[refactor] refactor deepseek-related files (#2849)

### What this PR does / why we need it?
This PR deletes ~2K lines of code about deepseek modeling. It falls back
CustomDeepseekV2 modules to original vllm implementations and adapts
some modifications in vllm about deepseek and moe.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
E2E  vllm serving with torchair graph mode and eager mode.

- vLLM version: v0.10.2
- vLLM main:
759ef49b15

---------

Signed-off-by: linfeng-yuan <1102311262@qq.com>
Signed-off-by: Yizhou Liu <liu_yizhou@outlook.com>
Co-authored-by: yiz-liu <136800916+yiz-liu@users.noreply.github.com>
Co-authored-by: Yizhou Liu <liu_yizhou@outlook.com>
This commit is contained in:
linfeng-yuan
2025-09-16 14:13:07 +08:00
committed by GitHub
parent 18ca7861f6
commit 1c5900327b
18 changed files with 295 additions and 1899 deletions

View File

@@ -36,6 +36,7 @@ class TestNPUPlatform(TestBase):
mock_ascend_config = MagicMock()
mock_ascend_config.torchair_graph_config.enabled = False
mock_ascend_config.ascend_scheduler_config.enabled = False
mock_ascend_config.enable_shared_expert_dp = False
return mock_ascend_config
def setUp(self):
@@ -479,6 +480,7 @@ class TestNPUPlatform(TestBase):
def test_get_attn_backend_cls_use_v1_and_mla(self, mock_get_ascend_config):
mock_config = MagicMock()
mock_config.torchair_graph_config.enabled = False
mock_config.enable_shared_expert_dp = False
mock_get_ascend_config.return_value = mock_config