[diffusion] amd: fix SGLANG_DIFFUSION_ATTENTION_BACKEND env var for diffusion attention backend selection (#16325)

Co-authored-by: root <root@mi300x8-008.atl1.do.cpe.ice.amd.com>
This commit is contained in:
sunxxuns
2026-01-09 20:16:58 +08:00
committed by GitHub
co-authored by root
parent 2babf88f24
commit 64a31d4b75
5 changed files with 36 additions and 19 deletions
-2
View File
@@ -409,7 +409,6 @@ jobs:
-e SGLANG_DENOISE_STEP_TOLERANCE=0.6 \
-e SGLANG_DENOISE_AGG_TOLERANCE=0.3 \
-e SGLANG_TEST_NUM_INFERENCE_STEPS=5 \
-e SGLANG_DIFFUSION_ATTENTION_BACKEND=AITER \
-e AITER_JIT_DIR=/sgl-data/aiter-kernels \
-e MIOPEN_USER_DB_PATH=/sgl-data/miopen-cache \
-e HF_HUB_ENABLE_HF_TRANSFER=1 \
@@ -529,7 +528,6 @@ jobs:
-e SGLANG_DENOISE_STEP_TOLERANCE=0.6 \
-e SGLANG_DENOISE_AGG_TOLERANCE=0.3 \
-e SGLANG_TEST_NUM_INFERENCE_STEPS=5 \
-e SGLANG_DIFFUSION_ATTENTION_BACKEND=AITER \
-e AITER_JIT_DIR=/sgl-data/aiter-kernels \
-e MIOPEN_USER_DB_PATH=/sgl-data/miopen-cache \
-e HF_HUB_ENABLE_HF_TRANSFER=1 \