Upload generate-responses.py with huggingface_hub

generate-responses.py

@@ -2,6 +2,7 @@
 # requires-python = ">=3.10"
 # dependencies = [
 #     "datasets",
+#     "hf_transfer",
 #     "huggingface-hub",
 #     "hf-xet>=1.1.7",
 #     "torch",
@@ -231,15 +232,16 @@ def main(
 
     # Initialize vLLM
     logger.info(f"Loading model: {model_id}")
+    if max_model_len is None:
+        max_model_len = 16384  # sensible default; most prompts are <8K tokens
     vllm_kwargs = {
         "model": model_id,
         "tensor_parallel_size": tensor_parallel_size,
         "gpu_memory_utilization": gpu_memory_utilization,
         "enable_prefix_caching": True,
+        "max_model_len": max_model_len,
     }
-    if max_model_len is not None:
-        vllm_kwargs["max_model_len"] = max_model_len
-        logger.info(f"Using max_model_len={max_model_len}")
+    logger.info(f"Using max_model_len={max_model_len}")
 
     llm = LLM(**vllm_kwargs)
     logger.info("Prefix caching enabled (system prompt shared across all cases)")