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

Fix typo of VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE (#1112)

Browse Source 
### What this PR does / why we need it?
Fix typo of VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE

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

No

### How was this patch tested?

CI passed

Signed-off-by: linfeng-yuan <1102311262@qq.com>
This commit is contained in:
linfeng-yuan
2025-06-07 19:45:33 +08:00
committed by GitHub
parent 20dedba5d1
commit b80a484864
4 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/multicard/test_offline_inference_distributed.py
View File

@@ -61,7 +61,7 @@ def test_models_distributed_DeepSeek():
vllm_model.generate_greedy(example_prompts, max_tokens) vllm_model.generate_greedy(example_prompts, max_tokens)
@patch.dict(os.environ, {"VLLM_ASCEND_ENABLE_TOPK_OPTIMZE": "1"}) @patch.dict(os.environ, {"VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE": "1"})
def test_models_distributed_topk() -> None: def test_models_distributed_topk() -> None:
example_prompts = [ example_prompts = [
"vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs.", "vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs.",

2
tests/singlecard/test_offline_inference.py
View File

@@ -83,7 +83,7 @@ def test_multimodal(model, prompt_template, vllm_runner):
max_tokens=64) max_tokens=64)
@patch.dict(os.environ, {"VLLM_ASCEND_ENABLE_TOPK_OPTIMZE": "1"}) @patch.dict(os.environ, {"VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE": "1"})
def test_models_topk() -> None: def test_models_topk() -> None:
example_prompts = [ example_prompts = [
"Hello, my name is", "Hello, my name is",

4
vllm_ascend/envs.py
View File

@@ -57,8 +57,8 @@ env_variables: Dict[str, Callable[[], Any]] = {
lambda: bool(int(os.getenv("VLLM_ENABLE_MC2", '0'))), lambda: bool(int(os.getenv("VLLM_ENABLE_MC2", '0'))),
# Whether to enable the topk optimization. It's disabled by default for experimental support # Whether to enable the topk optimization. It's disabled by default for experimental support
# We'll make it enabled by default in the future. # We'll make it enabled by default in the future.
"VLLM_ASCEND_ENABLE_TOPK_OPTIMZE": "VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE":
lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_TOPK_OPTIMZE", '0'))), lambda: bool(int(os.getenv("VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE", '0'))),
# Whether to use LCCL communication. If not set, the default value is False. # Whether to use LCCL communication. If not set, the default value is False.
"USING_LCCL_COM": "USING_LCCL_COM":
lambda: bool(int(os.getenv("USING_LCCL_COM", '0'))), lambda: bool(int(os.getenv("USING_LCCL_COM", '0'))),

2
vllm_ascend/patch/worker/patch_common/patch_sampler.py
View File

@@ -97,5 +97,5 @@ def topk_topp_forward_native(
Sampler.apply_min_p = apply_min_p Sampler.apply_min_p = apply_min_p
if envs.VLLM_ASCEND_ENABLE_TOPK_OPTIMZE: if envs.VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE:
TopKTopPSampler.forward_native = topk_topp_forward_native TopKTopPSampler.forward_native = topk_topp_forward_native