Source: Deep Learning on Medium
The deep learning notebook by Lambda Labs
I had been working on Deep Learning (DL) projects for a year, but it had all been from a work laptop. I did not own a personal laptop equipped for DL projects. So, I searched for one to buy online!
Using GPU from a cloud computing service such as Azure, AWS, or Google Cloud Platform, can get really expensive really fast! One should buy a computer or laptop with a GPU, at least while experimenting with DL.
When buying a laptop for DL, it should have an NVIDIA GPU (which is supported by most DL frameworks) and lots of memory. Gamer laptops are very popular for DL, but they usually come with the Windows OS by default. Ubuntu OS is usually preferred for DL, so most people who buy a Windows gaming laptop for DL will wipe out the Windows OS and install Ubuntu. This can cause compatibility issues with the hardware.
Setting up a laptop for DL work can be tedious. Installing the correct CUDA and cuDNN versions along with the compatible python package versions for TensorFlow, Keras, and other frameworks can be a huge hassle and may take you days or weeks to complete along with a lot of googling and hair loss.
I wanted to hit the ground running once I received my new laptop. I did not want to waste time setting it up. I also am not very savvy when it comes to uninstalling and installing operating systems. I did not want the frustration of setting up a laptop for DL to discourage me from working on DL outside of work.
I decided to buy a TensorBook because it comes with Ubuntu installed along with the following popular DL frameworks: TensorFlow, PyTorch, Keras, Caffe, Caffe2, and Theano.
The TensorBook is sold by Lambda Labs, a United States company located in California. Lambda Labs provided good customer service to me before and after I received the laptop; they were very attentive.
If the price is not a problem for you and you want the DL basics to already be setup, consider getting a TensorBook! The laptop’s been good so far. I quickly ran a Tensorflow project using the GPU and it was straightforward, no issues with dependencies or hardware setup.
The TensorBook laptop can be purchased here.