What is Arpaggers?
Arpaggers is a group of socially idle adults.
This site shares Arpaggers works, thoughts, and parts of the making process over time.
How We Started
Arpaggers began in a room of an apartment building.
After work, a group of coworkers would gather—sometimes studying fluid dynamics and relativity, other times watching anime or sharing a hot pot together.
What started as a casual meetup gradually took on a new shape. “With this group, we might be able to create something interesting.”
With that conviction, we began developing a project management web application, despite having almost no programming experience. This marked the starting point of our creative work.
Around that time, deep learning was beginning to gain attention, and we naturally found ourselves drawn into it. One of our early ideas was to visualize neural networks using Unity—connecting layers and simulating backpropagation in real time. Although we never completed the project, this period of exploration and experimentation became an important part of our journey.
We then moved on to developing an AI for horse race prediction. Before transformers became mainstream, we approached the problem as a sequence transformation task using an LSTM-based seq2seq model—mapping horse numbers to predicted finishing orders. We even brought laptops to the racetrack and ran predictions on-site. While it didn’t make us rich, it gave us a real sense of excitement in seeing our ideas come to life.
Then came an unexpected opportunity: a request to develop a game.
The result was Japanese Style Tien-gow, which we released on Steam—marking a major turning point for Arpaggers.
After work, a group of coworkers would gather—sometimes studying fluid dynamics and relativity, other times watching anime or sharing a hot pot together.
What started as a casual meetup gradually took on a new shape. “With this group, we might be able to create something interesting.”
With that conviction, we began developing a project management web application, despite having almost no programming experience. This marked the starting point of our creative work.
Around that time, deep learning was beginning to gain attention, and we naturally found ourselves drawn into it. One of our early ideas was to visualize neural networks using Unity—connecting layers and simulating backpropagation in real time. Although we never completed the project, this period of exploration and experimentation became an important part of our journey.
We then moved on to developing an AI for horse race prediction. Before transformers became mainstream, we approached the problem as a sequence transformation task using an LSTM-based seq2seq model—mapping horse numbers to predicted finishing orders. We even brought laptops to the racetrack and ran predictions on-site. While it didn’t make us rich, it gave us a real sense of excitement in seeing our ideas come to life.
Then came an unexpected opportunity: a request to develop a game.
The result was Japanese Style Tien-gow, which we released on Steam—marking a major turning point for Arpaggers.
