Torch

What do you like best?

Torch, has many packages in machine learning, signal processing, audio, video and parallel processing. It also has neural networks and deep learning algorithms. Basically. It allows people to conduct research at a rapid pace, develop ideas more quickly without being bogged down by the details of implementation. At the same time, it allows us to deepen and understand the fundamentals, making the learning cycle faster and more effective.

What do you dislike?

A negative point of TORCH is that it needs the LuaJIT environment when it is running and is slowing the road to large-scale production. Otherwise, this product is as good as the tensorflow. If someone is willing to learn a new language, this product is highly recommended.

Recommendations to others considering the product:

Basically, Torch comes with the integration of the LuaJIT programming language with the underlying implementation of C. It has better computational features where managing dependencies is easier than python. The most important reason why I like this framework is that it does not have compile time, which means that its code is interpreted and debugging is much easier.

What problems are you solving with the product? What benefits have you realized?

Torch helped me accelerate the result of the investigation, I was able to develop neural networks faster and better understand their internal functioning. I use Torch to develop solutions to the problem of classifying complete slide images of idiopathic inflammatory myopathies. Using Torch has allowed me to test solutions quickly.Torch allows you to browse the Internet in addition to other multiple tasks such as displaying websites, sharing websites through social networks, downloading torrents, accelerating downloads and recording multimedia files, all directly from the browser.