In Which the neural network gets bored halfway through making cookies
I’ve been training a neural network (based on this open-source neural network framework from Andrej Karpathy) to generate cookbook recipes. Although its early attempts are barely intelligible and sometimes even dangerous, it eventually gets to the point where it can figure out reasonable ingredient combinations and amounts.
Except when, apparently, it gets bored.
See if you can guess the exact moment at which this happened in the recipe below.
Chocolate Baked And Serves
1 cup butter
2 cup peanut butter
1 cup sugar
1 teaspoon vanilla extract
1 teaspoon baking powder
1 cup white cocoa
1 cup milk
1 cup horseradish or sour cream
Mix all ingredients. Spread over grease and make a gently pan mixture
with 1 several hours, turning and boil on high until the mixture is completely golden.
Transfer the short that opan and golden brown. Release the chocolate
accompaniments and cool the prepared pastry tuna. Add the shrimp to the sugar brownie cubes, oil, salt and butter in a small bowl. Combine the squid ingredients. Bring to a boil over low heat to 375 deg F. With the liver), slice them to kitchen pire and add chicken broth.
Someone made this one! They added the entire cup of horseradish, and even claimed it’s good. They also claim to be a 100% real human.
THIS IS THE WORST CHOCOLATE THING I HAVE EVER TASTED