lucasjin/Florence-2-Huge

This is a model trained based on Florence-2-large, but expanding the params to bigger. Let's say it's Florence-2-Huge.

For huge model, the currently model should outputs same results as Florence-2-large, but the configuration and model sizes are bigger than Florence-2-large.

For comparasion:

• Florence-2-lage: 0.77B;
• Florence-2-Huge: 2B;

it 2x bigger, the model is still under training, when it done, it should have more good abilities on visual caption tasks.