| 12345678910111213141516171819202122232425262728 |
- # flake8: noqa
- """Tests experts_int8 quantization startup and generation,
- doesn't test correctness
- """
- import pytest
- from tests.quantization.utils import is_quant_method_supported
- MODELS = ["ai21labs/Jamba-tiny-random"]
- @pytest.mark.skipif(not is_quant_method_supported("experts_int8"),
- reason="ExpertsInt8 is not supported on this GPU type.")
- @pytest.mark.parametrize("model", MODELS)
- @pytest.mark.parametrize("dtype", ["bfloat16"])
- @pytest.mark.parametrize("max_tokens", [10])
- def test_model_experts_int8_startup(
- hf_runner,
- aphrodite_runner,
- example_prompts,
- model: str,
- dtype: str,
- max_tokens: int,
- ) -> None:
- with aphrodite_runner(model, dtype=dtype,
- quantization="experts_int8") as aphrodite_model:
- aphrodite_model.generate_greedy(example_prompts, max_tokens)
|
