It's a Laravel app with a docker-compose.yml file in the project root, which looks a bit like this (documentation, unrelated services, and unrelated configuration settings removed for readability): version: '3'

'Running in Docker' is not very specific, so I'll start by explaining my project setup. I spent a couple of hours getting everything to work nicely, and encountered a weird (but in hindsight, sensible) issue.

Most of the online guides I've found while hoping for a quick copy and paste configuration didn't end up working, and were in fact aimed at Xdebug 2 - whereas the new Xdebug 3 version changed some of the configuration setting keys. I was also recently starting work on a new Laravel project and decided to setup Xdebug so I could, well, debug. I hoped this would've happened more often, but oh well.Īfter many years of Sublime Text usage, I recently switched to VS Code, as the Sublime Text ecosystem for PHP development seems to be somewhat less active lately, based on my personal experience. # training=False is only needed if there are layers with differentSo here's my second annual (as in once-per-year) blog post. Train_accuracy(labels, test_step(images, labels): Optimizer.apply_gradients(zip(gradients, ainable_variables)) Gradients = adient(loss, ainable_variables) Predictions = model(images, training=True) # behavior during training versus inference (e.g. # training=True is only needed if there are layers with different Test_accuracy = train_step(images, labels): Train_accuracy = tf.(name='train_accuracy') Test_ds = tf._tensor_slices((x_test, y_test)).batch(32) (x_train, y_train), (x_test, y_test) = mnist.load_data() Print('TensorFlow version:', tf._version_)įrom import Dense, Flatten, Conv2D I was able to run this example code from a P圜harm project on my M1 Max MBP. Did you install conda using the instructions from Apple here, and then make a Scientific P圜harm project pointed at the conda executable? Also make sure you conda install tensorflow.

Download pycharm for apple silicon