| module AlgorithmW | |
| // HINDLEY-MILNER TYPE INFERENCE | |
| // Based on http://catamorph.de/documents/AlgorithmW.pdf | |
| // (Now at http://web.archive.org/web/20170704013532/http://catamorph.de/documents/AlgorithmW.pdf) | |
| type Lit = | |
| | LInt of int | |
| | LBool of bool |
Upon completion you will have a sane, productive Haskell environment adhering to best practices.
sudo apt-get install libtinfo-dev libghc-zlib-dev libghc-zlib-bindings-dev
| " Vim syntax file | |
| " Language: Todo | |
| " Maintainer: Huy Tran | |
| " Latest Revision: 14 June 2020 | |
| if exists("b:current_syntax") | |
| finish | |
| endif | |
| " Custom conceal |
At the beginning of 2030, I found this essay in my archives. From what I know today, I think it was very insightful at the moment of writing. And I feel it should be published because it can teach us, Rust developers, how to prevent that sad story from happening again.
What killed Haskell, could kill Rust, too
What killed Haskell, could kill Rust, too. Why would I even mention Haskell in this context? Well, Haskell and Rust are deeply related. Not because Rust is Haskell without HKTs. (Some of you know what that means, and the rest of you will wonder for a very long time). Much of the style of Rust is similar in many ways to the style of Haskell. In some sense Rust is a reincarnation of Haskell, with a little bit of C-ish like syntax, a very small amount.
Is Haskell dead?
| # Sources used in figuring this out: | |
| # | |
| # - https://developer.mozilla.org/en-US/docs/WebAssembly/C_to_wasm | |
| # - https://emscripten.org/docs/compiling/Building-Projects.html#building-projects | |
| # Set up Emscripten SDK | |
| git clone https://github.com/emscripten-core/emsdk.git | |
| cd emsdk/ | |
| ./emsdk install latest | |
| ./emsdk activate latest |
| fun Any.prettyPrint(): String { | |
| var indentLevel = 0 | |
| val indentWidth = 4 | |
| fun padding() = "".padStart(indentLevel * indentWidth) | |
| val toString = toString() | |
| val stringBuilder = StringBuilder(toString.length) |
Chose between natural mode like MacOS or Windows default mode.
You can do this by going to Start Menu, type PowerShell, and click Run as Administrator.
$mode = Read-host "How do you like your mouse scroll (0 or 1)?"; Get-PnpDevice -Class Mouse -PresentOnly -Status OK | ForEach-Object { "$($_.Name): $($_.DeviceID)"; Set-ItemProperty -Path "HKLM:\SYSTEM\CurrentControlSet\Enum\$($_.DeviceID)\Device Parameters" -Name FlipFlopWheel -Value $mode; "+--- Value of FlipFlopWheel is set to " + (Get-ItemProperty -Path "HKLM:\SYSTEM\CurrentControlSet\Enum\$($_.DeviceID)\Device Parameters").FlipFlopWheel + "`n" }
| #!/usr/bin/env python3 | |
| import argparse | |
| import glob | |
| import os | |
| import struct | |
| import sys | |
| from sentencepiece import SentencePieceProcessor | |
| HPARAMS = keys = ["vocab_size", "dim", "multiple_of", "n_heads", "n_layers"] |
Yoav Goldberg, April 2023.
With the release of the ChatGPT model and followup large language models (LLMs), there was a lot of discussion of the importance of "RLHF training", that is, "reinforcement learning from human feedback". I was puzzled for a while as to why RL (Reinforcement Learning) is better than learning from demonstrations (a.k.a supervised learning) for training language models. Shouldn't learning from demonstrations (or, in language model terminology "instruction fine tuning", learning to immitate human written answers) be sufficient? I came up with a theoretical argument that was somewhat convincing. But I came to realize there is an additional argumment which not only supports the case of RL training, but also requires it, in particular for models like ChatGPT. This additional argument is spelled out in (the first half of) a talk by John Schulman from OpenAI. This post pretty much
The target audience is people who are familiar with Urbit's architecture, though not necessarily much of its code.
As some of you already know, i recently left my job as a core dev for the Urbit Foundation to work on a similar system called Plunder. Plunder was created in 2020 by two former Tlon employees, after their proposal for a new version of Nock was rejected. They have since reworked that significantly and built a reference implementation of their own system. You can follow its continued development on its mailing list.
I've known about Plunder for quite some time now, but their recently released demo -- in which the system is used to serve a 70 GB dataset, complete with metadata and searchable -- made me feel the need to explore it again and in greater detail. Doing this with my personal server doesn't feel like a big ask, but there is currentl