Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -4,19 +4,29 @@ import bitsandbytes
|
|
4 |
import accelerate
|
5 |
import scipy
|
6 |
from PIL import Image
|
7 |
-
|
8 |
from transformers import Blip2Processor, Blip2ForConditionalGeneration, InstructBlipProcessor, InstructBlipForConditionalGeneration
|
9 |
|
10 |
|
11 |
def load_caption_model(blip2=False, instructblip=True):
|
12 |
|
|
|
|
|
|
|
|
|
13 |
if blip2:
|
14 |
processor = Blip2Processor.from_pretrained("Salesforce/blip2-opt-2.7b", load_in_8bit=True,torch_dtype=torch.float16)
|
15 |
model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-2.7b", load_in_8bit=True,torch_dtype=torch.float16)
|
|
|
|
|
|
|
16 |
#model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-2.7b", torch_dtype=torch.float16, device_map="auto")
|
17 |
|
18 |
if instructblip:
|
19 |
model = InstructBlipForConditionalGeneration.from_pretrained("Salesforce/instructblip-vicuna-7b", load_in_8bit=True,torch_dtype=torch.float16)
|
|
|
|
|
|
|
20 |
processor = InstructBlipProcessor.from_pretrained("Salesforce/instructblip-vicuna-7b", load_in_8bit=True,torch_dtype=torch.float16)
|
21 |
|
22 |
return model, processor
|
|
|
4 |
import accelerate
|
5 |
import scipy
|
6 |
from PIL import Image
|
7 |
+
import torch.nn as nn
|
8 |
from transformers import Blip2Processor, Blip2ForConditionalGeneration, InstructBlipProcessor, InstructBlipForConditionalGeneration
|
9 |
|
10 |
|
11 |
def load_caption_model(blip2=False, instructblip=True):
|
12 |
|
13 |
+
|
14 |
+
model = YourModel()
|
15 |
+
|
16 |
+
|
17 |
if blip2:
|
18 |
processor = Blip2Processor.from_pretrained("Salesforce/blip2-opt-2.7b", load_in_8bit=True,torch_dtype=torch.float16)
|
19 |
model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-2.7b", load_in_8bit=True,torch_dtype=torch.float16)
|
20 |
+
if torch.cuda.device_count() > 1:
|
21 |
+
model = nn.DataParallel(model)
|
22 |
+
model.to('cuda')
|
23 |
#model = Blip2ForConditionalGeneration.from_pretrained("Salesforce/blip2-opt-2.7b", torch_dtype=torch.float16, device_map="auto")
|
24 |
|
25 |
if instructblip:
|
26 |
model = InstructBlipForConditionalGeneration.from_pretrained("Salesforce/instructblip-vicuna-7b", load_in_8bit=True,torch_dtype=torch.float16)
|
27 |
+
if torch.cuda.device_count() > 1:
|
28 |
+
model = nn.DataParallel(model)
|
29 |
+
model.to('cuda')
|
30 |
processor = InstructBlipProcessor.from_pretrained("Salesforce/instructblip-vicuna-7b", load_in_8bit=True,torch_dtype=torch.float16)
|
31 |
|
32 |
return model, processor
|