178ca1607e
### What this PR does / why we need it?
The main goal of this PR to alleviate the high maintenance burden from
model duplication when we are going to do the model optimization. Some
of our optimized models diverges a little from the vllm's modeling, but
needs to rewrite several part of original one, brings negligible
maintenance bruden to the vllm-ascend.In order to solve that, we propose
to leverage `torch.compile` and `inductor pattern matcher`,
automatically fuse the pattern we want to merge. For more details can
refer to the RFC https://github.com/vllm-project/vllm-ascend/issues/4239
This pr integrates `AddRMSNorm` and the `Quant` operator, which can
improve the inference speed of models using `w8a8 `quantization.
### Does this PR introduce _any_ user-facing change?
Yes, add new additional_config
### How was this patch tested?
```python
def main():
prompts = [
"The president of the United States is Mr.",
]
# Create a sampling params object.
sampling_params = SamplingParams(max_tokens=100, temperature=0.6, top_k=40, top_p=0.95)
# Create an LLM.
llm = LLM(
model="/root/.cache/modelscope/hub/models/vllm-ascend/Qwen3-8B-W8A8",
# enforce_eager=True,
tensor_parallel_size=1,
trust_remote_code=True,
gpu_memory_utilization=0.7,
quantization="ascend",
)
# Generate texts from the prompts.
outputs = llm.generate(prompts, sampling_params)
for output in outputs:
prompt = output.prompt
generated_text = output.outputs[0].text
print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
```
```text
Prompt: 'The president of the United States is Mr.', Generated text: ' Trump. The president of the United States is Mr. Biden. Which of the following statements is correct? \n\nA. Mr. Trump is Mr. Biden. \nB. Mr. Trump is not Mr. Biden. \nC. The president of the United States is not Mr. Trump. \nD. The president of the United States is not Mr. Biden.\n\nThe question presents a contradiction: it states that "The president of the United States is Mr. Trump" and "The president of'
```
- vLLM version: 86e178f7c4d8c3b0eaf3c8e3f810a83f63b90e24
- vLLM main:
86e178f7c4
---------
Signed-off-by: Icey <1790571317@qq.com>
Signed-off-by: wxsIcey <1790571317@qq.com>
58 lines
2.0 KiB
Python
58 lines
2.0 KiB
Python
from unittest.mock import patch
|
|
|
|
import pytest
|
|
import torch
|
|
from vllm.model_executor.layers.layernorm import RMSNorm
|
|
|
|
from vllm_ascend.utils import AscendDeviceType
|
|
|
|
|
|
@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
|
|
|
|
|
|
@pytest.mark.parametrize("is_310p", [True, False])
|
|
@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)
|
|
def test_RMSNorm_forward(mock_add_rmsnorm, mock_rmsnorm, is_310p, residual,
|
|
dummy_tensor):
|
|
|
|
with patch("vllm_ascend.utils.get_ascend_device_type",
|
|
return_value=AscendDeviceType._310P
|
|
if is_310p else AscendDeviceType._910_93):
|
|
layer = RMSNorm(hidden_size=8, eps=1e-05)
|
|
if residual is not None:
|
|
out_x, out_residual = layer.forward_oot(dummy_tensor, residual)
|
|
|
|
if is_310p:
|
|
expected_arg_x = dummy_tensor + residual.to(dummy_tensor.dtype)
|
|
expected_out_x = expected_arg_x + 1
|
|
expected_out_residual = expected_arg_x.to(residual.dtype)
|
|
|
|
mock_rmsnorm.assert_called_once()
|
|
assert torch.allclose(out_x, expected_out_x)
|
|
assert torch.allclose(out_residual, expected_out_residual)
|
|
else:
|
|
expected_out_x = 2 * dummy_tensor
|
|
expected_out_residual = 2 * residual
|
|
mock_add_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)
|