fix: update bench_speculative (#5649)

python/sglang/bench_serving.py

@@ -993,6 +993,8 @@ async def benchmark(
         async with semaphore:
             return await request_func(request_func_input=request_func_input, pbar=pbar)
 
+    if not hasattr(args, "warmup_requests"):
+        args.warmup_requests = 1
     # Warmup
     print(f"Starting warmup with {args.warmup_requests} sequences...")
 

scripts/playground/bench_speculative.py

@@ -85,7 +85,7 @@ def send_one_batch(base_url, num_prompts, batch_size):
             request_rate=float("inf"),
             max_concurrency=batch_size,
             disable_tqdm=False,
-            lora_name=None,
+            lora_names=None,
             extra_request_body={},
             profile=None,
         )