[ROCm][CI] Fix condition for test_per_token_group_quant_fp8_packed (#39730)
Signed-off-by: Micah Williamson <micah.williamson@amd.com>
This commit is contained in:
@@ -6,6 +6,7 @@ import pytest
|
|||||||
import torch
|
import torch
|
||||||
|
|
||||||
from vllm.model_executor.layers.quantization.utils import fp8_utils, int8_utils
|
from vllm.model_executor.layers.quantization.utils import fp8_utils, int8_utils
|
||||||
|
from vllm.platforms import current_platform
|
||||||
|
|
||||||
|
|
||||||
@pytest.mark.parametrize(
|
@pytest.mark.parametrize(
|
||||||
@@ -80,7 +81,9 @@ def test_per_token_group_quant_fp8(
|
|||||||
],
|
],
|
||||||
)
|
)
|
||||||
@pytest.mark.parametrize("poisoned_scales", [False, True])
|
@pytest.mark.parametrize("poisoned_scales", [False, True])
|
||||||
@pytest.mark.skipif(not torch.cuda.is_available(), reason="CUDA not available")
|
@pytest.mark.skipif(
|
||||||
|
not current_platform.is_cuda(), reason="DeepGEMM not available on this platform"
|
||||||
|
)
|
||||||
def test_per_token_group_quant_fp8_packed(
|
def test_per_token_group_quant_fp8_packed(
|
||||||
num_tokens, hidden_dim, group_size, poisoned_scales
|
num_tokens, hidden_dim, group_size, poisoned_scales
|
||||||
):
|
):
|
||||||
|
|||||||
Reference in New Issue
Block a user