1. 6
  1.  

  2. 3

    I like it with high fuzziness.

    Cheese Dough, Store bought (1 cuplunger or refrigerated coconut )/50 percent norm leg; if vasily, in quesadillas. “Oulla”R’b”-ling”: ▪︎ In cutting board, puree wine in 4 ounces. Mixture should hiftened; arrange over melted, up shots. If 2 trouvers refrigerate, as vesom. Use a spider, spoon onto gratins. Place asparagus in bread loaf for later.

    1. 3

      Haha, yeah, huge fuzziness makes that RNN a little “drunk”

    2. 3

      Salmon Mousse of Beef and Stilton Salad with Jalapenos

      Thank you, I’ll pass.

      But I did enjoy the ride!

      1. 2

        Haha, nice! I had “Cream Soda with Onions” and “Zucchini flavour Tea”. I would also pass

      2. 2

        Funny application! Keep going!

        Small improvement, you could sum up the ingredients present in the list. For example I had 4 times “1/4 tbps ground allspice”, which could be factored to a full one :)

        1. 1

          Yeah, that’s a good idea and it might improve the output. For now I just wanted to see how good/bad RNN performance is. I was thinking maybe by adding another LSTM layer the network would learn more complex things and will some up ingredients for me. But I didn’t check this assumption yet (the training is taking too long :).