Fix pickle issue for CLIP training (NVIDIA#8722)

* fix local pickle issue

Signed-off-by: yaoyu-33 <yaoyu.094@gmail.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: yaoyu-33 <yaoyu.094@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

nemo/collections/multimodal/data/clip/clip_dataset.py

@@ -85,23 +85,25 @@ def get_preprocess_fns(model_cfg, tokenizer=None, is_train=True):
     return img_transform, text_transform
 
 
+# This function maps data that are tuples to dictionary.
+def tuple_to_dict(inp):
+    for input in inp:
+        out_dict = dict()
+        out_dict['images'] = input[0]
+        out_dict['captions'] = input[1]
+        yield out_dict
+
+
+def transform_fn(sample, img_transform, text_transform):
+    image, text = sample["jpg"], sample["txt"]
+    return img_transform(image), text_transform(text)
+
+
 def build_train_valid_datasets(
     model_cfg, consumed_samples, tokenizer=None,
 ):
     data_cfg = model_cfg.data
 
-    # This function maps data that are tuples to dictionary.
-    def tuple_to_dict(inp):
-        for input in inp:
-            out_dict = dict()
-            out_dict['images'] = input[0]
-            out_dict['captions'] = input[1]
-            yield out_dict
-
-    def transform_fn(sample, img_transform, text_transform):
-        image, text = sample["jpg"], sample["txt"]
-        return img_transform(image), text_transform(text)
-
     train_img_transform, text_transform = get_preprocess_fns(model_cfg, tokenizer, is_train=True)
     train_data = WebDatasetCommon(
         dataset_cfg=data_cfg,