init!
Browse files
app.py
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# test.py
|
2 |
+
import torch
|
3 |
+
from PIL import Image
|
4 |
+
from transformers import AutoModel, AutoTokenizer
|
5 |
+
|
6 |
+
model = AutoModel.from_pretrained('openbmb/MiniCPM-Llama3-V-2_5', trust_remote_code=True, torch_dtype=torch.float16)
|
7 |
+
model = model.to(device='cuda')
|
8 |
+
|
9 |
+
tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-Llama3-V-2_5', trust_remote_code=True)
|
10 |
+
model.eval()
|
11 |
+
|
12 |
+
image = Image.open('xx.jpg').convert('RGB')
|
13 |
+
question = 'What is in the image?'
|
14 |
+
msgs = [{'role': 'user', 'content': question}]
|
15 |
+
|
16 |
+
res = model.chat(
|
17 |
+
image=image,
|
18 |
+
msgs=msgs,
|
19 |
+
tokenizer=tokenizer,
|
20 |
+
sampling=True, # if sampling=False, beam_search will be used by default
|
21 |
+
temperature=0.7,
|
22 |
+
# system_prompt='' # pass system_prompt if needed
|
23 |
+
)
|
24 |
+
print(res)
|
25 |
+
|
26 |
+
## if you want to use streaming, please make sure sampling=True and stream=True
|
27 |
+
## the model.chat will return a generator
|
28 |
+
res = model.chat(
|
29 |
+
image=image,
|
30 |
+
msgs=msgs,
|
31 |
+
tokenizer=tokenizer,
|
32 |
+
sampling=True,
|
33 |
+
temperature=0.7,
|
34 |
+
stream=True
|
35 |
+
)
|
36 |
+
|
37 |
+
generated_text = ""
|
38 |
+
for new_text in res:
|
39 |
+
generated_text += new_text
|
40 |
+
print(new_text, flush=True, end='')
|