Hugging Face's logo

hieudinhpro
/
git-base-on-diffuision-dataset2

Image-to-Text
Transformers
PyTorch
English
git
image-text-to-text
Generated from Trainer
Inference Endpoints
Model card Files Community
git-base-on-diffuision-dataset2 / README.md
hieudinhpro's picture
hieudinhpro
Update README.md
ef27788
preview code
|
raw
history blame contribute delete
1.91 kB
metadata
license: mit
base_model: microsoft/git-base
tags:
  - generated_from_trainer
model-index:
  - name: git-base-on-diffuision-dataset2
    results: []
language:
  - en
library_name: transformers
pipeline_tag: image-to-text

git-base-on-diffuision-dataset2

This model is a fine-tuned version of microsoft/git-base on hieudinhpro/diffuision-dataset2 dataset.

Model description

GIT (short for GenerativeImage2Text) model, base-sized version.
It was introduced in the paper GIT: A Generative Image-to-text Transformer for Vision and Language

Model train for task : Sketch Scene image to text

How to use mdoel 

# Load model directly
from transformers import AutoProcessor, AutoModelForCausalLM

processor = AutoProcessor.from_pretrained("microsoft/git-base")
model = AutoModelForCausalLM.from_pretrained("hieudinhpro/git-base-on-diffuision-dataset2")

# load image
from PIL import Image

image = Image.open('/content/image_3.jpg')

# pre image
inputs = processor(images=image, return_tensors="pt")
pixel_values = inputs.pixel_values

# predict 
generated_ids = model.generate(pixel_values=pixel_values, max_length=50)

# decode to text
generated_caption = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(generated_caption)

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 2e-05
  • train_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 2
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 1

Framework versions

  • Transformers 4.34.0
  • Pytorch 2.0.1+cu118
  • Datasets 2.14.5
  • Tokenizers 0.14.0