minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[NPU]Add a full test pipeline on NPU, resolve issues in the NPU test architecture (#20751)

Browse Source
This commit is contained in:
Cherry_ming
2026-04-01 19:56:31 +08:00
committed by GitHub
parent ac039bd04e
commit e67b95d66b
43 changed files with 673 additions and 106 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/pyproject_npu.toml
+2
View File
@@ -26,6 +26,8 @@ dependencies = [
"einops",
"fastapi",
"gguf",
"hf_transfer",
"huggingface_hub",
"interegular",
"llguidance>=0.7.11,<0.8.0",
"modelscope",
python/sglang/test/ascend/test_ascend_utils.py
+9 -9
View File
@@ -117,9 +117,18 @@ QWEN3_1_7B_GPTQ_INT8_WEIGHTS_PATH = os.path.join(
QWEN3_235B_A22B_W8A8_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "vllm-ascend/Qwen3-235B-A22B-W8A8"
)
QWEN3_30B_A3B_GPTQ_2507_INT4_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "Qwen/Qwen3-30B-A3B-GPTQ-Int4"
)
QWEN3_30B_A3B_INSTRUCT_2507_INT4_AUTOROUND_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "Intel/Qwen3-30B-A3B-Instruct-2507-int4-AutoRound"
)
QWEN3_30B_A3B_INSTRUCT_2507_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "Qwen/Qwen3-30B-A3B-Instruct-2507"
)
QWEN3_8B_INT4_AUTOROUND_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "Intel/Qwen3-8B-int4-AutoRound"
)
QWEN3_8B_WEIGHTS_PATH = os.path.join(MODEL_WEIGHTS_DIR, "Qwen/Qwen3-8B")
QWEN3_8B_EAGLE3_WEIGHTS_PATH = os.path.join(MODEL_WEIGHTS_DIR, "Qwen/Qwen3-8B_eagle3")
QWEN3_32B_WEIGHTS_PATH = os.path.join(MODEL_WEIGHTS_DIR, "Qwen/Qwen3-32B")
@@ -133,15 +142,6 @@ QWEN3_32B_EAGLE3_WEIGHTS_PATH = os.path.join(MODEL_WEIGHTS_DIR, "Qwen/Qwen3-32B-
QWEN3_32B_W8A8_MINDIE_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "aleoyang/Qwen3-32B-w8a8-MindIE"
)
QWEN3_235B_A22B_W8A8_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "vllm-ascend/Qwen3-235B-A22B-W8A8"
)
QWEN3_CODER_480B_A35B_INSTRUCT_W8A8_QUAROT_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "Qwen3-Coder-480B-A35B-Instruct-w8a8-QuaRot"
)
QWEN3_NEXT_80B_A3B_INSTRUCT_WEIGHTS_PATH = os.path.join(
MODEL_WEIGHTS_DIR, "Qwen/Qwen3-Next-80B-A3B-Instruct"
)
QWQ_32B_W8A8_WEIGHTS_PATH = os.path.join(MODEL_WEIGHTS_DIR, "vllm-ascend/QWQ-32B-W8A8")
SMOLLM_1_7B_WEIGHTS_PATH = os.path.join(MODEL_WEIGHTS_DIR, "HuggingFaceTB/SmolLM-1.7B")
STABLELM_2_1_6B_WEIGHTS_PATH = os.path.join(