EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[MM][Bugfix] Update hf_config to hf_text_config (#5319)

Browse Source 
### What this PR does / why we need it?

Following https://github.com/vllm-project/vllm-ascend/pull/5205, update
`hf_config` to `hf_text_config`.

Find more details at
https://github.com/vllm-project/vllm-ascend/pull/5205#issuecomment-3675417534
and
https://github.com/vllm-project/vllm-ascend/pull/5205#issuecomment-3677920872.

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: release/v0.13.0
- vLLM main:
5fbfa8d9ef

Signed-off-by: shen-shanshan <467638484@qq.com>
This commit is contained in:
Shanshan Shen
2026-01-06 16:41:39 +08:00
committed by GitHub
parent 293b2275df
commit b94d589769
23 changed files with 44 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/ut/core/test_scheduler_dynamic_batch.py
View File

@@ -130,8 +130,8 @@ class TestSchedulerDynamicBatch(TestBase):
)
model_config.pooler_config = MagicMock()
model_config.multimodal_config = MagicMock()
model_config.hf_config = MagicMock()
model_config.hf_config.is_encoder_decoder = False
model_config.hf_text_config = MagicMock()
model_config.hf_text_config.is_encoder_decoder = False
# Cache config, optionally force APC
kwargs_cache: Dict[str,
Any] = ({} if ENABLE_PREFIX_CACHING is None else {

4
tests/ut/ops/test_mla.py
View File

@@ -87,7 +87,7 @@ class TestAscendMultiHeadLatentAttention(TestBase):
mock_tp_size.return_value = 2
mock_ascend_config.return_value.enable_shared_expert_dp = True
mock_vllm_config = MagicMock(spec=VllmConfig)
mock_vllm_config.model_config.hf_config = MagicMock(
mock_vllm_config.model_config.hf_text_config = MagicMock(
num_hidden_layers=32, first_k_dense_replace=True)
mock_get_vllm_config.return_value = mock_vllm_config
mock_vllm_config.compilation_config = CompilationConfig()
@@ -122,7 +122,7 @@ class TestAscendMultiHeadLatentAttention(TestBase):
mock_tp_size.return_value = 1
mock_ascend_config.return_value.enable_shared_expert_dp = False
mock_vllm_config = MagicMock(spec=VllmConfig)
mock_vllm_config.model_config.hf_config = MagicMock(
mock_vllm_config.model_config.hf_text_config = MagicMock(
num_hidden_layers=32, first_k_dense_replace=False)
mock_get_vllm_config.return_value = mock_vllm_config
mock_vllm_config.compilation_config = CompilationConfig()

12
tests/ut/ops/test_rotary_embedding.py
View File

@@ -115,7 +115,7 @@ class TestAscendRotaryEmbedding(unittest.TestCase):
model_config = ModelConfig(MODEL,
tokenizer=MODEL,
max_model_len=MAX_NUM_BATCHED_TOKEND)
model_config.hf_config = PretrainedConfig()
model_config.hf_text_config = PretrainedConfig()
vllm_config.model_config = model_config
with set_ascend_forward_context(None, vllm_config):
result_q, result_k = self.layer.forward(self.positions, self.query,
@@ -141,7 +141,7 @@ class TestAscendRotaryEmbedding(unittest.TestCase):
model_config = ModelConfig(MODEL,
tokenizer=MODEL,
max_model_len=MAX_NUM_BATCHED_TOKEND)
model_config.hf_config = PretrainedConfig()
model_config.hf_text_config = PretrainedConfig()
vllm_config.model_config = model_config
with set_ascend_forward_context(None, vllm_config):
result_q, result_k = self.layer.forward(self.positions,
@@ -164,7 +164,7 @@ class TestAscendRotaryEmbedding(unittest.TestCase):
model_config = ModelConfig(MODEL,
tokenizer=MODEL,
max_model_len=MAX_NUM_BATCHED_TOKEND)
model_config.hf_config = PretrainedConfig()
model_config.hf_text_config = PretrainedConfig()
vllm_config.model_config = model_config
with set_ascend_forward_context(None, vllm_config):
self.layer.forward(self.positions, self.query, self.key,
@@ -184,7 +184,7 @@ class TestAscendRotaryEmbedding(unittest.TestCase):
model_config = ModelConfig(MODEL,
tokenizer=MODEL,
max_model_len=MAX_NUM_BATCHED_TOKEND)
model_config.hf_config = PretrainedConfig()
model_config.hf_text_config = PretrainedConfig()
vllm_config.model_config = model_config
with set_ascend_forward_context(None, vllm_config):
result_q, result_k = self.layer.forward(
@@ -213,7 +213,7 @@ class TestAscendRotaryEmbedding(unittest.TestCase):
model_config = ModelConfig(MODEL,
tokenizer=MODEL,
max_model_len=MAX_NUM_BATCHED_TOKEND)
model_config.hf_config = PretrainedConfig()
model_config.hf_text_config = PretrainedConfig()
vllm_config.model_config = model_config
with set_ascend_forward_context(None, vllm_config):
result_q, result_k = self.layer.forward(self.positions, self.query,
@@ -404,7 +404,7 @@ class TestAscendMRotaryEmbedding(unittest.TestCase):
model_config = ModelConfig(MODEL_VL,
tokenizer=MODEL_VL,
max_model_len=MAX_NUM_BATCHED_TOKEND)
model_config.hf_config = PretrainedConfig()
model_config.hf_text_config = PretrainedConfig()
vllm_config.model_config = model_config
return vllm_config

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

@@ -79,7 +79,7 @@ class TestAscendQuantConfig(TestBase):
def test_get_quant_method_for_linear(self):
mock_config = MagicMock()
mock_config.model_config.hf_config.model_type = None
mock_config.model_config.hf_text_config.model_type = None
linear_layer = MagicMock(spec=LinearBase)
# Test skipped layer
with patch("vllm_ascend.quantization.quant_config.get_current_vllm_config", return_value=mock_config), \
@@ -103,7 +103,7 @@ class TestAscendQuantConfig(TestBase):
def test_get_quant_method_for_attention(self):
attention_layer = MagicMock(spec=Attention)
mock_config = MagicMock()
mock_config.model_config.hf_config.model_type = None
mock_config.model_config.hf_text_config.model_type = None
with patch("vllm_ascend.quantization.quant_config.get_current_vllm_config", return_value=mock_config), \
patch('vllm_ascend.quantization.quant_config.AscendKVCacheMethod', \
return_value=MagicMock()) as mock_ascend_kvcache:
@@ -117,7 +117,7 @@ class TestAscendQuantConfig(TestBase):
fused_moe_layer.moe = MagicMock(spec=FusedMoEConfig)
fused_moe_layer.moe_config = MagicMock(spec=FusedMoEConfig)
mock_config = MagicMock()
mock_config.model_config.hf_config.model_type = None
mock_config.model_config.hf_text_config.model_type = None
# Test skipped layer
with patch.object(self.ascend_config, 'is_layer_skipped_ascend', return_value=True), \

2
tests/ut/spec_decode/test_mtp_proposer.py
View File

@@ -41,7 +41,7 @@ class TestMtpProposer:
config.model_config.dtype = torch.float16
config.model_config.max_model_len = 2048
config.model_config.uses_mrope = False
config.model_config.hf_config = None
config.model_config.hf_text_config = None
config.load_config = None