21b6779a33
### What this PR does / why we need it?
Refactor swiglu and rms_norm unittest case for 310P and 910B.
Apply attention_v1 get_kv_cache_shape and build metadata on all of
platforms
### Does this PR introduce _any_ user-facing change?
NA
### How was this patch tested?
CI UT test
- vLLM version: v0.14.1
- vLLM main:
dc917cceb8
---------
Signed-off-by: pu-zhe <zpuaa@outlook.com>
85 lines
2.9 KiB
Python
85 lines
2.9 KiB
Python
from unittest.mock import MagicMock, patch
|
|
|
|
import pytest
|
|
import torch
|
|
from vllm.config import set_current_vllm_config
|
|
from vllm.model_executor.layers.layernorm import RMSNorm
|
|
|
|
from vllm_ascend.utils import AscendDeviceType, enable_custom_op
|
|
from vllm_ascend.utils import is_310p as is_310p_hw
|
|
|
|
enable_custom_op()
|
|
|
|
|
|
@pytest.fixture
|
|
def dummy_tensor():
|
|
return torch.randn(4, 8, dtype=torch.float16)
|
|
|
|
|
|
def mock_rms_norm(x, weight, eps):
|
|
return x + 1, None
|
|
|
|
|
|
def mock_add_rms_norm(x, residual, weight, eps):
|
|
return 2 * x, None, 2 * residual
|
|
|
|
|
|
def mock_add_rms_norm_bias(x, residual, weight, bias, eps):
|
|
if bias is None:
|
|
return 2 * x, None, 2 * residual
|
|
else:
|
|
return 2 * x + bias, None, 2 * residual
|
|
|
|
|
|
@pytest.fixture(autouse=True)
|
|
def default_vllm_config():
|
|
mock_config = MagicMock()
|
|
mock_config.compilation_config.custom_ops = ["all"]
|
|
|
|
with set_current_vllm_config(mock_config):
|
|
yield mock_config
|
|
|
|
|
|
@pytest.mark.skipif(is_310p_hw(), reason="non_310P device unittest case.")
|
|
@pytest.mark.parametrize("residual", [None, torch.randn(4, 8, dtype=torch.float32)])
|
|
@patch("torch_npu.npu_rms_norm", side_effect=mock_rms_norm)
|
|
@patch("torch_npu.npu_add_rms_norm", side_effect=mock_add_rms_norm)
|
|
@patch("torch.ops._C_ascend.npu_add_rms_norm_bias", side_effect=mock_add_rms_norm_bias)
|
|
def test_RMSNorm_forward(
|
|
mock_add_rms_norm_bias, mock_add_rmsnorm, mock_rmsnorm, residual, dummy_tensor, default_vllm_config
|
|
):
|
|
layer = RMSNorm(hidden_size=8, eps=1e-05)
|
|
if residual is not None:
|
|
out_x, out_residual = layer.forward_oot(dummy_tensor, residual)
|
|
expected_out_x = 2 * dummy_tensor
|
|
expected_out_residual = 2 * residual
|
|
mock_add_rms_norm_bias.assert_called_once()
|
|
assert torch.allclose(out_x, expected_out_x)
|
|
assert torch.allclose(out_residual, expected_out_residual)
|
|
else:
|
|
out_x = layer.forward_oot(dummy_tensor, residual)
|
|
expected_out_x = dummy_tensor + 1
|
|
|
|
mock_rmsnorm.assert_called_once()
|
|
assert torch.allclose(out_x, expected_out_x)
|
|
|
|
|
|
@pytest.mark.skipif(not is_310p_hw(), reason="310P device unittest case.")
|
|
@pytest.mark.parametrize("residual", [None, torch.randn(4, 8, dtype=torch.float16)])
|
|
@patch("torch_npu.npu_rms_norm", side_effect=mock_rms_norm)
|
|
def test_RMSNorm_forward_310p(
|
|
mock_rmsnorm, residual, dummy_tensor, default_vllm_config
|
|
):
|
|
layer = RMSNorm(hidden_size=8, eps=1e-05)
|
|
if residual is not None:
|
|
out_x, out_residual = layer.forward_oot(dummy_tensor, residual)
|
|
expected_out_residual = dummy_tensor + residual
|
|
expected_out_x = expected_out_residual + 1
|
|
mock_rmsnorm.assert_called_once()
|
|
assert torch.allclose(out_x, expected_out_x)
|
|
assert torch.allclose(out_residual, expected_out_residual)
|
|
else:
|
|
out_x = layer.forward_oot(dummy_tensor, residual)
|
|
expected_out_x = dummy_tensor + 1
|
|
mock_rmsnorm.assert_called_once()
|
|
assert torch.allclose(out_x, expected_out_x) |