From 01cd2fe05211279deb29846f5599867d958057b0 Mon Sep 17 00:00:00 2001 From: Graeme Malcolm Date: Sun, 14 Mar 2021 11:57:03 -0700 Subject: [PATCH] Fixed typo --- 05b - Convolutional Neural Networks (PyTorch).ipynb | 2 +- 05b - Convolutional Neural Networks (Tensorflow).ipynb | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/05b - Convolutional Neural Networks (PyTorch).ipynb b/05b - Convolutional Neural Networks (PyTorch).ipynb index 6971679..7ecc977 100644 --- a/05b - Convolutional Neural Networks (PyTorch).ipynb +++ b/05b - Convolutional Neural Networks (PyTorch).ipynb @@ -235,7 +235,7 @@ "\n", "In the example below, we use 5 epochs to train the model using the batches of images loaded by the data loaders, holding back the data in the test data loader for validation. After each epoch, a loss function measures the error (*loss*) in the model and adjusts the weights (which were randomly generated for the first iteration) to try to improve accuracy. \n", "\n", - "> **Note**: We're only using 5 epochs to minimze the training time for this simple example. A real-world CNN is usually trained over more epochs than this. CNN model training is processor-intensive, involving a lot of matrix and vector-based operations; so it's recommended to perform this on a system that can leverage GPUs, which are optimized for these kinds of calculation. This will take a while to complete on a CPU-based system - status will be displayed as the training progresses." + "> **Note**: We're only using 5 epochs to minimize the training time for this simple example. A real-world CNN is usually trained over more epochs than this. CNN model training is processor-intensive, involving a lot of matrix and vector-based operations; so it's recommended to perform this on a system that can leverage GPUs, which are optimized for these kinds of calculation. This will take a while to complete on a CPU-based system - status will be displayed as the training progresses." ] }, { diff --git a/05b - Convolutional Neural Networks (Tensorflow).ipynb b/05b - Convolutional Neural Networks (Tensorflow).ipynb index 28ff8e7..c4f41aa 100644 --- a/05b - Convolutional Neural Networks (Tensorflow).ipynb +++ b/05b - Convolutional Neural Networks (Tensorflow).ipynb @@ -193,7 +193,7 @@ "\n", "With the layers of the CNN defined, we're ready to train the model using our image data. In the example below, we use 5 iterations (*epochs*) to train the model in 30-image batches, holding back 30% of the data for validation. After each epoch, the loss function measures the error (*loss*) in the model and adjusts the weights (which were randomly generated for the first iteration) to try to improve accuracy. \n", "\n", - "> **Note**: We're only using 5 epochs to minimze the training time for this simple example. A real-world CNN is usually trained over more epochs than this. CNN model training is processor-intensive, involving a lot of matrix and vector-based operations; so it's recommended to perform this on a system that can leverage GPUs, which are optimized for these kinds of calculation. This will take a while to complete on a CPU-based system - status will be displayed as the training progresses." + "> **Note**: We're only using 5 epochs to minimize the training time for this simple example. A real-world CNN is usually trained over more epochs than this. CNN model training is processor-intensive, involving a lot of matrix and vector-based operations; so it's recommended to perform this on a system that can leverage GPUs, which are optimized for these kinds of calculation. This will take a while to complete on a CPU-based system - status will be displayed as the training progresses." ] }, { @@ -430,4 +430,4 @@ }, "nbformat": 4, "nbformat_minor": 2 -} +} \ No newline at end of file