Victor Taelin

CanonicalLean. A Lean tactic for Canonical, a search procedure for terms in dependent type theory.

public-apis. A collective list of free APIs

HRM-Text. HRM-Text is a 1B text generation model based on the HRM architecture, strengthened by task completion and latent space reasoning.

plouf. Proof assistant

llm-as-a-verifier. Python

dflash. DFlash: Block Diffusion for Flash Speculative Decoding

tailslayer. Library for reducing tail latency in RAM reads

hermes-agent. The agent that grows with you

autonomous-loop. Codex autonomous loop controller that prevents premature agent exits with frozen contracts, trusted gates, and fail-closed stop hooks.

ANE. Training neural networks on Apple Neural Engine via reverse-engineered private APIs

openclaw. Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

claude-mem. Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

beads. Beads - A memory upgrade for your coding agent

claude-rolling-context. Smart context window manager for ClaudeCode transcripts - maintains continuous conversations with intelligent trimming and optional LLM summaries

hvm. A Zig implementation of HVM - the Higher-Order Virtual Machine based on Interaction Calculus.

ZkLinalg. Formally proving the security of Fast Reed-Solomon interactive oracle proofs of proximity

soma. ⚗️ | Soma is a general-purpose dependently-typed functional programming language powered by Interaction Nets with a minimal runtime.

nanochat. The best ChatGPT that $100 can buy.

TinyRecursiveModels. Python

codex. Lightweight coding agent that runs in your terminal

syrup_artifacts. OCaml

L2. λ² is a tool for synthesizing functional programs from input-output examples.

trio_artifacts. OCaml

BurstArtifactEvaluation. Artifact Evaluation for Bottom-up Synthesis of Recursive Functional Programs using Angelic Execution

leon. The Leon system for verification, synthesis, repair

myth. Type-and-example-driven program synthesis for functional programming languages

MPMCQueue. A bounded multi-producer multi-consumer concurrent queue written in C++11

fastqueue2. Yet another blazingly fast SPSC queue.

lfqueue. A Scalable, Portable, and Memory-Efficient Lock-Free FIFO Queue (DISC '19)

Kimina-Prover-Preview. Technical report of Kimina-Prover Preview.

unsloth. Unsloth Studio is a web UI for training and running open models like Gemma 4, Qwen3.6, DeepSeek, gpt-oss locally.

opencode. The open source coding agent.

claude-code-router. Use Claude Code as the foundation for coding infrastructure, allowing you to decide how to interact with the model while enjoying updates from Anthropic.

qwen-code. An open-source AI coding agent that lives in your terminal.

Qwen3-Coder. Qwen3-Coder is the code version of Qwen3, the large language model series developed by Qwen team.

robin. Robin: A multi-agent system for automating scientific discovery

hoc-notes. Notes written while consulting for the Higher Order Company

multi-llm-ts. A Typescript library to use LLM providers APIs in a unified way.

gemini-cli. An open-source AI agent that brings the power of Gemini directly into your terminal.

opencode. A powerful AI coding agent. Built for the terminal.

LLMs-from-Scratch. Build a Large Language Model (From Scratch) book and Finetuned Models

HVM3-Strict. [Loc:48 | Lab:8 | Tag:8] version

ghc-wasm-miso-examples. Haskell

par-lang. Par (⅋) is an experimental concurrent programming language. It's an attempt to bring the expressive power of linear logic into practice.

asm-lessons. FFmpeg Assembly Language Lessons

hvm-regex. initial exploration how hvm3 regex implementation will hold up, documenting the attempt, much to be improved

TypeChecker. A simple type checker in Haskell for the simply typed lambda calculus.

hvm3bountysubmission. Python

noise_step. noise_step: Training in 1.58b With No Gradient Memory

scrybe. Type-and-example directed program synthesis using example propagation, as described in Program Synthesis Using Example Propagation.

smyth. Programming-by-example in a typed functional language with sketches. Smyth = Sketching + Myth.

