Update deprecated Python 3.8 typing (#13971)

tests/models/decoder_only/language/test_gguf.py

@@ -5,7 +5,7 @@ Note: To pass the test, quantization higher than Q4 should be used
 """
 
 import os
-from typing import List, NamedTuple, Type
+from typing import NamedTuple
 
 import pytest
 from huggingface_hub import hf_hub_download
@@ -90,8 +90,8 @@ MODELS = [
 @pytest.mark.parametrize("tp_size", [1, 2])
 def test_models(
     num_gpus_available: int,
-    vllm_runner: Type[VllmRunner],
-    example_prompts: List[str],
+    vllm_runner: type[VllmRunner],
+    example_prompts: list[str],
     model: GGUFTestConfig,
     dtype: str,
     max_tokens: int,

tests/models/decoder_only/language/test_modelopt.py

@@ -5,7 +5,6 @@
 Note: these tests will only pass on H100
 """
 import os
-from typing import List
 
 import pytest
 from transformers import AutoTokenizer
@@ -65,7 +64,7 @@ def test_models(example_prompts, model_name) -> None:
         for prompt in example_prompts
     ]
     params = SamplingParams(max_tokens=20, temperature=0)
-    generations: List[str] = []
+    generations: list[str] = []
     # Note: these need to be run 1 at a time due to numerical precision,
     # since the expected strs were generated this way.
     for prompt in formatted_prompts: