Topics / #optimization

Optimization projects on GitHub

Performance tuning and mathematical optimization

As of , the most-starred Optimization projects on GitHub are annotated_deep_learning_paper_implementations (67.6K stars), winutil (64.4K stars) and dive (54.6K stars). GitHub.Locker tracks 74 Optimization repositories with 938.6K stars combined. The fastest-growing on October 16 was solo_jev, with +149 stars in a day.

74Projects
938.6KStars in total
+797Stars in a day
Browse in the catalog →github.com/topics/optimization ↗

Hot right now

Most stars gained in a day · Oct 16

+149

solo_jev

0814wdwd/solo_jev · ★ 622

SOLO reorders the rows and fields of a dataset so an LLM decision backend with prefix caching can reuse computation across requests, running a natural-language decision over every row while returning results in the original row order.

+106

heretic

p-e-w/heretic · ★ 35.6K

Heretic is a command-line tool that automatically removes censorship (safety alignment) from transformer-based language models using directional ablation combined with an Optuna-powered parameter optimizer, without any training.

+90

Winhance

memstechtips/Winhance · ★ 13.8K

Winhance is a free Windows enhancement utility that debloats, optimizes and customizes Windows 10 and 11, combining app management, privacy and performance tweaks, and theming in one GUI app.

+74

winutil

ChrisTitusTech/winutil · ★ 64.4K

Chris Titus Tech's Windows Utility (WinUtil) is a curated compilation of Windows system tasks: it streamlines installs, debloats with tweaks, troubleshoots, and configures Windows updates.

The all-time top

74 projects · by stars

  1. #1 · labmlai/annotated_deep_learning_paper_implementations

    annotated_deep_learning_paper_implementations

    A collection of 60+ PyTorch implementations of deep learning papers, each annotated with side-by-side explanatory notes so you can read the code and the math/explanation together.

    67.6K
  2. #2 · ChrisTitusTech/winutil

    winutil

    Chris Titus Tech's Windows Utility (WinUtil) is a curated compilation of Windows system tasks: it streamlines installs, debloats with tweaks, troubleshoots, and configures Windows updates.

    64.4K
  3. #3 · wagoodman/dive

    dive

    dive is a CLI tool that breaks a Docker or OCI image into its layers, shows what each layer contains, and points out where the image wastes space so you can shrink it.

    54.6K
  1. 4

    ray

    44.1Kray-project/ray · Python

    Ray is an open-source framework for scaling AI and Python applications: a core distributed runtime plus a set of AI libraries that simplify ML compute.

  2. 5

    llvm-project

    41.1Kllvm/llvm-project · LLVM

    LLVM is an open-source toolkit for constructing highly optimized compilers, optimizers, and run-time environments; this repository holds its source code.

  3. 6

    dspy

    38.7Kstanfordnlp/dspy · Python

    DSPy is a Python framework for programming — rather than prompting — language models: you build modular AI systems in code, and its algorithms automatically optimize the prompts and weights.

  4. 7

    heretic

    35.6Kp-e-w/heretic · Python

    Heretic is a command-line tool that automatically removes censorship (safety alignment) from transformer-based language models using directional ablation combined with an Optuna-powered parameter optimizer, without any training.

  5. 8

    czkawka

    34.3Kqarmin/czkawka · Fluent

    Czkawka is a multi-functional Rust app for finding and cleaning duplicate files, empty folders, similar images/videos/music, broken files, temporary files and more on your disk.

  6. 9

    tuning_playbook

    30.4Kgoogle-research/tuning_playbook

    A free, in-depth guide from five Google Research Brain Team researchers (plus a Harvard co-author) on systematically maximizing the performance of deep learning models — the 'Deep Learning Tuning Playbook'.

  7. 10

    fasthttp

    23.5Kvalyala/fasthttp · Go

    FastHTTP is a fast HTTP server and client library for Go, tuned for high performance: zero memory allocations in hot paths and up to 10x the speed of net/http.

  8. 11

    slim

    23.4Kslimtoolkit/slim · Go

    SlimToolkit (formerly DockerSlim) is a free, open-source CLI toolkit that inspects, optimizes and debugs container images. It can shrink an image by up to 30x without you changing anything in your Dockerfile — and makes the result more secure too.

  9. 12

    svgo

    22.7Ksvg/svgo · JavaScript

    SVGO (SVG Optimizer) is a Node.js library and command-line application that optimizes SVG files — cutting away redundant information so they weigh less while rendering the same.

