[ROCm][Quark][6/N] Use MXFP4 linear kernel abstraction for aiter backend (#49348)

Signed-off-by: Felix Marty <Felix.Marty@amd.com>
Co-authored-by: Andreas Karatzas <akaratza@amd.com>
This commit is contained in:
fxmarty-amd
2026-07-27 21:35:57 -05:00
committed by GitHub
co-authored by Andreas Karatzas
parent 73af7a362a
commit 7aea73d83d
6 changed files with 370 additions and 166 deletions
@@ -0,0 +1,5 @@
model_name: "amd-quark/Qwen3-1.7B-MXFP4"
accuracy_threshold: 0.27
num_questions: 1319
num_fewshot: 5
server_args: "--enforce-eager --max-model-len 4096"