Upload 2 files
Browse filesUpload a handler.py and requirements.txt to run on inference endpoints
- handler.py +30 -0
- requirements.txt +3 -0
handler.py
ADDED
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from io import BytesIO
|
2 |
+
import base64
|
3 |
+
|
4 |
+
from PIL import Image
|
5 |
+
import torch
|
6 |
+
from transformers import CLIPProcessor, CLIPTextModel, CLIPVisionModelWithProjection
|
7 |
+
|
8 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
9 |
+
|
10 |
+
class EndpointHandler():
|
11 |
+
def __init__(self, path=""):
|
12 |
+
self.text_model = CLIPTextModel.from_pretrained("rbanfield/clip-vit-large-patch14")
|
13 |
+
self.image_model = CLIPVisionModelWithProjection.from_pretrained("rbanfield/clip-vit-large-patch14")
|
14 |
+
self.processor = CLIPProcessor.from_pretrained("rbanfield/clip-vit-large-patch14")
|
15 |
+
|
16 |
+
def __call__(self, data):
|
17 |
+
text_input = data.pop("text", None)
|
18 |
+
image_input = data.pop("image", None)
|
19 |
+
|
20 |
+
if text_input:
|
21 |
+
processor = self.processor(text=text_input, return_tensors="pt", padding=True)
|
22 |
+
with torch.no_grad():
|
23 |
+
return self.text_model(**processor).pooler_output.tolist()
|
24 |
+
elif image_input:
|
25 |
+
image = Image.open(BytesIO(base64.b64decode(image_input)))
|
26 |
+
processor = self.processor(images=image, return_tensors="pt")
|
27 |
+
with torch.no_grad():
|
28 |
+
return self.image_model(**processor).image_embeds.tolist()
|
29 |
+
else:
|
30 |
+
return None
|
requirements.txt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
Pillow
|
2 |
+
transformers
|
3 |
+
torch
|