---
library_name: peft
base_model: ybelkada/blip2-opt-2.7b-fp16-sharded
license: apache-2.0
pipeline_tag: image-to-text
---

# Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->
Lora for Blip2 to generate QAs from a picture.



## Infertece Demo

```python
from datasets import load_dataset 
from peft import PeftModel
import torch
from transformers import AutoProcessor, Blip2ForConditionalGeneration

# prepare the model
processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-2.7b")
model = Blip2ForConditionalGeneration.from_pretrained("ybelkada/blip2-opt-2.7b-fp16-sharded", device_map="auto", load_in_8bit=True)
model = PeftModel.from_pretrained(model, "curlyfu/blip2-OCR-QA-generation")

# prepare inputs
dataset = load_dataset("howard-hou/OCR-VQA", split="test")
example = dataset[10]
image = example["image"]

inputs = processor(images=image, return_tensors="pt").to("cuda", torch.float16)
pixel_values = inputs.pixel_values

generated_ids = model.generate(pixel_values=pixel_values, max_length=100)
generated_caption = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(generated_caption)
```

## Thanks
[huggingface/notebooks](!https://github.com/huggingface/notebooks)