biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Convert formatting to use ruff instead of yapf + isort (#26247)

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-10-05 15:06:22 +01:00
committed by GitHub
parent 17edd8a807
commit d6953beb91
1508 changed files with 115244 additions and 94146 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
tests/kernels/quantization/test_gptq.py
View File

@@ -8,25 +8,22 @@ from vllm import _custom_ops as ops # noqa: F401
def test_gptq_shuffle_opcheck():
weight = torch.randint(-2000000,
2000000, (1792, 4096),
device='cuda',
dtype=torch.int32)
perm = torch.empty((0, ), device='cuda', dtype=torch.int32)
weight = torch.randint(
-2000000, 2000000, (1792, 4096), device="cuda", dtype=torch.int32
)
perm = torch.empty((0,), device="cuda", dtype=torch.int32)
bit = 4
opcheck(torch.ops._C.gptq_shuffle, (weight, perm, bit))
def test_gptq_gemm_opcheck():
a = torch.rand((240, 4096), device='cuda', dtype=torch.float16)
weight = torch.randint(-2000000,
2000000, (512, 6144),
device='cuda',
dtype=torch.int32)
zeros = torch.zeros((32, 768), device='cuda', dtype=torch.int32)
scales = torch.rand((32, 6144), device='cuda', dtype=torch.float16)
idx = torch.empty((0, ), device='cuda', dtype=torch.int32)
a = torch.rand((240, 4096), device="cuda", dtype=torch.float16)
weight = torch.randint(
-2000000, 2000000, (512, 6144), device="cuda", dtype=torch.int32
)
zeros = torch.zeros((32, 768), device="cuda", dtype=torch.int32)
scales = torch.rand((32, 6144), device="cuda", dtype=torch.float16)
idx = torch.empty((0,), device="cuda", dtype=torch.int32)
use_exllama = True
bit = 4
opcheck(torch.ops._C.gptq_gemm,
(a, weight, zeros, scales, idx, use_exllama, bit))
opcheck(torch.ops._C.gptq_gemm, (a, weight, zeros, scales, idx, use_exllama, bit))