60ee4af6d0
### What this PR does / why we need it?
1. Add custom op to nightly tests, fix
https://github.com/vllm-project/vllm-ascend/pull/3665
2. Correctly pass github secrets when using workflow_call, see
https://docs.github.com/en/actions/how-tos/reuse-automations/reuse-workflows
3. Fix the single node mutual cancellation issue
- vLLM version: v0.11.0rc3
- vLLM main:
c9461e05a4
---------
Signed-off-by: wangli <wangli858794774@gmail.com>
38 lines
1.1 KiB
Python
38 lines
1.1 KiB
Python
import pytest
|
|
import torch
|
|
import torch_npu
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
'B',
|
|
[1, 16, 64, 128, 32768],
|
|
)
|
|
@pytest.mark.parametrize(
|
|
'D',
|
|
[8, 16, 32, 64, 128],
|
|
)
|
|
@pytest.mark.parametrize(
|
|
'top_k',
|
|
[1, 2, 4, 8],
|
|
)
|
|
@pytest.mark.parametrize(
|
|
"dtype, atol, rtol",
|
|
[
|
|
(torch.float16, 1e-3, 1e-3),
|
|
(torch.bfloat16, 1e-3, 1e-3),
|
|
],
|
|
)
|
|
def test_quant_fpx_linear(B: int, D: int, top_k: int, dtype, atol, rtol):
|
|
x = torch.rand((B, D), dtype=dtype).to("npu")
|
|
# finished = torch.randint(1, size=(B,), dtype=torch.bool).to("npu")
|
|
finished = None
|
|
y, expert_idx, row_idx = torch_npu.npu_moe_gating_top_k_softmax(x,
|
|
finished,
|
|
k=top_k)
|
|
|
|
topk_weights = x.softmax(dim=-1)
|
|
topk_weights, topk_ids = topk_weights.topk(top_k, dim=-1)
|
|
topk_ids = topk_ids.to(torch.int32)
|
|
torch.allclose(y, topk_weights, atol=atol, rtol=rtol)
|
|
torch.allclose(expert_idx, topk_ids, atol=atol, rtol=rtol)
|