louise. Polynomial-time Meta-Interpretive Learning

mazeppa. A modern supercompiler for call-by-value functional languages

llvm-hs. Haskell bindings for LLVM

eclair-lang. A minimal, fast Datalog implementation in Haskell that compiles to LLVM IR

node-addon-api. Module for using Node-API from C++

BitNet. Official inference framework for 1-bit LLMs

aider. aider is AI pair programming in your terminal

MisguidedAttention. A collection of prompts to challenge the reasoning abilities of large language models in presence of misguiding information

GPT-Haskell. A pure Haskell implementation of a decoder-only transformer (GPT)

entropix. Entropy Based Sampling and Parallel CoT Decoding

ai.c. gpt written in plain c

cctt. high-performance cubical evaluation

agda-pkg. apkg - package manager for Agda

ComfyUI. The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.

zipperposition. An automatic theorem prover in OCaml for typed higher-order logic with equality and datatypes, based on superposition+rewriting; and Logtk, a supporting library for manipulating terms, formulas, clauses, etc.

GPU-Benchmarks-on-LLM-Inference. Multiple NVIDIA GPUs or Apple Silicon for Large Language Model Inference?

typst. A markup-based typesetting system that is powerful and easy to learn.

cline. Autonomous coding agent as an SDK, IDE extension, or CLI assistant.

agda-check. Check agda commands from your terminal

vim-agda. Asynchronous type-checking for Agda 2.6.2.

yatima. A zero-knowledge Lean4 compiler and kernel

goodai-ltm-benchmark. A library for benchmarking the Long Term Memory and Continual learning capabilities of LLM based agents. With all the tests and code you need to evaluate your own agents. See more in the blogpost:

matmulfreellm. Implementation for MatMul-free LM.

chatbox. Powerful AI Client

lazy.nvim. 💤 A modern plugin manager for Neovim

lean.nvim. Neovim support for the Lean theorem prover

gateway. A blazing fast AI Gateway with integrated guardrails. Route to 1,600+ LLMs, 50+ AI Guardrails with 1 fast & friendly API.

IntelliNode. Access the latest AI models like ChatGPT, LLaMA, Deepseek, Diffusion, Hugging face, and beyond through a unified prompt layer and performance evaluation

litellmjs. JavaScript implementation of LiteLLM.

litellm. Python SDK, Proxy Server (AI Gateway) to call 100+ LLM APIs in OpenAI (or native) format, with cost tracking, guardrails, loadbalancing and logging. [Bedrock, Azure, OpenAI, VertexAI, Cohere, Anthropic, Sagemaker, HuggingFace, VLLM, NVIDIA NIM]

z3. The Z3 Theorem Prover

MicroHs. Haskell implemented with combinators

ingpu. GPU-based interaction nets evaluator

posit-javascript. JavaScript

bfp. Beyond Floating Point - Posit C/C++ implementation

universal. Large collection of number systems providing custom arithmetic for mixed-precision algorithm development and optimization for AI, Machine Learning, Computer Vision, Signal Processing, CAE, EDA, control, optimization, estimation, and approximation.

simple-evals. Python

llm.c. LLM training in simple, raw C/CUDA

clevergpt. Training GPTs to solve interaction nets

schedule_free. Schedule-Free Optimization in PyTorch

tinyllama-bitnet. Train your own small bitnet model

hvm-compare-perf. Compare `hvm-core`'s historical performance

grok. Python

Bend. A massively parallel, high-level programming language

anthropic-sdk-typescript. Access to Anthropic's safety-first language model APIs in TypeScript

picoGPT. An unnecessarily tiny implementation of GPT-2 in NumPy.

