Matěj Kripner

flow_matching. A PyTorch library for implementing flow matching algorithms, featuring continuous and discrete flow matching implementations. It includes practical examples for both text and image modalities.

AverageJoe. The first superhuman generals.io bot, trained from scratch with self-play reinforcement learning.

openpi. Python

opentofu. OpenTofu lets you declaratively manage your cloud infrastructure.

rustbpe. The missing tiktoken training code

pi. AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI

ProofNet. Port of the ProofNet formal theorem proving benchmark to Lean 4.27.0

sbx. SBX: Stable Baselines Jax (SB3 + Jax) RL algorithms

rsl_rl. A fast and simple implementation of learning algorithms for robotics.

IMOLean. Suggested conventions and examples for Lean formalization of IMO problem statements

IsaacLab. Unified framework for robot learning built on NVIDIA Isaac Sim

openprover. Automated Theorem Prover inspired by Aletheia. Claude Code for mathematicians.

envpool. C++-based high-performance parallel environment execution engine (vectorized env) for general RL environments.

newton. An open-source, GPU-accelerated physics simulation engine built upon NVIDIA Warp, specifically targeting roboticists and simulation researchers.

cosmos-predict1. Cosmos-Predict1 is a collection of general-purpose world foundation models for Physical AI that can be fine-tuned into customized world models for downstream applications.

nanoclaw. A lightweight alternative to OpenClaw that runs in containers for security. Connects to WhatsApp, Telegram, Slack, Discord, Gmail and other messaging apps,, has memory, scheduled jobs, and runs directly on Anthropic's Agents SDK

autoresearch. AI agents running research on single-GPU nanochat training automatically

ClickHouse. ClickHouse® is a real-time analytics database management system

cslib. The Lean Computer Science Library (CSLib)

numina-lean-agent. Python

PutnamBench. An evaluation benchmark for undergraduate competition math in Lean4, Isabelle, Coq, and natural language.

mlflow. The open source AI engineering platform for agents, LLMs, and ML models. MLflow enables teams of all sizes to debug, evaluate, monitor, and optimize production-quality AI applications while controlling costs and managing access to models and data.

QED-Nano. Training tiny models to prove hard theorems

goodseed. ML experiment tracker. Simple alternative for neptune.ai or Tensorboard.

htps. HTPS go brr

mini-sglang. A compact implementation of SGLang, designed to demystify the complexities of modern LLM serving systems.

sglang. SGLang is a high-performance serving framework for large language models and multimodal models.

superhuman. Lean

erdosproblems. A community database for the problems on the erdosproblems.com site

minnt. Minimalistic Neural Network Trainer for PyTorch inspired by Keras

nanoproof. Minimal open-source implementation of AlphaProof and HyperTree Proof Search.

aim. Aim 💫 — An easy-to-use & supercharged open-source experiment tracker.

memray. Memray is a memory profiler for Python

pympler. Development tool to measure, monitor and analyze the memory behavior of Python objects in a running Python application.

ppo-implementation-details. The source code for the blog post The 37 Implementation Details of Proximal Policy Optimization

sam2. The repository provides code for running inference with the Meta Segment Anything Model 2 (SAM 2), links for downloading the trained model checkpoints, and example notebooks that show how to use the model.

nano-trm. An implementation of Tiny Recursive Models (TRM)

miniF2F. A fork of openai/miniF2F adapted to Lean 4, with corrections to formalizations and informal descriptions. for human readers.

termplotlib. :chart_with_upwards_trend: Plotting on the command line

leantree. Structured data extraction and programmatic interaction with Lean 4.

VARC. Python

HolisticTraceAnalysis. A library to analyze PyTorch traces.

mup. maximal update parametrization (µP)

recurrent_arc25. Python

nle. The NetHack Learning Environment

PufferLib. Puffing up reinforcement learning

bsuite. bsuite is a collection of carefully-designed experiments that investigate core capabilities of a reinforcement learning (RL) agent

procgen. Procgen Benchmark: Procedurally-Generated Game-Like Gym-Environments

open-genie. Pytorch implementation of "Genie: Generative Interactive Environments", Bruce et al. (2024).

ReProver. Retrieval-Augmented Theorem Provers for Lean

triton. Development repository for the Triton language and compiler

monitorch. A plug-and-use python module to monitor neural network learning.

adam-atan2-pytorch. Implementation of the proposed Adam-atan2 from Google Deepmind in Pytorch

MOSE-api. [ICCV 2023] MOSE: A New Dataset for Video Object Segmentation in Complex Scenes

pelican. Static site generator that supports Markdown and reST syntax. Powered by Python.

openevolve. Open-source implementation of AlphaEvolve

hierarchical-reasoning-model-analysis. Python

TinyRecursiveModels. Python

nanochat. The best ChatGPT that $100 can buy.

lovely-tensors. Tensors, for human consumption

ec. Slash

ark_framework. The Ark framework.

rendergit. Render any git repo into a single static HTML page for humans or LLMs

formal-conjectures. A collection of formalized statements of conjectures in Lean.

