I shared the ny times with gemini to help me to learn how to play soduko but it wasnt capable to see the blank gris appropriately. How can I train the model to identify them?
I opened a tab websudoku and shared it with Google AI studio’s stream realtime feature.
I then asked to give me a generic strategy to get started and it told me to look for the rows and columns that are most complete, and so on. I could ask further questions if I wanted.
Hope this helps …