None:
         args.tokenizer = args.model
     validate_args(args)
-    main(args)
+    main(args)
\ No newline at end of file

vllm/engine/arg_utils.py

@@ -788,6 +788,12 @@ class EngineArgs:
             default=None,
             help="The configurations for speculative decoding. Should be a "
             "JSON string.")
+        parser.add_argument(
+            '--num-speculative-heads',
+            type=int,
+            default=EngineArgs.num_speculative_heads,
+            help='The number of speculative heads to sample from '
+                 'the draft model in speculative decoding.')
 
         # Observability arguments
         observability_kwargs = get_kwargs(ObservabilityConfig)

vllm/perf/benchmark_throughput.py

@@ -103,9 +103,11 @@ def run_vllm(
         "prompt_token_ids": batch
     } for batch in dummy_prompt_token_ids.tolist()]
     
+    use_beam_search = False
+    
     print("Warming up...")
     for _ in tqdm(range(num_iters_warmup), desc="Warmup iterations"):
-        if not args.use_beam_search:
+        if not use_beam_search:
             llm.generate(dummy_prompts, sampling_params=sampling_params, use_tqdm=False)
         else:
             llm.beam_search(
@@ -117,8 +119,6 @@ def run_vllm(
                 ),
             )
 
-    use_beam_search = False
-
     outputs = None
     if not use_beam_search:
         if args.profile: