Merge pull request ‘[Feat][Platform] 适配 –attention-config / –ray-workers-use-nsight 到 Ascend NPU’ (#1) from ccf-attention-config-nsight into master
This submission adapts --attention-config and --ray-workers-use-nsight for Ascend NPU.
--attention-config
--ray-workers-use-nsight
Source-level delivery is included in:
attention-config-ray-workers-use-nsight/
That directory contains changed vllm_ascend source files, focused unit tests, user documentation, and the parameter-specific SKILL.md.
vllm_ascend
SKILL.md
The same implementation is also provided as a unified patch under patches/, with Ascend 910B2C validation records in verification/.
patches/
verification/
Validation summary:
attention_config or nsight focused tests: 5 passed full tests/ut/test_platform.py: 57 passed, 1 skipped
本仓库用于CCF开源创新大赛
版权所有:中国计算机学会技术支持:开源发展技术委员会 京ICP备13000930号-9 京公网安备 11010802047560号
vLLM Ascend Parameter Adaptation
This submission adapts
--attention-configand--ray-workers-use-nsightfor Ascend NPU.Source-level delivery is included in:
That directory contains changed
vllm_ascendsource files, focused unit tests, user documentation, and the parameter-specificSKILL.md.The same implementation is also provided as a unified patch under
patches/, with Ascend 910B2C validation records inverification/.Validation summary: