3 ms·
The transformer (Gemini) predicts text with image and text in the context window. That's it. OCR, Object detection etc all come from the transformer predicting
by famouswaffles 3y ago
The transformer (Gemini) predicts text with image and text in the context window. That's it.
OCR, Object detection etc all come from the transformer predicting text. Read the Flamingo paper.