Follow Datanami:
December 30, 2020

DL Framework ‘SmallTrain 0.2.0’ Released for Professional and Commercial Use

KYOTO, Japan, Dec. 30, 2020 — Geek Guild Co., Ltd. announced the Open-Source Software (OSS) project, “SmallTrain,” which generates user-friendly deep learning models for high accuracy and high functionality as a standalone deep learning library and as a wrapper for TensorFlow and PyTorch. The trained model built with SmallTrain 0.2.0 can be used commercially. The source code is open-source, free to use, and released for commercial use. We have released the source code with the desire to “Deliver Artificial Intelligence (AI) that can withstand commercial and professional use.”

Artificial General Intelligence OSS Goal

During the recent AI boom, companies develop customized “Narrow AI,” which are active only in limited use cases. Here in Geek Guild, we believe AI is fundamental for machines to help humanity build symbiotic spheres as “Harmonic AI,” aiming to integrate independent AI for developing general-purpose AI.

SmallTrain is an OSS framework that enables you to develop transfer learning models pre-trained in various data. Here, you can rapid-develop hi-accuracy deep neural-network without large training-data requirements. “SmallTrain” uses 60+ layers of pyramid neural-network architecture build from state-of-the-art cutting edge algorithms, our “Harmonic AI” which are pre-trained for different use cases and data-science problems, the accuracy doesn’t drop even with minimal available training data as normal “Narrow AI” does.

Background of SmallTrain OSS Project

As AI development from amateur to advance is a steep learning curve, developing an AI model is a tedious task. It requires a lot of in-depth knowledge of data science. Due to this, many data science enthusiasts find difficulty in this field. Here at Geek Guild, we developed SmallTrain with the philosophy that “Instead of you, we do the heavy lifting.” This means “SmallTrain” enables you to quickly and efficiently build a learning model, even with the minimal data-science background, while maintaining quality service operable condition.

Advantages of SmallTrain

  • Rapid development from POC to production with a minimal data-science background.
  • Almost no programming for building your pre-trained model.
  • Available as TensorFlow and PyTorch wrapper.
  • Build using state-of-the-art algorithms from AI research papers.
  • Accuracy will be better, even with minimal data and training time, SmallTrain is pre-trained in Multi data-Science problems.
  • Licensed under MIT Open Source. So, no worries about bug fixes or improvements on your own. Collaboration is the key.

SmallTrain Overview

The AI model that Geek Guild uses for R&D and contract projects is OSS. It’s a deep learning framework developed as a library for convenient use and as a wrapper. Kyoto Economic Gardening Program adopts the project. Please visit our website and GitHub page for more info.

https://github.com/geek-guild/smalltrain
https://www.smalltrain.org/

About Geek Guild Co., Ltd.

Established in June 2018 as a Guild (organization) that gathers Geek (human resources with extensive knowledge of IT). The goal is to create General Artificial-Intelligence based on harmony between humans and machines. https://www.geek-guild.jp/


Source: Geek Guild Co., Ltd.

Datanami