Use float32 for test_completion.py (vllm-project#22385)

Signed-off-by: Michael Goin <[email protected]>

Author: mgoin

tests/v1/entrypoints/openai/test_completion.py

@@ -20,9 +20,8 @@
 @pytest.fixture(scope="module")
 def default_server_args():
     return [
-        # use half precision for speed and memory savings in CI environment
         "--dtype",
-        "bfloat16",
+        "float32",
         "--max-model-len",
         "2048",
         "--max-num-seqs",