removing quantization

actualize-ae · Dec 31, 2023 · ac8bd88 · ac8bd88
1 parent aa64610
commit ac8bd88
Showing 1 changed file with 1 addition and 2 deletions.
diff --git a/start.py b/start.py
@@ -61,8 +61,7 @@ def run_vllm(model_id_or_path, tasks):
         "tensor_parallel_size": 8,
         "dtype": 'auto',
         "gpu_memory_utilization": 0.90,
-        "trust_remote_code": True,
-        "quantization": "awq"
+        "trust_remote_code": True
     }
     model_args_str = make_model_args_str(model_args)
     cmd = f"lm_eval --model=vllm \