zcbenz
Re:My thousand dollar iPhone can't do math
on Monday February 02, 2026 @06:46AM (#46854898)
It is a bug in MLX that has been fixed a few days ago:
https://github.com/ml-explore/mlx/pull/3083
Re:Apple's MLX adding CUDA support
on Monday July 14, 2025 @07:20PM (#44566412)
In the absence of hardware unified memory, CUDA will automatically copy data between CPU/GPU when there are page faults.
Re:PyTorch Internals: Ezyang's Blog
on Saturday March 22, 2025 @08:54PM (#43450005)
Most features are already supported on x86 CPUs, you can pip install mlx on Linux , and you can even use it on Windows (no official binary release yet but it is building and tests are passing).
Re:PyTorch Internals: Ezyang's Blog
on Saturday March 22, 2025 @06:18PM (#43449094)
For learning internals of ML frameworks I recommend reading the source code of MLX: https://github.com/ml-explore/mlx .
It is a modern and clean codebase without legacies, and I could understand most things without seeking external articles.
Re:Show HN: You need a few lines of JS, not a vector database
on Tuesday October 01, 2024 @07:30PM (#41715505)
You can use any serialization that supports numbers, in the repo I used BSER (a binary format of JSON), you can even use raw JSON but it would be very inefficient for float numbers though.
Show HN: Chrohime – cross-platform C library wrapping Chromium's native UI code
https://github.com/photoionization/chrohime
I wrote this library to explore the possibility of adding native UI APIs to the Electron framework, and I think the concept works quite well.The development (or experiment) stopped partly because I left MSFT, and partly because a CI infrastructure for building Chromium is too high for my own pocket.There is probably no future for it unless some organization is willing to sponsor the...
Re:Show HN: Sisi – Semantic Image Search CLI tool, locally without thi...
on Monday September 16, 2024 @09:48AM (#41556050)
No it is not normal, I only tested x64/arm64 macs, I will try on linux.
Re:Show HN: Sisi – Semantic Image Search CLI tool, locally without thi...
on Monday September 16, 2024 @09:46AM (#41556025)
That is sad, the name sisi comes from the sisi empress:
https://en.m.wikipedia.org/wiki/Empress_Elisabeth_of_Austria
Show HN: Sisi – Semantic Image Search CLI tool, locally without third party APIs
https://github.com/frost-beta/sisi
I wrote this tool to get familiar with CLIP model, I know many people have written similar tools with CLIP before, but I'm new to machine learning and writing a classic tool helps my study.The unusual thing with my version is, it is in pure Node.js, with the power of node-mlx, a Node.js machine learning framework.The repo in the link is mostly about implementing indexing and CLI, the code of...
V8 is damn fast, faster than C++ when it comes to numerics
V8 is damn fast, faster than C++ when it comes to numerics
Show HN: Train a language model from scratch with JavaScript
Re:Converting Codebases with LLMs
on Saturday July 20, 2024 @08:37AM (#41016032)
A few months ago I ported ~15k lines of python code (10k are tests) to typescript, using GPT4. It cost me ~$70.
The python project is https://github.com/ml-explore/mlx and the converted project is https://github.com/frost-beta/node-mlx
I wrote a long prompt: https://github.com/frost-beta/node-mlx/blob/main/tests/promp...
The first result was almost always bad, but after manually modifying the assistant's answer, following generation usually went much better.
Show HN: node-mlx – Machine learning framework for Node.js
https://github.com/frost-beta/node-mlx
I started studying machine learning this year, and I found python in my way to understand how things actually work, so I decided to port a machine learning framework to JavaScript first.I chose to port MLX, a library focused on Mac GPUs, because I mainly use Mac and I don't want to spend time on NVIDEA's environment, also I believe MLX and JavaScript are perfect match since so many JavaScript...
