EngineX-Ascend/enginex-ascend-910-vllm
10
0
Fork 1
Code Issues Pull Requests Actions Projects Releases Wiki Activity

init v0.11.0rc0

Browse Source
This commit is contained in:
luopingyi
2025-10-14 10:38:28 +08:00
parent 67afd0ea78
commit 66dc16f966
278 changed files with 28130 additions and 11708 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tests/ut/quantization/test_w8a8.py
View File

@@ -5,8 +5,8 @@ import torch
from tests.ut.base import TestBase
from vllm_ascend.attention.attention_v1 import AscendAttentionState
from vllm_ascend.ops.layers.experts_selector import (_native_grouped_topk,
select_experts)
from vllm_ascend.ops.moe.experts_selector import (_native_grouped_topk,
select_experts)
from vllm_ascend.quantization.w8a8 import (AscendC8KVCacheMethod,
AscendW8A8FusedMoEMethod,
AscendW8A8LinearMethod,
@@ -784,7 +784,7 @@ class TestSelectExperts(TestBase):
self.assertEqual(ids.shape, (self.num_tokens, self.top_k))
self.assertEqual(ids.dtype, torch.int32)
@patch('vllm_ascend.ops.layers.experts_selector._native_grouped_topk')
@patch('vllm_ascend.ops.moe.experts_selector._native_grouped_topk')
def test_grouped_topk_with_correction_bias(self, mock_grouped_topk):
"""Test grouped topk with expert score correction bias"""
mock_grouped_topk.return_value = torch.ones(self.num_tokens,