---
datasets:
- MMInstruction/M3IT
pipeline_tag: image-to-text
---

This model is fintuned on instruction dataset using `SalesForce/blip-imagecaptioning-base` model.
## Usage:
```
from transformers import BlipProcessor, BlipForConditionalGeneration
import torch
from PIL import Image

processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
if processor.tokenizer.eos_token is None:
    processor.tokenizer.eos_token = '<|eos|>'
model = BlipForConditionalGeneration.from_pretrained("prasanna2003/Instruct-blip-v2")

image = Image.open('file_name.jpg').convert('RGB')

prompt = """Instruction: Answer the following input according to the image.
Input: Describe this image.
output: """

inputs = processor(image, prompt, return_tensors="pt")

output = model.generate(**inputs, max_length=100)
print(tokenizer.decode(output[0]))
```