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 16, 05:45 UTC
[CI/Build] Fix broken multimodal test (#10950)
Cyrus Leung committed
Dec 6, 2024 at 18:41 UTC
222f5b082a62d0b2675cb461e223ae43368eea92
1 file changed
+4
tests/models/embedding/vision_language/test_llava_next.py
+4
@@ -2,6 +2,7 @@ from typing import List, Type
2
3
import pytest
4
import torch.nn.functional as F
5
+import transformers
6
from transformers import AutoModelForVision2Seq
7
8
from ....conftest import IMAGE_ASSETS, HfRunner, PromptImageInput, VllmRunner
@@ -85,6 +86,9 @@ def _run_test(
86
)
87
88
89
+@pytest.mark.skipif(transformers.__version__.startswith("4.46"),
90
+ reason="Model broken with changes in transformers 4.46")
91
+@pytest.mark.core_model
92
@pytest.mark.parametrize("model", MODELS)
93
@pytest.mark.parametrize("dtype", ["half"])
94
def test_models_text(