  10. 13

    react-scan

    21.9Kaidenybai/react-scan · TypeScript

    React Scan is a dev tool that automatically detects performance issues in your React app: it highlights exactly the components you need to optimize, requires no code changes, and stays accessible through a toolbar on the page.

  11. 14

    Atlas

    21.9KAtlas-OS/Atlas · Batchfile

    Atlas is an open-source modification for Windows that optimizes an existing install for performance, privacy and usability: it strips telemetry, debloats the system and applies tweaks — no custom ISO involved.

  12. 15

    SkillOpt

    18.5Kmicrosoft/SkillOpt · Python

    A text-space optimizer that trains reusable natural-language skills for frozen LLM agents without touching model weights. The trained skill ships as a compact markdown file (typically 300–2,000 tokens) that adds zero inference-time model calls at deployment.

  13. 16

    DeepLearning

    17.8KMikoto10032/DeepLearning · Jupyter Notebook

    A Chinese-language deep learning tutorial — a huge curated collection of links to courses, books, lecture notes, surveys and code, taking you from math basics through CNNs, GANs, RNNs and GNNs to Kaggle practice and interview prep.

  14. 17

    million

    17.7Kaidenybai/million · TypeScript

    Million.js is an extremely fast and lightweight optimizing compiler for React that makes components up to 70% faster by speeding up reconciliation.

  15. 18

    Front-End-Performance-Checklist

    17.4Kthedaviddias/Front-End-Performance-Checklist

    It's an exhaustive performance checklist for front-end developers: things you should check, or at least be aware of, when building a website — for personal and professional projects alike.

  16. 19

    codon

    16.9Kexaloop/codon · Python

    Codon is a high-performance Python implementation that compiles your code straight to native machine code — typically 10-100x faster than vanilla Python on a single thread, on par with C/C++.

  17. 20

    Modern-CPP-Programming

    16.2Kfederico-busato/Modern-CPP-Programming · HTML

    A free, open-access Modern C++ Programming course by Federico Busato covering standards from C++03 to C++26 — from language basics to advanced features, tools and optimization.

  18. 21

    Nuitka

    15.2KNuitka/Nuitka · Python

    Nuitka is a Python compiler, written in Python itself, that translates Python code into C and then compiles it into an executable or extension module.

  19. 22

    scipy

    15.1Kscipy/scipy · Python

    SciPy (pronounced 'Sigh Pie') is an open-source Python library for mathematics, science, and engineering.

  20. 23

    ants

    14.5Kpanjf2000/ants · Go

    ants is a goroutine pool library for Go: it manages and recycles a massive number of goroutines with a fixed capacity, letting you limit how many run concurrently in your program.

  21. 24

    or-tools

    14.2Kgoogle/or-tools · C++

    OR-Tools is Google's open-source software suite for combinatorial optimization: solvers and algorithms for problems like scheduling, routing, packing, and resource allocation.

  22. 25

    prepack

    14.1Kfacebookarchive/prepack · JavaScript

    Prepack is a partial evaluator for JavaScript that rewrites a JS bundle into code that executes more efficiently, especially useful for initialization-heavy code.

  23. 26

    Winhance

    13.8Kmemstechtips/Winhance · C#

    Winhance is a free Windows enhancement utility that debloats, optimizes and customizes Windows 10 and 11, combining app management, privacy and performance tweaks, and theming in one GUI app.

  24. 27

    primitive

    13.5Kfogleman/primitive · Go

    It recreates a target image out of geometric primitives: the algorithm keeps adding one shape at a time, each chosen to minimize the difference from the original, until the result is recognizable yet artistic and abstract.

  25. 28

    scalene

    13.5Kplasma-umass/scalene · Python

    Scalene is a high-performance CPU, GPU, and memory profiler for Python. It is also the first profiler to add AI-powered optimization proposals, via providers like OpenAI, Azure, Bedrock, or local Ollama models.

  26. 29

    UglifyJS

    13.4Kmishoo/UglifyJS · JavaScript

    UglifyJS is a toolkit for JavaScript that parses, minifies, compresses, mangles, and beautifies code, mainly used to shrink JS files for production.

