frozenleaves/vllm
1
0
Fork 0
mirror of https://github.com/vllm-project/vllm.git synced 2025-10-20 14:53:52 +08:00
Code Issues Packages Projects Releases Wiki Activity

[CI] Add Qwen3 MoE NVFP4 to Blackwell lm-eval (#26316)

Browse Source 
Signed-off-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Michael Goin
2025-10-07 13:36:15 -04:00
committed by GitHub
parent a38c1bfe09
commit 30a3e5af69
3 changed files with 8 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.buildkite/test-pipeline.yaml
View File

@ -868,7 +868,7 @@ steps:
- pytest -s -v tests/quantization/test_blackwell_moe.py
- label: Blackwell LM Eval Small Models
timeout_in_minutes: 75
timeout_in_minutes: 120
gpu: b200
optional: true # run on nightlies
source_file_dependencies:

6
tests/evals/gsm8k/configs/Qwen3-30B-A3B-NVFP4.yaml Normal file
View File

@ -0,0 +1,6 @@
model_name: "nvidia/Qwen3-30B-A3B-FP4"
accuracy_threshold: 0.89
num_questions: 1319
num_fewshot: 5
max_model_len: 4096

1
tests/evals/gsm8k/configs/models-blackwell.txt
View File

@ -2,3 +2,4 @@ Qwen3-0.6B-FP8.yaml
Qwen2.5-VL-3B-Instruct-FP8-dynamic.yaml
Qwen1.5-MoE-W4A16-CT.yaml
DeepSeek-V2-Lite-Instruct-FP8.yaml
Qwen3-30B-A3B-NVFP4.yaml