[torchao] Support quantization configs using module swap (#21982)

Signed-off-by: Jerry Zhang <jerryzh168@gmail.com>

tests/quantization/test_torchao.py

@@ -75,5 +75,25 @@ def test_qwenvl_int8wo_model_loading_with_params(vllm_runner):
         print(output)
 
 
+@pytest.mark.skipif(not TORCHAO_AVAILABLE, reason="torchao is not available")
+@pytest.mark.skip(
+    reason="since torchao nightly is only compatible with torch nightly"
+    "currently https://github.com/pytorch/ao/issues/2919, we'll have to skip "
+    "torchao tests that requires newer versions (0.14.0.dev+) for now")
+def test_opt_125m_awq_int4wo_model_loading_with_params(vllm_runner):
+    torch._dynamo.reset()
+    model_name = ("torchao-testing/opt-125m-AWQConfig-Int4WeightOnlyConfig-v2"
+                  "-0.14.0.dev")
+    with vllm_runner(model_name=model_name,
+                     quantization="torchao",
+                     dtype="bfloat16",
+                     pt_load_map_location="cuda:0") as llm:
+        output = llm.generate_greedy(["The capital of France is"],
+                                     max_tokens=32)
+
+        assert output
+        print(output)
+
+
 if __name__ == "__main__":
     pytest.main([__file__])