  27. 30

    deep-learning-drizzle

    13Kkmario23/deep-learning-drizzle · HTML

    A curated index that collects links to full video lectures and course materials for Deep Learning, Reinforcement Learning, Machine Learning, Computer Vision and NLP courses taught at top universities worldwide.

  28. 31

    VasSonic

    11.8KTencent/VasSonic · Java

    VasSonic (Sonic for short) is a lightweight, high-performance hybrid framework from Tencent's VAS team that speeds up the first screen of websites loaded in hybrid apps on Android and iOS.

  29. 32

    BenchmarkDotNet

    11.5Kdotnet/BenchmarkDotNet · C#

    BenchmarkDotNet is a .NET library that turns ordinary methods into reliable, repeatable performance benchmarks and reports detailed statistics about the results.

  30. 33

    STL

    11.2Kmicrosoft/STL · C++

    Microsoft's official implementation of the C++ Standard Library (the STL), which ships as part of the MSVC toolset and the Visual Studio IDE. This repo hosts its source code and development.

  31. 34

    go-perfbook

    10.9Kdgryski/go-perfbook

    It is a free, work-in-progress book documenting best practices for writing high-performance Go code, covering both general optimization techniques and Go-specific ones.

  32. 35

    thumbor

    10.5Kthumbor/thumbor · Python

    Thumbor is an open-source HTTP image service that crops, resizes, transforms and optimizes photos on-demand, using AI for smart face/feature-based cropping.

  33. 36

    optimizerDuck

    10.5Kitsfatduck/optimizerDuck · C#

    A free, open-source Windows optimization app that gathers performance tweaks, privacy settings, bloatware removal and cleanup tools in one place, so you don't have to dig through the Registry, Group Policy or PowerShell.

  34. 37

    critical

    10.3Kaddyosmani/critical · JavaScript

    It extracts the above-the-fold (critical-path) CSS from your HTML pages, inlines it into the document, and loads the rest asynchronously so the browser can paint the first screen without waiting for the full stylesheet.

  35. 38

    tpot

    10.1KEpistasisLab/tpot · Jupyter Notebook

    TPOT, the Tree-based Pipeline Optimization Tool, is a Python AutoML library that uses genetic programming to automatically optimize machine learning pipelines for your data; the README calls it your Data Science Assistant.

  36. 39

    pi-autoresearch

    8.2Kdavebcn87/pi-autoresearch · TypeScript

    pi-autoresearch is an extension for pi, a terminal-based AI coding agent, that runs autonomous optimization loops: it tries a code change, benchmarks it, keeps the change if it's an improvement, reverts it if it isn't, and repeats indefinitely.

  37. 40

    closure-compiler

    7.7Kgoogle/closure-compiler · JavaScript

    A tool for making JavaScript download and run faster — a true compiler for JavaScript: it compiles JavaScript to better JavaScript, parsing your code, removing dead code and rewriting and minimizing what is left.

  38. 41

    pxpipe

    7.5Kteamchong/pxpipe · TypeScript

    pxpipe is a local proxy that cuts Claude Code's (and other LLM clients') input token usage by rendering bulky context — system prompts, tool docs, and conversation history — as compact PNG images instead of raw text, since an image's token cost depends on its pixel size, not the amount of text packed into it.

  39. 42

    ANE

    7.3Kmaderix/ANE · Objective-C

    ANE is a research project that trains neural networks directly on Apple's Neural Engine using reverse-engineered private APIs, instead of CoreML's inference-only path.

  40. 43

    turboquant_plus

    7KTheTom/turboquant_plus · Python

    TurboQuant+ is the research home for TurboQuant (ICLR 2026), a KV cache compression method for transformer LLMs. It shrinks the KV cache 3.8-6.4x using PolarQuant plus Walsh-Hadamard rotation, and this repo holds the Python reference implementation, validation papers and benchmark data.

  41. 44

    darwin-skill

    6.4Kalchaincyf/darwin-skill · HTML

    Darwin.skill is a system that evolves Claude Code Agent Skills through an automated evaluate-improve-test-keep-or-revert loop, inspired by Andrej Karpathy's autoresearch project.

  42. 45

    xquad

    5.6KQuipNetwork/xquad · Python

    XQuad is a hardware-agnostic toolchain for expressing and running quadratic optimization problems (QUBO / Ising / discrete) across quantum annealers and classical solvers, acting as a common intermediate representation you write once and retarget to any supported backend.

