A course by Andrej Karpathy on building neural networks, from scratch, in code. We start with the basics of backpropagation and build up to modern deep neural networks, like GPT. In my opinion language models are an excellent place to learn deep learning, even if your intention is to eventually go to other areas like computer vision because most of what you learn will be immediately transferable. This is why we dive into and focus on languade models.
What is the magic of GPT? I want to understand how GPT works but can easily get lost in so many terms and papers so by looking at a minimal implementation give me some ideas.
Don’t let the “Go code” put you off if you aren’t a Go user. Bit manipulation is the fundamental in programming. It can do quite a cool tricks such as check same sign(positive/negative) by looking at the last bit of both number or a fast way to compute avg.
FILTER and CASE can be use in an aggregate funcion in your SQL to caculate a value from a set of row that match the conditions. They looks similar in term of functionality, but FILTER is much nicer to read because its appear on the left of your aggregate function, not wrap around like case. So is there any perf impact? Generally, no, you can safely switch to FILTER.
You do not have to follow these rules every time. If you have a good reason to break any of them, do. But they are safe to follow every time.
There is a certain directory on MacOS has a specific icon in Finder. So let looks at how they work
explain how YouTube works under the hood, how it serves streams to the client, and also how you can exploit that knowledge to download videos from the site
A window function performs a calculation across a set of table rows that are somehow related to the current row. It’s very flexible because it doesn’t need to pre-group row, and can work It’s useful to calculate extra data between the current row and a group of row related to current row by any condition that you can express in code. This post has many practical use of window function to calculate ranking
Introduce many approachs we can use to do responsive design in 2023. If you never heard of
minxmax, Size Container Queries.. time to refresh your CSS skill for 2023.
A pathfinding visualizer made in Python and Pygame. This project aims to provide a fun and interactive way to learn about popular pathfinding algorithms such as Dijkstra’s, A* and other supported algorithms.Python
Better structured concurrency for goGo
a library preloader to speed up process startup, uses a client/server approach. The server loads the libraries and listens on a UNIX socket, and the client connects to that socket to run a process. For each client connection, the server forks a worker process, which uses the current directory, stdin, stdout, stderr, and environment of the client process. The worker process then processes the arguments provided by the client.Ruby
If you need to handle many common spreadsheet types, looks no more. It supports Excel, Libre, CSV, and even google sheet.Ruby
boxxy is a tool for boxing up misbehaving Linux applications and forcing them to put their files and directories in the right place, without symlinks! It uses Linux namespaces for its functionality. Kind of like docker, but here simply prefix your command with
boxxy and boxxy setup a namespace file system with dir mount etc then run your process inside that namespace. It’s Rust but you will learn a lot about this concept. Highly recomended to check it out.
A minimal PyTorch re-implementation of the OpenAI GPT (Generative Pretrained Transformer) trainingPython
Andrej implements nanoGPT which we linked to in issue #238 and this is a video to show some aspect of its
a Postgres extension for adding hypothetical indexes to Postgres so you can do index planning. Read the introduction article to get a sense of it.
To run a snippet of JS on any page, you can pop into the console
a distributed POSIX file system built on top of Redis and S3.
When working with message queue we sometimes need ability to peek into the queue from CLI. With so many solutions like Kafka, RabbitMQ, PubSub, SNS,Nats, Redis,… it’s hard to remember all the syntax. This tool rule them all using a consistent syntax. Support a lot of message queues system.
A cli tool to proxy and analyze TCP connections.
Query, provision and operate Cloud and SaaS resources(such as AWS, GCP, Azure, Okta, Github,…) and APIs using an extensible SQL. Super useful when you want to have a big picture of what resources we have in a big org, not just for AWS, GCP but say how many users we have in Github and who belongs to which group
aims to be a reliable, easy-to-use backup and restore solution that can seamlessly scale up to the largest databases and workloads by utilizing algorithms that are optimized for database-specific requirements.
a template for you to create your own customized, high-availability solution using Python and - for maximum accessibility - a distributed configuration store like ZooKeeper, etcd, Consul or Kubernetes.
List of ngrok alternatives and other ngrok-like tunneling software and services. Focus on self-hosting.
an observability tool that consumes logs from various inputs, transforms them, and exports logs and/or metrics to a chosen target.
An OpenShift / Kubernetes operator for network observability
Better Dev Link