With the summer melting away more and more brain cells I decided to employ some artificial help in creating beautiful cocktail recipes.
This project uses torch-rnn on data from The Cocktail DB to create new drinks!
The example given below assumes Ubuntu 18.04 and CUDA support. For everything else check out the torch-rnn repository.
- Install nvidia-docker
- Start the docker container
nvidia-docker run --rm -ti https://hub.docker.com/r/xoryouyou/torch-rnn/ bash - Copy the training data to data/cocktails.txt
- Preprocess the sample data
python scripts/preprocess.py \ --input_txt data/cocktails.txt \ --output_h5 data/cocktails.h5 \ --output_json data/cocktails.json - Train
th train.lua \ -input_h5 data/cocktails.h5 \ -input_json data/cocktails.json - Sample
th sample.lua -checkpoint cv/checkpoint_9000.t7 -length 2000
Check the examples.txt file for more cocktails!
Angeltula Mint
1 shot Water
2 cups Milk
3 oz Chocolate liqueur
3/4 oz chocolate CoWly High
1 oz Sweet Vermouth
1 cup Carapai
1/4 cup the Rum
125 oz L Lime juice
1/2 tsp Light cream
1 tsp Ice
Shake all ingredients with ice, strain into a chilled cocktail glass.
Serve in Cocktail glass
Bora Hitron
3/4 oz Absolut Vodka
1 oz frozen Orange juice
1 Old-fashioned
Pour in old-fashioned glass, add day kee de Coffeod, stir again to fill up with beer.
Serve in Highball glass
Brandy Tea
1 pint Cocoa powder
1 shle de Carbonated water
Burut it inatirich schnalm.
Whip peer.
Serve in Coffee mug
Mother's Milk
1 oz Baileys irish cream
15 ml Five Sugar
5- minnea
Oif your barter difst every toblos, this bitters, shake into a cocktail glass, and serve.
Serve in Cocktail glass
Feel free to create bots, apps, websites, spaceships, music, drawings, ... just don't use it for evil!