  43. 46

    hermes-agent-self-evolution

    5.5KNousResearch/hermes-agent-self-evolution · Python

    It is an evolutionary self-improvement system for Hermes Agent that uses DSPy and GEPA (Genetic-Pareto Prompt Evolution) to automatically optimize the agent's skills, tool descriptions, system prompts, and code, producing measurably better versions through reflective evolutionary search.

  44. 47

    original_performance_takehome

    4.2Kanthropics/original_performance_takehome · Python

    This is Anthropic's original performance take-home interview challenge, open-sourced for anyone to try. The goal is to optimize a simulated-machine program so it runs in as few clock cycles as possible, starting from a slow baseline of 18532 cycles.

  45. 48

    TurboDiffusion

    3.9Kthu-ml/TurboDiffusion · Python

    TurboDiffusion is a video generation acceleration framework that speeds up end-to-end video diffusion generation by 100-200x on a single RTX 5090 while keeping video quality, using SageAttention, Sparse-Linear Attention (SLA), and rCM timestep distillation.

  46. 49

    TOTK-Mods-collection

    3.1Khoverbike1/TOTK-Mods-collection · Shell

    A community repository of mods for The Legend of Zelda: Tears of the Kingdom (TOTK), put together for playing the game on a real Switch or in Switch emulators.

  47. 50

    ReBarUEFI

    2.6KxCuri0/ReBarUEFI · C

    ReBarUEFI is a UEFI DXE driver that enables Resizable BAR on systems whose firmware doesn't support it officially — a free performance boost, mostly noticeable in games.

  48. 51

    spinel

    2.2Kmatz/spinel · C

    Spinel is a Ruby ahead-of-time compiler that turns Ruby source code into standalone native executables by inferring types across the whole program and generating optimized C code.

  49. 52

    dqn-adaptive-cpu-scheduler

    1.5Kaaawangyh/dqn-adaptive-cpu-scheduler · Python

    This is a research-grade repository that implements an adaptive CPU task scheduler trained with a Deep Q-Network (DQN). It includes a simulator of multi-core CPUs with DVFS and a thermal-aware reward, plus training, evaluation, and plotting pipelines.

  50. 53

    hexstellar

    1.4Kbrayonpi/hexstellar · Python

    HexStellar lets an AI agent hand off hard computational problems — optimization, quantum computing, scientific computing, and decision intelligence — to a managed service called Cortex, which executes them and returns structured, verifiable results with certainty labels and receipts.

  51. 54

    AntOmniEvo

    1.2Kant-research/AntOmniEvo · Python

    AntOmniEvo is an auto-evolution framework: you give it your system's tunable artifacts (skills, prompts, workflow configs, or code) plus a way to score them, and it runs a concurrent evolution loop where a coding-agent 'Proposer' rewrites those files to produce better-performing candidates.

  52. 55

    cpu-performance-engineering

    1.2Kusamahz/cpu-performance-engineering · Python

    A curated reading path for CPU performance engineering: it walks you from what one instruction does on a modern core to serving ML inference on CPUs, ordered so each piece makes the next one legible.

  53. 56

    halo

    825whitecircle/halo · Python

    Halo is an open-source framework from White Circle for training large language and multimodal models: pre-training, SFT, preference optimization, and fully asynchronous multi-turn RL in one codebase, from a single GPU up to multi-node clusters.

  54. 57

    solo_jev

    6220814wdwd/solo_jev · Python

    SOLO reorders the rows and fields of a dataset so an LLM decision backend with prefix caching can reuse computation across requests, running a natural-language decision over every row while returning results in the original row order.

  55. 58

    DeepGEMM-Ascend

    583deepseek-ai/DeepGEMM-Ascend · C++

    DeepGEMM-Ascend is a port of DeepSeek's DeepGEMM matrix-multiplication kernel library to Huawei Ascend NPUs, fully API-compatible with the original and supporting BF16, FP8, and FP4 GEMM along with MQA logits and MegaMoE operators.

  56. 59

    Jet

    533CubeCoders/Jet · C++

    Jet is a tiny, dependency-free, fixed-function 3D software rasteriser in C++17, built for embedded chips with little CPU and RAM like the ESP32 and STM32; it also runs on desktop PCs and Raspberry Pi.

  57. 60

    ir

    509dstogov/ir · C

    IR is a lightweight framework for building JIT compilers in medium-size projects. It gives you an intermediate representation, an API to construct it, and passes for optimization, scheduling, register allocation and machine code generation that can be executed directly in memory.

