build(dev-dependency): Add torchvision for multimodal test (#755)

## Summary
<!--- This is a required section; please describe the main purpose of
this proposed code change. --->
`Qwen2VLVideoProcessor` requires torchvision. Related to #738
#723 (comment)
<!---
## Details
This is an optional section; is there anything specific that reviewers
should be aware of?
--->

## Testing Done
<!--- This is a required section; please describe how this change was
tested. --->

<!-- 
Replace BLANK with your device type. For example, A100-80G-PCIe

Complete the following tasks before sending your PR, and replace `[ ]`
with
`[x]` to indicate you have done them. 
-->

- Hardware Type: <BLANK>
- [ ] run `make test` to ensure correctness
- [ ] run `make checkstyle` to ensure code style
- [ ] run `make test-convergence` to ensure convergence

Author: Tcc0403

setup.py

@@ -43,6 +43,7 @@ def get_optional_dependencies():
             "seaborn",
             "mkdocs",
             "mkdocs-material",
+            "torchvision>=0.20",
         ]
     }
 

test/convergence/bf16/test_mini_models_multimodal.py

@@ -20,6 +20,7 @@
 from test.utils import UNTOKENIZED_DATASET_PATH
 from test.utils import MiniModelConfig
 from test.utils import assert_verbose_allclose
+from test.utils import is_torchvision_available
 from test.utils import load_image_processing_config
 from test.utils import load_processor_config
 from test.utils import load_tokenizer_config
@@ -788,6 +789,7 @@ def run_mini_model_multimodal(
                     not QWEN2_VL_AVAILABLE,
                     reason="Qwen2-VL not available in this version of transformers",
                 ),
+                pytest.mark.skipif(not is_torchvision_available(), reason="Qwen2VLVideoProcessor requires torchvision"),
             ],
         ),
         pytest.param(
@@ -826,6 +828,7 @@ def run_mini_model_multimodal(
                     not QWEN2_5_VL_AVAILABLE,
                     reason="Qwen2.5-VL not available in this version of transformers",
                 ),
+                pytest.mark.skipif(not is_torchvision_available(), reason="Qwen2VLVideoProcessor requires torchvision"),
             ],
         ),
         pytest.param(

test/convergence/fp32/test_mini_models_multimodal.py

@@ -20,6 +20,7 @@
 from test.utils import UNTOKENIZED_DATASET_PATH
 from test.utils import MiniModelConfig
 from test.utils import assert_verbose_allclose
+from test.utils import is_torchvision_available
 from test.utils import load_image_processing_config
 from test.utils import load_processor_config
 from test.utils import load_tokenizer_config
@@ -783,6 +784,7 @@ def run_mini_model_multimodal(
                     not QWEN2_VL_AVAILABLE,
                     reason="Qwen2-VL not available in this version of transformers",
                 ),
+                pytest.mark.skipif(not is_torchvision_available(), reason="Qwen2VLVideoProcessor requires torchvision"),
             ],
         ),
         pytest.param(
@@ -817,6 +819,7 @@ def run_mini_model_multimodal(
                     not QWEN2_5_VL_AVAILABLE,
                     reason="Qwen2.5-VL not available in this version of transformers",
                 ),
+                pytest.mark.skipif(not is_torchvision_available(), reason="Qwen2VLVideoProcessor requires torchvision"),
             ],
         ),
         pytest.param(

test/utils.py

@@ -228,6 +228,13 @@ def supports_bfloat16():
         return False
 
 
+def is_torchvision_available():
+    if importlib.util.find_spec("torchvision") is not None:
+        return True
+    else:
+        return False
+
+
 def revert_liger_kernel_to_granite(model_config: MiniModelConfig):
     """
     Revert all Liger kernel patches applied to Granite.