A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 3, 16:30 UTC
[CI][AMD] Skip tests where models have problems or fails on both HW types (#42126)
Signed-off-by: Randall Smith <Randall.Smith@amd.com>
rasmith committed
May 14, 2026 at 03:21 UTC
addef3299c95173926dfdbdb9469545b205e08dc
1 file changed
+25
-3
tests/models/multimodal/generation/test_common.py
+25
-3
@@ -316,7 +316,15 @@ VLM_TEST_SETTINGS = {
316
stop_str=["<|im_end|>"],
317
image_size_factors=[(0.10, 0.15)],
318
max_tokens=64,
319
- marks=[large_gpu_mark(min_gb=64)],
319
+ marks=[
320
+ pytest.mark.skip(
321
+ reason="Aria needs to update for latest transformers, "
322
+ "must have a vision_processor.py."
323
+ "An issue has been filed:"
324
+ "https://huggingface.co/rhymes-ai/Aria/discussions/23"
325
+ ),
326
+ large_gpu_mark(min_gb=64),
327
+ ],
328
),
329
"aya_vision": VLMTestInfo(
330
models=["CohereLabs/aya-vision-8b"],
@@ -468,7 +476,14 @@ VLM_TEST_SETTINGS = {
476
max_tokens=8,
477
num_logprobs=10,
478
auto_cls=AutoModelForCausalLM,
471
- marks=[large_gpu_mark(min_gb=32)],
479
+ marks=[
480
+ pytest.mark.skip(
481
+ reason="The code for this model has a bug."
482
+ "Please see the issue here:"
483
+ "https://huggingface.co/zai-org/glm-4v-9b/discussions/46."
484
+ ),
485
+ large_gpu_mark(min_gb=32),
486
+ ],
487
),
488
"glm4_1v": VLMTestInfo(
489
models=["zai-org/GLM-4.1V-9B-Thinking"],
@@ -513,7 +528,14 @@ VLM_TEST_SETTINGS = {
528
num_logprobs=10,
529
image_size_factors=[(0.25,), (0.25, 0.25, 0.25), (0.25, 0.2, 0.15)],
530
auto_cls=AutoModelForImageTextToText,
516
- marks=[large_gpu_mark(min_gb=32)],
531
+ marks=[
532
+ pytest.mark.skip(
533
+ reason="This test fails on both AMD and NV"
534
+ "hardware. please see the issue:"
535
+ "https://github.com/vllm-project/vllm/issues/42016"
536
+ ),
537
+ large_gpu_mark(min_gb=32),
538
+ ],
539
),
540
"granite4_vision": VLMTestInfo(
541
models=["ibm-granite/granite-vision-4.1-4b"],