  58. 61

    mcopt

    471noahdunnagan/mcopt · Java

    mcopt is a Minecraft mod that adds a native Metal renderer for Minecraft: Java Edition, aimed specifically at making the game run faster on Apple Silicon Macs.

  59. 62

    rebalancer

    461facebook/rebalancer · C++

    Rebalancer is a domain-specific language and library for defining and solving assignment problems, like putting objects into containers under constraints and goals.

  60. 63

    benjamin-plus-skill

    330JetBrains/benjamin-plus-skill · Shell

    Benjamin-Plus is a token-efficiency skill for coding agents: it teaches five habits around looking things up and waiting (not what the agent builds), measured to cut cost by up to 18% with quality unchanged.

  61. 64

    EvoVLM

    319zzzz7788990213-ops/EvoVLM · Python

    EvoVLM is a research project that runs LLM-guided, multi-objective evolutionary search over vision-language inference programs. It targets the accuracy-latency trade-off on scientific chart understanding while keeping model weights frozen.

  62. 65

    imp

    254deepfates/imp · Elixir

    Imp is a full port of DSPy to the BEAM: an Elixir framework for building and automatically improving language-model programs. You declare a step's inputs and outputs as a typed signature, and an optimizer tunes its prompts or examples against labeled data.

  63. 66

    ComfyUI-H3VAE_TRT

    254lihaoyun6/ComfyUI-H3VAE_TRT · Python

    A ComfyUI custom node that runs the MiniMax-H3 VAE as a compiled TensorRT engine instead of the regular one, making it up to 1.7x faster.

  64. 67

    Hyperloom

    241AMD-AGI/Hyperloom · Python

    A multi-agent harness that autonomously optimizes LLM inference on AMD Instinct GPUs: it profiles your workload, searches framework and GPU-kernel optimizations, validates every candidate end to end, and saves proven results into a recipe knowledge base.

  65. 68

    Game-Optimizer-CPUs-Threads-Optimizer

    211charlie754/Game-Optimizer-CPUs-Threads-Optimizer · C++

    Game Optimizer is a Windows tray app that pins a chosen game (and its child processes) to the correct CPU cache complex using the Windows CPU Sets API, instead of using affinity masks, disabling SMT, or changing BIOS settings.

  66. 69

    InstinctFlash

    200General-Instinct/InstinctFlash · Python

    InstinctFlash is a high-performance serving runtime for robotics foundation models - it loads checkpoints like GR00T, Cosmos3, pi0.5, LingBot-VA/VLA and DreamZero, applies acceleration (FP8, fused Triton kernels, caching) and serves them over Python or WebSocket, reporting up to 33.78x speedup on Jetson Thor.

  67. 70

    explaindb

    167explaindb/explaindb · Jupyter Notebook

    ExplainDB is teaching material for a database systems course: a collection of Jupyter notebooks paired with a small Python DBMS (the system/ package) written to be readable, used alongside Prof. Jens Dittrich's 'Database Systems 2024/25' lecture at Saarland University.

  68. 71

    RingOut

    163jackpoison-prog/RingOut · C

    RingOut is a native PC port of a GameCube fighting game (disc ID GRSEAF), made by static recompilation rather than emulation: the original PowerPC executable is translated ahead of time into C, compiled for x86-64, and run as native code inside a Dolphin-derived runtime that handles graphics, audio, input and hardware emulation around it.

  69. 72

    toolrush

    157OnlyTerp/toolrush · Python

    ToolRush is a low-overhead execution layer plugin for Hermes Agent (Nous Research's agent harness) that eliminates the latency overhead of tool calls like file reads, searches, and terminal commands while keeping the same tools, outputs, and safety gates.

  70. 73

    cumcm-c-problem-lfs

    148liufanshan11/cumcm-c-problem-lfs · TeX

    A skill for AI agents that handles Problem C of CUMCM, China's national undergraduate math-modeling contest: a full modeling-and-paper production workflow distilled from 2018–2025 official C-problem papers, expert commentary and grading standards.

  71. 74

    Singeli

    136mlochbaum/Singeli · BQN

    Singeli is a domain-specific language for building high-performance algorithms, including SIMD code. It provides flexible abstractions over code that maps to individual CPU instructions, and it compiles your programs down to C.

Often tagged together