readme.md

# Italian CLIP

....

# Novel Contributions
The original CLIP model was trained on 400millions text-image pairs; this amount of data is not available for Italian and the only datasets for captioning in the literature are MSCOCO-IT (translated version of MSCOCO) and WIT. To get competitive results we follewed three directions: 1) more data 2) better augmentation and 3) better training.

## More Data

## Better Augmentations

## Better Training

different optimizer and backbone freezing

# Scientific Validity
To better understand how well our clip-italian model works we run an experimental evaluation. Since this is the first clip-based model in Italian, we used the multilingual CLIP model as a comparison baseline. 

We selected two different tasks: 
+ image-retrieval 
+ zero-shot classification
            
## Image Retrieval

## Zero-shot classification

# Broader Outlook