mathisto

Orchestrating cheap agents for f3decomp: notes from a Swarm day

These are my notes from a long session: one expensive model orchestrating a fleet of cheaper ones through the OpenCode CLI. It took the account from Power to Swarm in a day. Everything below is a pattern you can copy; the model results are one day's data, so treat them as direction, not gospel.

For the matching technique itself (twins, literal ports, the symptom → fix table), see MATCHING_TIPS.md in mathisto/f3twins. This doc is only about running many agents at once.

mathisto

Fallout 3 matching: agent prompt (VS2005 SP1 /O2)

You are matching one Fallout 3 PC 1.7.0.3 function on f3decomp: write C++ that VS2005 SP1 (cl 14.00.50727.762) compiles to exactly the target bytes.

The core idea

You are reproducing the compiler's decisions, not guessing the programmer's style. Once the code is basically right, almost every leftover difference comes from one of three things the compiler counts: swapped or rotated registers, x87 operand order, stack slot layout, one load in the wrong place.

mathisto

Title: MemGPT: Making AI Chatbots and Systems Smarter

Imagine your computer could only remember what happened in the last five minutes. That's a bit like how large language models (LLMs)—the brains behind AI like chatbots—work today. They're impressive when responding to recent conversations or small bits of information, but they hit a wall when the conversation drags on or they have a mountain of data to process.

What’s the Problem?

LLMs can only deal with a limited amount of information at once, sort of like trying to cram all your groceries into a small bag. This means they're not great at keeping up with long chats or analyzing big documents. Extending their memory isn't as simple as it sounds and gets expensive in terms of computing power.

Introducing MemGPT

mathisto

Here's a high-level breakdown of your project in bullet points:

Project Overview: Eidolon - Infinite Contextual Memory for LLMs

• Objective:
  • Provide an infinite memory solution for any LLM by storing and retrieving conversational context dynamically.

---

Key Features

mathisto

Eidolon: Infinite Memory Chat Companion

Objectives

1. OpenAI Compatible Endpoints:

   • Expose RESTful API endpoints compatible with OpenAI’s API for seamless integration.

2. Infinite Archival Memory:

mathisto

High-Level Instructions

1.	Create a new branch for your work.
2.	Switch to the new branch.
3.	Make changes and commit them.
4.	Check your work (status and log).
5.	Push the branch to the remote repository.

Git Bash Commands

mathisto

Heaven Tunnel X-Junction

{
  "name": "Heaven Tunnel X-Junction",
  "statePosArrayList": "{blockstatemap:[{Name:\"minecraft:air\"},{Name:\"chisel_chipped_integration:factory_wireframe\"},{Name:\"antiblocksrechiseled:bright_white\"},{Name:\"chisel_chipped_integration:laboratory_tiles\"},{Name:\"antiblocksrechiseled:bright_black\"},{Name:\"antiblocksrechiseled:bright_yellow\"}],endpos:{X:0,Y:10,Z:32},startpos:{X:-32,Y:0,Z:0},statelist:[I;0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,3,3,3,3,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,3,3,3,4,3,3,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,3,3,0,0,0,0,0,3,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,3,3,0,0,0,0,0,3,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,3,3,0,0,0,0,0,3,3,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,3,3,0,0,0,0,0,3,3,0,0,0,0,

mathisto

Pyro's Comprehensive Tips on Bees 1

The majority of this post is written under the assumption that you’re using Industrial Apiaries and the Advanced Mutatron. Everything is doable with standard apiaries and alvearies, but will be more difficult.

Bee Breeding is a fairly large part of Meatballcraft and understanding how to do it efficiently can save you a lot of time in the long run and also aid you in generating some less-enjoyable materials, such as Gaia Spirits and Flux. I want to start with an approach where I first present the end goal of your infrastructure and then how we approach reaching that point, followed with an example in how to use this.

You will need, in addition to the Industrial Apiaries and Advanced Mutatron, a Genetic Sampler, a Genetic Transposer, Genetic Replicator, DNA Extractor, Protein Liquefier, and Genetic Imprinter. These are listed in order of priority. It is also strongly recommended that you have access to generation of Rocky Hives and a Hibeescus.

Our end goal is to, as w

mathisto

An opinionated list of excellent Clojure learning materials

These resources (articles, books, and videos) are useful when you're starting to learn the language, or when you're learning a specific part of the language. This an opinionated list, no doubt. I've compiled this list from writing and teaching Clojure over the last 10 years.

• 🔴 Mandatory (for both beginners and intermediates)
• 🟩 For beginners
• 🟨 For intermediates

Table of contents

1. Getting into the language

mathisto

Keybase proof

I hereby claim:

• I am mathisto on github.
• I am mathisto (https://keybase.io/mathisto) on keybase.
• I have a public key ASBMGtLCM_Ibq4VDVHFe3zH8dKWOPTCvGBPCHhyAR0z5Awo

To claim this, I am signing this object: