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

13
tests/ut/models/test_deepseek_mtp.py
View File

@@ -13,10 +13,13 @@ from vllm_ascend.models.deepseek_mtp import (
class TestCustomDeepSeekMultiTokenPredictorLayer(PytestBase):
@pytest.fixture
def setup_mtp_layer(self, mocker: MockerFixture):
def setup_mtp_layer(self, mocker: MockerFixture, vllm_config: VllmConfig,
mock_distributed):
config = PretrainedConfig(vocab_size=1000,
hidden_size=768,
rms_norm_eps=1e-5)
mocker.patch("vllm_ascend.models.deepseek_mtp.get_current_vllm_config",
return_value=vllm_config)
mocker.patch(
"vllm.model_executor.layers.vocab_parallel_embedding.VocabParallelEmbedding.__init__",
return_value=None)
@@ -29,15 +32,15 @@ class TestCustomDeepSeekMultiTokenPredictorLayer(PytestBase):
"vllm_ascend.models.deepseek_mtp.CustomDeepSeekShareHead.__init__",
return_value=None)
mocker_deepseek_v2_decode_layer = mocker.patch(
"vllm_ascend.models.deepseek_v2.CustomDeepseekV2DecoderLayer.__init__",
"vllm.model_executor.models.deepseek_v2.DeepseekV2DecoderLayer.__init__",
return_value=None)
mocker.patch(
"vllm_ascend.ops.vocab_parallel_embedding.AscendVocabParallelEmbedding.__init__",
return_value=None)
mocker.patch("vllm_ascend.utils.get_ascend_config",
mocker.patch("vllm_ascend.models.deepseek_v2.get_ascend_config",
return_value=mocker.Mock())
mtp_layer = CustomDeepSeekMultiTokenPredictorLayer(config, "", None)
mtp_layer = CustomDeepSeekMultiTokenPredictorLayer(config, "0", None)
mocker_deepseek_v2_decode_layer.assert_called_once()
return mtp_layer
@@ -165,8 +168,6 @@ class TestCustomDeepSeekMTP(PytestBase):
mocker.patch(
"vllm_ascend.models.deepseek_mtp.CustomDeepSeekMultiTokenPredictorLayer.__call__",
return_value=None)
mocker.patch("vllm.model_executor.layers.sampler.get_sampler",
return_value=None)
mocker.patch(
"vllm_ascend.ops.vocab_parallel_embedding.AscendVocabParallelEmbedding.__init__",
return_value=None)