Mastering Computer Vision with TensorFlow 2.x

Creating a notebook instance and beginning training

Let's follow these steps:

1. Select the instance type (select an instance for accelerated computing, such as ml.p2.nxlarge, where n can be 1, 2, 8, and so on). Note that if the instance type is standard, such as ml.m5.nxlarge, or compute-optimized, such as ml.c5.nxlarge, then the training will fail. So, request a service limit increase, as we described previously.
2. Select a maximum run time—start with 1 hour and increase it for very large jobs.
3. Assign a path to the S3 bucket for each of the four channels described previously so that the algorithm knows where to pull data from.
4. Assign the path to the output folder mentioned previously. An example of an output path is shown in the preceding code ...