File size: 1,498 Bytes
551bc08
 
886514c
 
 
 
551bc08
a7be8e9
886514c
a7be8e9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
886514c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
---
license: mit
language:
- en
library_name: transformers
inference: False
---
## Sharded BLIP-2 Model Card - flan-t5-xl

This is a sharded version of the [blip2-flan-t5-xl](https://huggingface.co/models/Salesforce/blip2-flan-t5-xl) which leverages [Flan T5-xl](https://huggingface.co/google/flan-t5-xl) for image-to-text tasks such as image captioning and visual question answering.

Refer to the [original model card](https://huggingface.co/models/Salesforce/blip2-flan-t5-xl) for more details about the model description, intended uses, and limitations, as well as instructions for how to use the model on CPU and GPU in different precisions.

## Usage

Refer to the original model card for details or see [this blog post](https://huggingface.co/blog/blip-2#using-blip-2-with-hugging-face-transformers). Here is how you can use it on CPU:


```python
import requests
from PIL import Image
from transformers import BlipProcessor, Blip2ForConditionalGeneration

model_name = "Salesforce/blip2-flan-t5-xl")
processor = BlipProcessor.from_pretrained(model_name)
model = Blip2ForConditionalGeneration.from_pretrained(model_name)

img_url = 'https://storage.googleapis.com/sfr-vision-language-research/BLIP/demo.jpg' 
raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')

question = "how many dogs are in the picture?"
inputs = processor(raw_image, question, return_tensors="pt")

out = model.generate(**inputs)
print(processor.decode(out[0], skip_special_tokens=True))
```