JavaScript's garbage collector is not very ready for machine learning
Show HN: Inference Llama2 with High-Level C++
Re:Sudo for Windows
on Thursday February 08, 2024 @07:33PM (#39309807)
I wrote a Node.js module that implemented similar feature years ago.
https://github.com/atom-archive/node-runas
(An updated fork can be found at https://www.npmjs.com/package/runas-redux)
Re:Dumbar, a Not So Smart Menubar App
on Saturday August 26, 2023 @08:24PM (#37278177)
Shameless to insert my app here, also open source and provides quick access from menu bar.
https://chie.app/posts/minimalist-approach-to-using-chatgpt/
Re:Show HN: Chie – a cross-platform, native, and extensible desktop cl...
on Saturday July 29, 2023 @08:26PM (#36926361)
I would argue it is not a large part a webview. The webview related code is 900 lines vs the whole project 10k lines.
Re:Show HN: Chie – a cross-platform, native, and extensible desktop cl...
on Saturday July 29, 2023 @08:13PM (#36926266)
I actually asked ChatGPT what chie might mean in other languages, for multiple times, but it did not tell me this :(
Show HN: Chie – a cross-platform, native, and extensible desktop client for LLMs
https://chie.app/posts/introduce-the-chie-app/
I'm submitting this before going to bed, so by the time it hits front page (if it does at all) I'll likely be asleep, so I won't be able to answer questions in time.
Re:What Is a Vector Database
on Friday May 05, 2023 @07:02AM (#35827656)
Suppose one is writing a desktop app doing searches of embeddings, is there a lightweight vector database that runs on all major desktop platforms? I'm aware of redis and sqlite-vss, but none runs natively on Windows.
Re:Plain C API design, the real world Kobayashi Maru test
on Sunday April 16, 2023 @11:06PM (#35596456)
Adding fields to context will break ABI compatibility.
To work around the problem, some libraries add padding fields, and some require passing the size of struct to determine ABI version.
Re:Show HN: GPT-4-powered web searches for developers
on Wednesday April 12, 2023 @10:46PM (#35550245)
Will you provide APIs for this service? I'm writing an extensive desktop app for all ChatGPT alike services, and this service is definitely a thing I want to include.
Re:Programmer interrupted: The cost of interruption and context switch...
on Wednesday April 05, 2023 @05:54PM (#35460914)
Since several years ago I was starting to have health issues that if I sat still for too long my eyes/arms/legs/waist would feel pain, sometimes fiercely. So I set a timer to remind me to stop staring at the monitor and stand up to do some light exercise, at first the timer was 40 minutes and in recent years it became 20 minutes as my body wears.
So now my work is interrupted every 20 minutes, at first the timer was annoying but after several months I got used to it completely and I don’t think it affects my productivity. And actually I wish I could do it from the begining so I wouldn’t have the health issues at all.
However my interruption is short and during it I was uaually just thinking about my work while stretching my body, so it is no compare to hard context switches like being dragged to a conversation.
Re:John Carmack: I loaned money to anyone that asked me
on Monday January 09, 2023 @01:11PM (#34313705)
When I was teen I lent some high value items in online game, and of course they never got back.
I’m still thankful that I learnt the lesson online instead of in real life.
Re:An M1 for Curl
on Friday December 30, 2022 @06:36AM (#34184118)
I have no relationship with Apple, but from my experience, the usual way big corp contributes to open source is to have their engineers sending patches, instead of sponsoring money/hardware directly.
For example Apple sent patches to some open source projects to add support for Apple Silicon: https://twitter.com/markvillacampa/status/127520044676491264...
Re:Everyone seems to forget why GNOME and GNOME 3 and Unity happened
on Wednesday July 27, 2022 @08:42PM (#32259060)
From my memory GNOME 3 and Unity were created because there was a trend to make all interfaces both desktop and mobile friendly, Ubuntu tried to ship a phone using Unity, and GNOME 3 was advertised for its tablet friendly interface.
For Windows world there was Metro UI of Windows 8, Apple insisted on using different UIs for desktop and mobile and was laughed at.