Code More

Using TensorFlow addons on Mac with M1

These days I am building an image classifier using an unbalanced dataset. So instead of Accuracy, I am more interested in viewing Precision and Recall, or the F1 score. The F1 score metric is implemented by https://www.tensorflow.org/addons/api_docs/python/

tensorflow

Using the Keras Tuner

Recently I read about the Keras Tuner. A hyperparameter tuner in machine learning is a piece of software that will train random combinations of models in order to find the best architecture. See https://www.tensorflow.org/tutorials/keras/keras_tuner. The idea is

m1

Homebrew on Mac with M1

Recently I needed to install graphviz. On their official site, they say one way is to use Homebrew. Workaround: use Rosetta 2Here's what I got: % brew install graphviz Error: Cannot install in Homebrew on ARM processor in Intel default prefix (/usr/local)! Please create

m1

Installing scikit-learn, panda, scipy, numpy on a Mac with M1

Before installing scikit-learn, you should install its dependencies. Installing PandasPandas installs fine with pip install pandas. However, when I tried to import it in my Jupyter Lab notebook, it crashed with this error: ValueError Traceback (most recent call last) <ipython-input-2-b6c7f9fc9652> in <

machine learning

Survey of Keras's image related generators

In François Chollet's Deep learning with Python, he teaches how to build image classification models using ImageDataGenerators. Let's look into how these are made and how to use them. tf.keras.preprocessing.image.ImageDataGeneratorGenerate batches of tensor image data with real-time data augmentation.TensorFlowBasic

tensorflow

Plotting discrimination thresholds in TensorFlow

scikit-learn can render precision, recall and F1 score depending on thresholds in a chart. I wanted the same to evaluate my binary classifier made with TensorFlow. See how it looks like on scikit-learn. The scikit-learn chart I set out to buildO(data × number of

tensorflow

Installing TensorFlow 2.5 and Jupyter Lab on Mac with M1

Last month, I finally painstakingly installed TensorFlow 2.4 and Jupyter Lab on my Mac with M1 (see the blog post). It worked nicely: 10 times faster than Colab, but also had a few issues like working only with Python 3.8, having to

python

How to compute things in parallel in Python

I had a loop that could be parallelized and wanted to take advantage of my CPU cores. A quick search led me to https://docs.python.org/3/library/asyncio-task.html#running-tasks-concurrently. It looked simple enough: define your methods as async, then run them

numpy

Fixing `TypeError: only integer scalar arrays can be converted to a scalar index`

This will be a short post. I was trying to concatenate two numpy arrays, when I got this exception: TypeError: only integer scalar arrays can be converted to a scalar indexI was following https://numpy.org/doc/stable/reference/generated/numpy.concatenate.html. It

machine learning

Displaying a confusion matrix in TensorFlow

I recently started dabbling in TensorFlow by training an image classifier. My classifier has a few classes, and instead of blindly judging its accuracy, I decided to look at the confusion matrix to see which classes it recognized well and which it didn't. Multiclass

flutter

Profiling a Flutter app

Today I ran into some performance issue when I used my image gallery app on a folder of 17k images. Lots of jank. So I decided to use the Flutter profiler for the first time. I started from the Flutter performance profiling page. Flutter

flutter

Fixing `Unable to load asset: FontManifest.json` when using golden_toolkit

I recently started using eBay's golden_toolkit for screenshot tests instead of Flutter's built-in matchesGoldenFile. My main issue with the built-in function was that it would use the Ahem font, which is all blocks. It makes it much harder to see if your app

firestore

How come Firestore won't write anything anymore?

I had been scratching my head since last night. For some reason, when creating lots of new documents, only the first batch would show up in my UI. I'd put a print before and after the await batch.commit() and see that it would

Using Firebase in Flutter on macOS

Firebase AuthAuthentication methodsI typically use Google Sign In for my projects, but as you can see on https://pub.dev/packages/google_sign_in, it only supports Android, iOS and Web, and there is no GitHub issue tracking Desktop. Phone authentication also required extra

tensorflow

TensorFlow reference docs

In a previous post, I collected reference documentation for Python at http://blog.wafrat.com/finding-the-python-reference-docs/. Let's do the same for TensorFlow: https://www.tensorflow.org/api_docs/python/tf/dtypes/DTypehttps://keras.io/api/preprocessing/image/https://www.tensorflow.org/api_docs/python/

flutter

Getting stack traces for published Flutter apps

The tool recommended by Flutter is Sentry. I've tried it and it's perfect for me. First you check out the tutorial about it. Report errors to a serviceHow to keep track of errors that users encounter.Flutter LogoWhat does it look like?With Sentry,

flutter web

Setting up Sentry for Flutter Web

On the official Flutter docs, they say you can run flutter build web to compile for a release, and it'll generate everything you need, including source maps. Build and release a web appHow to prepare for and release a web app.Flutter LogoSee main.

flutter

GridView with fixed height in Flutter

I wanted to convert a ListView into a GridView so that it would show more information on a large screen such as Web. My ListView code looked like this: return ListView.builder( physics: scrollable ? null : NeverScrollableScrollPhysics(), shrinkWrap: shrinkWrap, padding: EdgeInsets.only(top: 16.0)

flutter

Migrating a large Flutter project to null safety

According to the guide at https://dart.dev/null-safety/migration-guide, migrating to null safety is straightforward. Let's try it on a large project. So today I switched the sdk version to 2.12 in my pubspec.yaml, and 300+ compilation errors appeared. I tried

flutter

On the importance of the Flutter environment sdk version

Yesterday I made a quick update to my app. All the code compiled and all the tests succeeded, so I published an update. This morning, I wake up to a half broken app. What happened? The simplified snippet was: await Firebase.initializeApp(); final instance

dart

Working with JSON in Dart

Working with JSON in Dart is relatively easy, but there are still a few caveats. Handling dynamic typesThere's a page dedicated to JSON in Dart at https://dart.dev/guides/json. The first way is to use dart:convert's jsonDecode(String) => dynamic. I

dart

Map through a list with index in Dart

In Javascript, it's easy to map through a list with the value and index. In Dart, you only get the value, but not the index. One of the top results on Google for this issue is this StackOverflow post from 2 years ago: https:

flutter

Upgrading pre 1.12 Android projects based on Kotlin

My old Flutter app has been complaining I need to do this migration for a while. Today I finally decided to do it. flutter/flutterFlutter makes it easy and fast to build beautiful apps for mobile and beyond. - flutter/flutterflutterGitHubBased on the criteria

flutter

My new Flutter project is stuck in assembleGradle

I created a new project by running flutter create .. Then I modified the Android app's package by modifying it in: android/app/build.gradleandroid/app/src/main/AndroidManifest.xmlI also integrated Firestore to it, and added an Android app there, with the new package

flutter

Fixing `com.android.builder.dexing.DexArchiveMergerException: Error while merging dex archives: The number of method references in a .dex file cannot exceed 64K.` in Flutter when using Firestore

I created a new project, integrated Firestore to it, and on the first run, I got this error. A quick search got me to https://github.com/flutter/flutter/issues/21005. The error also shares this link: https://developer.android.com/tools/building/multidex.