frozenleaves/vllm
1
0
Fork 0
mirror of https://github.com/vllm-project/vllm.git synced 2025-10-20 14:53:52 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Chore] Remove unused PolyNorm layer (#27110)

Browse Source 
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
This commit is contained in:
Isotr0py
2025-10-18 03:03:43 +08:00
committed by GitHub
parent e33ee23ee3
commit 3125d79950
7 changed files with 1 additions and 524 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
tests/kernels/core/test_layernorm.py
View File

@ -6,7 +6,7 @@ import torch
from tests.kernels.quant_utils import FP8_DTYPE
from tests.kernels.utils import opcheck
from vllm.model_executor.layers.layernorm import PolyNorm, RMSNorm
from vllm.model_executor.layers.layernorm import RMSNorm
from vllm.platforms import current_platform
DTYPES = [torch.half, torch.bfloat16, torch.float]
@ -70,38 +70,6 @@ def test_rms_norm(
)
@pytest.mark.parametrize("num_tokens", NUM_TOKENS)
@pytest.mark.parametrize("hidden_size", HIDDEN_SIZES)
@pytest.mark.parametrize("dtype", DTYPES)
@pytest.mark.parametrize("seed", SEEDS)
@pytest.mark.parametrize("device", CUDA_DEVICES)
@torch.inference_mode()
def test_poly_norm(
num_tokens: int,
hidden_size: int,
dtype: torch.dtype,
seed: int,
device: str,
) -> None:
current_platform.seed_everything(seed)
torch.set_default_device(device)
layer = PolyNorm().to(dtype=dtype)
layer.weight.data.normal_(mean=1.0, std=0.1)
layer.bias.data.normal_(mean=1.0, std=0.1)
scale = 1 / (2 * hidden_size)
x = torch.randn(num_tokens, hidden_size, dtype=dtype)
x *= scale
ref_out = layer.forward_native(x)
out = layer(x)
torch.testing.assert_close(out, ref_out, atol=1e-2, rtol=1e-2)
opcheck(
torch.ops._C.poly_norm,
(out, x, layer.weight.data, layer.bias.data, layer.variance_epsilon),
)
@pytest.mark.parametrize("num_tokens", NUM_TOKENS)
@pytest.mark.parametrize("hidden_size", HIDDEN_SIZES)
@pytest.mark.parametrize("add_residual", ADD_RESIDUAL)