lpn. Latent Program Network (from the "Searching Latent Program Spaces" paper)

pypdf. A pure-python PDF library capable of splitting, merging, cropping, and transforming the pages of PDF files

py-spy. Sampling profiler for Python programs

clean. Lean circuit DSL

LeanInteract. LeanInteract: A Python Interface for Lean 4

ts-fsrs. ts-fsrs is a versatile package written in TypeScript that supports ES modules, CommonJS, and UMD.

repl. A simple REPL for Lean 4, returning information about errors and sorries.

proof-wala. Library for interaction with proof-environments, proof search and training of transformers for proof-step prediction

itp-interface. Generic interface for hooking up to any Interactive Theorem Prover (ITP) and collecting data for training ML models for AI in formal theorem proving.

rllm. Democratizing Reinforcement Learning for LLMs

alectryon. A collection of tools for writing technical documents that mix Rocq code and prose.

logical_verification_2023. Hitchhiker's Guide to Logical Verification (2023 Edition)

paperproof. Lean theorem proving interface which feels like pen-and-paper proofs.

lean_proof_recording. Proof recording for Lean 3

Pantograph. (Mirror) A Machine-to-Machine Interaction System for Lean 4

PyPantograph. A Machine-to-Machine Interaction System for Lean 4.

lean-training-data. Lean

generals-bots. Develop your agent for generals.io!

Bend. A massively parallel, high-level programming language

HVM2. A massively parallel, optimal functional runtime in Rust

alacritty. A cross-platform, OpenGL terminal emulator.

litellm. The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]

mathlib4. The math library of Lean 4

OpenHands. 🙌 OpenHands: AI-Driven Development

python-fire. Python Fire is a library for automatically generating command line interfaces (CLIs) from absolutely any Python object.

autogen. A programming framework for agentic AI

supervision. We write your reusable computer vision tools. 💜

awesome-RLHF. A curated list of reinforcement learning with human feedback resources (continually updated)

trl. Train transformer language models with reinforcement learning.

Implicit-Language-Q-Learning. Official code from the paper "Offline RL for Natural Language Generation with Implicit Language Q Learning"

trlx. A repo for distributed training of language models with Reinforcement Learning via Human Feedback (RLHF)

open_clip. An open source implementation of CLIP.

RL4LMs. A modular RL library to fine-tune language models to human preferences

vllm. A high-throughput and memory-efficient inference and serving engine for LLMs

bertviz. BertViz: Visualize Attention in Transformer Models

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

xformers. Hackable and optimized Transformers building blocks, supporting a composable construction.

betaflight. Open Source Flight Controller Firmware

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

postprocessing. A post processing library for three.js.

text-generation-inference. Large Language Model Text Generation Inference

longformer. Longformer: The Long-Document Transformer

mistral-inference. Official inference library for Mistral models

model-catalog. A collection of standardized JSON descriptors for Large Language Model (LLM) files.

faiss. A library for efficient similarity search and clustering of dense vectors.

awesome-openai-vision-api-experiments. Must-have resource for anyone who wants to experiment with and build on the OpenAI vision API 🔥

SummEval. Resources for the "SummEval: Re-evaluating Summarization Evaluation" paper

MINERVA. Meandering In Networks of Entities to Reach Verisimilar Answers

codellama. Inference code for CodeLlama models

elasticsearch. Free and Open Source, Distributed, RESTful Search Engine

lama. 🦙 LaMa Image Inpainting, Resolution-robust Large Mask Inpainting with Fourier Convolutions, WACV 2022

openai-cookbook. Examples and guides for using the OpenAI API

EGG. EGG: Emergence of lanGuage in Games

detectron2. Detectron2 is a platform for object detection, segmentation and other visual recognition tasks.

Awesome-Diffusion-Models. A collection of resources and papers on Diffusion Models

langchain. The agent engineering platform.

makemore. An autoregressive character-level language model for making more things

cs11dotnet7. Repository for the Packt Publishing book titled "C# 11 and .NET 7 - Modern Cross-Platform Development Fundamentals" by Mark J. Price

alzaergo. C++

arxiv-sanity-preserver. Web interface for browsing, search and filtering recent arxiv submissions

npfl114. Materials for the Deep Learning -- ÚFAL course NPFL114

4PrivacyEngine-Core. C++ implementation of the 4PrivacyEngine core

MinAtar. Python

videos. Source code to my videos.

pastaroids. Asteroids, but with pasta

slama.dev. The source code to my personal website.

manim. Animation engine for explanatory math videos

perun. [WIP] Modern compiled low-level programming language in development

scheme. Tiny, readable Scheme interpreter

gitignore. A collection of useful .gitignore templates

MPS. JetBrains Meta programming System

flutter_architecture_samples. TodoMVC for Flutter

rich_editor. This is an RichEditor for the Flutter platform.

example-code. Example code for the book Fluent Python, 1st Edition (O'Reilly, 2015)

flutter. Flutter makes it easy and fast to build beautiful apps for mobile and beyond

WebChecker. Library for Java, which let you manipulate with web pages from outside.