OS-Copilot. An self-improving embodied conversational agent seamlessly integrated into the operating system to automate our daily tasks.

pcc116-2021-2. JavaScript

mamba-minimal. Simple, minimal implementation of the Mamba SSM in one file of PyTorch.

mergekit. Tools for merging pretrained large language models.

self-rewarding-lm-pytorch. Implementation of the training framework proposed in Self-Rewarding Language Model, from MetaAI

AlphaCodium. Official implementation for the paper: "Code Generation with AlphaCodium: From Prompt Engineering to Flow Engineering""

TOVA. Token Omission Via Attention

llama2.c. Inference Llama 2 in one file of pure C

rinha-de-algoritmos. Na Rinha de Algoritmos você deve utilizar suas habilidades para a criação de algoritmos eficientes para resolver problemas!

BitNet. Implementation of "BitNet: Scaling 1-bit Transformers for Large Language Models" in pytorch

lean-riemann-hypothesis. Riemann Hypothesis in Lean

logpy. Logic Programming in Python

itrees. Rust

LLM_Tree_Search. (ICML 2024) Alphazero-like Tree-Search can guide large language model decoding and training

image2ascii. :foggy: Convert image to ASCII

presenting.vim. A simple tool for presenting slides in vim based on text files.

miniF2F. Formal to Formal Mathematics Benchmark

gptme. Your agent in your terminal, equipped with local tools: writes code, uses the terminal, browses the web. Make your own persistent autonomous agent on top!

ChatGLM3. ChatGLM3 series: Open Bilingual Chat LLMs | 开源双语对话语言模型

DreamCraft3D. [ICLR 2024] Official implementation of DreamCraft3D: Hierarchical 3D Generation with Bootstrapped Diffusion Prior

telegram-math-quiz. JavaScript

hvm-core. Cuda

tinygrad. You like pytorch? You like micrograd? You love tinygrad! ❤️

compiler. Compiler for Elm, a functional language for reliable webapps.

nvim-metals. A Metals plugin for Neovim

rinha-de-compiler. 🥖 | Rinha de compiladores (ou interpretadores kkkk

AgentBench. A Comprehensive Benchmark to Evaluate LLMs as Agents (ICLR'24)

clint-llm.github.io. Clint LLM GitHub Pages

CUDA-Stream-Compaction. Cuda

WizardVicunaLM. LLM that combines the principles of wizardLM and vicunaLM

cornelis. agda-mode for neovim

PGMax. Loopy belief propagation for factor graphs on discrete variables in JAX

nanoT5. Fast & Simple repository for pre-training and fine-tuning T5-style models

gpt-engineer. CLI platform to experiment with codegen. Precursor to: https://lovable.dev

ZyPlugin. Java

OpenLLM. Run any open-source LLMs, such as DeepSeek and Llama, as OpenAI compatible API endpoint in the cloud.

interact. Interaction net / combinator language based on https://core.ac.uk/download/pdf/81113716.pdf

MEGABYTE-pytorch. Implementation of MEGABYTE, Predicting Million-byte Sequences with Multiscale Transformers, in Pytorch

qlora. QLoRA: Efficient Finetuning of Quantized LLMs

textgen. Open-source desktop app for local LLMs. Text, vision, tool-calling, OpenAI/Anthropic-compatible API. 100% private.

agda2hvm. Haskell

Neural-Networks-on-Silicon. This is originally a collection of papers on neural network accelerators. Now it's more like my selection of research on deep learning and computer architecture.

rwkv.cpp. INT4/INT5/INT8 and FP16 inference on CPU for RWKV language model

langchain. The agent engineering platform.

tree-of-thought-llm. [NeurIPS 2023] Tree of Thoughts: Deliberate Problem Solving with Large Language Models

micrograd. A tiny scalar-valued autograd engine and a neural net library on top of it with PyTorch-like API

trazodone. 🧪 | LLVM backend for HVM

gpt4all. GPT4All: Run Local LLMs on Any Device. Open-source and available for commercial use.

