diff --git a/src/clip/__init__.py b/src/clip/__init__.py
index c6d7dd09f..1d4154a7b 100644
--- a/src/clip/__init__.py
+++ b/src/clip/__init__.py
@@ -1,7 +1,7 @@
 from .factory import create_model_and_transforms
 from .loss import ClipLoss
 from .model import CLIP, CLIPTextCfg, CLIPVisionCfg, convert_weights_to_fp16
-from .openai_clip import load_openai
+from .openai import load_openai
 from .pretrained import list_pretrained, list_pretrained_tag_models, list_pretrained_model_tags,\
     get_pretrained_url, download_pretrained
 from .tokenizer import SimpleTokenizer, tokenize
diff --git a/src/clip/factory.py b/src/clip/factory.py
index 677f477b4..c834329e1 100644
--- a/src/clip/factory.py
+++ b/src/clip/factory.py
@@ -5,8 +5,8 @@
 
 import torch
 
-from .openai_clip import load_openai
 from .model import CLIP, convert_weights_to_fp16
+from .openai import load_openai
 from .pretrained import get_pretrained_url, download_pretrained
 from .transform import image_transform
 
diff --git a/src/clip/openai_clip.py b/src/clip/openai.py
similarity index 96%
rename from src/clip/openai_clip.py
rename to src/clip/openai.py
index 1ebceb5ce..9e2c2e00e 100644
--- a/src/clip/openai_clip.py
+++ b/src/clip/openai.py
@@ -68,8 +68,8 @@ def load_openai(
         if str(device) == "cpu":
             model.float()
         return model, \
-               image_transform(model.visual.image_size, is_train=True), \
-               image_transform(model.visual.image_size, is_train=False)
+            image_transform(model.visual.image_size, is_train=True), \
+            image_transform(model.visual.image_size, is_train=False)
 
     # patch the device names
     device_holder = torch.jit.trace(lambda: torch.ones([]).to(torch.device(device)), example_inputs=[])