ahumanft

LLM Wiki V3: Segmentation

Karpathy gave us the foundation. Rohitg00 warned us what breaks. V3 is how you structure it to scale.

This is a concept document in the same spirit as V1 and V2.

V1 was intentionally vague. Build on it. V2 was intentionally open. Solve it.

CypherpunkSamurai

Here’s a list with the model/project name, a DOI (or stable arXiv identifier), and the GitHub repository where available.

• AlphaTensor - 10.1038/s41586-022-05172-4 - https://github.com/deepmind/alphatensor
• AlphaDev - 10.1038/s41586-023-06004-9 - https://github.com/google-deepmind/alphadev
• AlphaEvolve - 10.48550/arXiv.2409.09756 - N/A
• AlgoPilot - 10.48550/arXiv.2410.18944 - N/A
• FunSearch - 10.1038/s41586-023-06924-6 - https://github.com/google-deepmind/funsearch
• Latent Program Network (LPN) - 10.48550/arXiv.2306.05278 - https://github.com/ziyu-chen/lpn
• Neural Logic Induction (Gradient-Based Program Synthesis) - 10.48550/arXiv.2109.04870 - https://github.com/nju-websoft/NLI
• AI Feynman - 10.1126/sciadv.aay2631 - https://github.com/SJ001/AI-Feynman

karpathy

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

OmerFarukOruc

AI Coding Agent Guidelines (claude.md)

These rules define how an AI coding agent should plan, execute, verify, communicate, and recover when working in a real codebase. Optimize for correctness, minimalism, and developer experience.

---

Operating Principles (Non-Negotiable)

• Correctness over cleverness: Prefer boring, readable solutions that are easy to maintain.
• Smallest change that works: Minimize blast radius; don't refactor adjacent code unless it meaningfully reduces risk or complexity.

peteristhegreat

Quick AI analysis of this funny history:

Original Post: A Brief, Incomplete, and Mostly Wrong History of Programming Languages

Original Post by James Iry on Thursday, May 7, 2009

https://james-iry.blogspot.com/2009/05/brief-incomplete-and-mostly-wrong.html

Quick Analysis/Dissection for the Uninitiated

eylenburg

Step by step guide: How to install Microsoft Office in any Linux distribution

There are multiple options how to install MS Office on Linux.

VM-based - Integrate Windows apps running in a Windows virtual machine as native-looking in Linux

1. LinOffice - Microsoft Office Launcher for Linux, my own fork of Winapps which is focused on only running Microsoft Office, with some Office-specific improvements over Winapps and a fully automated setup. Eventually I would like to create a GUI for it. Decribed below
2. Winapps, based on KVM, QEMU, Docker/Podman and FreeRDP. Still actively maintained (getting Github commits). Decribed below
3. Cassowary, based on KVM, QEMU, libvirt/virt-manager, and FreeRDP. Last release in Feb 2022 and seems to be abandoned.

veekaybee

Anti-hype LLM reading list

Goals: Add links that are reasonable and good explanations of how stuff works. No hype and no vendor content if possible. Practical first-hand accounts of models in prod eagerly sought.

Foundational Concepts

Pre-Transformer Models

lostintangent

Expandable Animated Card Slider

We have made an expandable animated card slider, it will expand and collapse based on card click. We used owl carousel and jQuery for variable width and responsive slider.

A Pen by Yudiz Solutions Limited on CodePen.

License.

yoavg

Reinforcement Learning for Language Models

Yoav Goldberg, April 2023.

Why RL?

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

rain-1

Purpose

Bootstrap knowledge of LLMs ASAP. With a bias/focus to GPT.

Avoid being a link dump. Try to provide only valuable well tuned information.

Prelude

Neural network links before starting with transformers.