ReplitLM. Inference code and configs for the ReplitLM model family

adaptive. :chart_with_upwards_trend: Adaptive: parallel active learning of mathematical functions

deque. Purely Functional, Real-Time Deques with Catenation (Kaplan & Tarjan)

rwkv-cpp-accelerated. A torchless, c++ rwkv implementation using 8bit quantization, written in cuda/hip/vulkan for maximum compatibility and minimum dependencies

safari. Convolutions for Sequence Modeling

flash-attention. Fast and memory-efficient exact attention

works. The code of several works on oimo.io/works

machine-learning-book. Code Repository for Machine Learning with PyTorch and Scikit-Learn

minGPT. A minimal PyTorch re-implementation of the OpenAI GPT (Generative Pretrained Transformer) training

nanoGPT. The simplest, fastest repository for training/finetuning medium-sized GPTs.

mathlib3. Lean 3's obsolete mathematical components library: please use mathlib4

AutoGPT. AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.

rwkvk-rs. Rust

bsc. Bluespec Compiler (BSC)

RWKV-LM. RWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.

awesome-gpt4. A curated list of prompts, tools, and resources regarding the GPT-4 language model.

chat-with-gpt. An open-source ChatGPT app with a voice

bloop. bloop is a fast code search engine written in Rust.

vim-ai. AI-powered code assistant for Vim. OpenAI and ChatGPT plugin for Vim and Neovim.

Orbis. A general-purpose layer 2 zk-rollup scaling solution for Cardano

differentiable-plasticity. Implementations of the algorithms described in Differentiable plasticity: training plastic networks with gradient descent, a research paper from Uber AI Labs.

llama-dl. High-speed download of LLaMA, Facebook's 65B parameter GPT model

llama.cpp. LLM inference in C/C++

SpikeGPT. Implementation of "SpikeGPT: Generative Pre-trained Language Model with Spiking Neural Networks"

Grassmann.jl. ⟨Grassmann-Clifford-Hodge⟩ differential geometric algebra

llama. Inference code for Llama models

nohm. Nondeterministic Optimal Higher-order Machine

mm-cot. Official implementation for "Multimodal Chain-of-Thought Reasoning in Language Models" (stay tuned and more will be updated)

lafont. Animation of Yves Lafont's Interaction Combinators

Anime-Girls-Holding-Programming-Books. Anime Girls Holding Programming Books

kind2lisp. a small one-file lisp implemented in kind2, intentionally incomplete

Haskell4Noobs. Único roadmap de Haskell possível, do noob ao Thanos

circcash. Circcash is the only cryptocurrency with a mining algorithm that is designed to advance science. Circcash mining will accelerate the development of reversible computational hardware.

Production-RWKV. This project aims to make RWKV Accessible to everyone using a Hugging Face like interface, while keeping it close to the R and D RWKV branch of code.

ChatRWKV. ChatRWKV is like ChatGPT but powered by RWKV (100% RNN) language model, and open source.

klo-gba.js. 🧢 Reverse engineering tool for the Klonoa's GBA game

BOHM1.1. Bologna Optimal Higher-Order Machine, Version 1.1

SICAM. A parallel general purpose computer based on interaction combinators

evm. Pure Rust implementation of Ethereum Virtual Machine

celloc. cell allocator with hard-realtime garbage collection AND DEDUPLICATION

cranelift-jit-demo. JIT compiler and runtime for a toy language, using Cranelift

link_cuda_kernel. HowTo: Compile CUDA with nvcc, and link to Rust

rust-cuda. Ecosystem of libraries and tools for writing and executing fast GPU code fully in Rust.

ocl. OpenCL for Rust

inpla. Inpla: Interaction nets as a programming language (the current version)

stable-dreamfusion. Text-to-3D & Image-to-3D & Mesh Exportation with NeRF + Diffusion.

wyhash. The FASTEST QUALITY hash function, random number generators (PRNG) and hash map.

lean4. Lean 4 programming language and theorem prover

stable-diffusion-webui. Stable Diffusion web UI

pixelpusher. pushin' pix

stable-diffusion. A latent text-to-image diffusion model

cooltt. 😎TT

gpt-neox. An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries

stable-diffusion. Jupyter Notebook

cicada-solo. Cicada Language (solo version)