Havuz

This project is for learning: it shows how an agent loop works underneath, in a way more intuitive and visual than reading about it. Behind

A computer, compiled into the weights of a language model. An exact CPU inside Qwen3.5-4B, with Gated DeltaNet state as RAM.

A moddable training framework for pre-training ablations.

Runs Hugging Face ONNX models on any Nx backend, with no per-architecture code

Illustrated comic-style cheat sheets for ML, LLMs, AI safety, LeetCode and Git, plus a SwiftUI study app (Panels) to read them and quiz your

Interactive desktop app with 58 labs covering ML fundamentals, transformers, RAG, agents, fine-tuning and quantization and more.

Guía gratuita de IA generativa avanzada en español (311 págs.): Transformer, post-entrenamiento, inferencia, GPU, difusión, agentes, segurid

Instinct: single-forward-pass decision models (choice / true-false / score) returning a probability per candidate

AnyJev, open LLM’lerden typed kararlar ve eşiklenebilir olasılıklar üretmek için calibration ve karar katmanları sunan bir Python kütüphanesidir.

Neden yükseliyor · Yedi günde 802 yıldız almış; dikkat çekmesinde etiket gerektirmeyen ilk seviye, az etiketle kalibrasyon iddiası ve modeli fine-tune etmeden karar endpoint’i sunma yaklaşımı etkili görünüyor. Bu ilgi benimseme kanıtı değil; verilen sonuçlar README’deki sınırlı benchmark’a dayanıyor.

an open-source, from-first-principles reconstruction of the ideas behind TypeSafe AI's Jev, written in pytorch

Run SemIf (Jev-style semantic-if decisions) on a CPU — no GPU. Reads typed option probabilities straight from an open model in one forward p

Open-source alternative to TypeSafe's Jev: a System One style model layer that gives typed, calibrated decisions from any open-weights LLM i

Research project on bilingual Urdu-English retrieval-augmented generation for Pakistani government and legal documents, with OCR, hybrid ret

[arXiv 2026] Code for "Elastic Token Compression for Pixel-Space Diffusion Transformers"

High-performance CUDA inference engine for Qwen models, written from scratch in C++/CUDA

Spark-optimized MLX LM for local inference and deployment on Apple Silicon

Halo is an open-source framework built by White Circle for training large language and multimodal models

Evaan — a fully local, offline-capable AI companion chatbot built with Python & Hugging Face Transformers, running on Qwen2.5-0.5B-Instruct

Low-memory C++20 LLM inference runtime for quantized GGUF models on CPU — mmap-backed weights, paged KV cache, Q4/Q8 kernels, and reproducib

A native macOS lab for teaching tiny language models to think — build the architecture, train the weights, and watch a small LLM emerge from

Interactive, animated architecture maps for any Hugging Face model — no weights downloaded

Distributed 56M-parameter LLM inference across 3 ESP32-S3 boards via ESP-NOW , Split-PLE + KV cache, fully offline.

Build and train a mini Kimi K3 from scratch — a pure-PyTorch playground from a custom 88M model to 2T-scale LLM profiles.

hummingbird is a lightweight, zero dependency runtime for massive open source Mixture of Experts (MoE) language models. It unifies SSD, RAM,

Interactive 3D visualization platform for exploring transformer architectures, tensors, and real-time LLM inference.

Build LLM systems you actually control. A free, open engineering book + course — from tokenization to serving your own models. Mechanisms, t

A mental-model-first field guide to Audio AI — from raw waveforms to production voice systems. Covers ASR, TTS, audio classification, Transf

Jacobian-Brainwash : A manual alignment tool for large language models built on Anthropic's Jacobian Lens. Results are exportable.

Transformer architecture explained step by step - the full architecture, every attention variant, positional embeddings, and every layer ins

To know what models don't say out loud.

OpenAI-compatible server with a live view into any HF model's residual stream. pip install brainscope

Native AI inference for PHP 8.3+ - run ONNX, GGUF (llama.cpp) and RubixML models directly in your PHP process via FFI. Chat, streaming, embe

Publication-quality LLM architecture figures from config.json — Raschka-style towers, diffs, posters, VRAM math. No weights needed.

🧠 One LLM, split across a Mac (Apple MPS) + a Windows PC (NVIDIA CUDA) — heterogeneous pipeline inference over a framework-neutral wire, bit

Why is LLM inference slow — and how do you make it fast? A hands-on, first-principles course: roofline → KV cache → quantization → paralleli

rizzo-pii, İtalyanca hukuk metinlerindeki kişisel verileri yerel olarak tespit edip geri döndürülebilir placeholder’larla maskeleyen bir yazılımdır.

Neden yükseliyor · 91 günde 1057 yıldız ve son 28 günde 25 commit ile 5 farklı yazar görünür ilgi ve yakın dönem geliştirme faaliyeti gösteriyor; yükselişin özel nedeni mevcut verilerden doğrulanamıyor.

A zero-dependency Python module for inspecting and converting coding-agent session files (.jsonl) — Claude Code, Codex, and Pi — into the me

Seven Minutes Is All I Can Spare To Play With You.

Open benchmark for expert collapse and routing efficiency in sparse MoE LLMs

Models Take Notes at Prefill: KV Cache Can Be Editable and Composable (arXiv:2606.17107) — paper, code, results, and interactive companion.

Mechanistic experiments on how LLMs represent and compute arithmetic internally, with strict no-parser controls, reproducible audits, and an

👻 MoEWatch: A lightweight diagnostic and audit library for Mixture-of-Experts (MoE) models in HuggingFace Transformers — the pytest for MoE

211 free visual cards about large language models, from tokenization to deployment. Full collection: 376 cards.

Train decoder-only language models from scratch with PyTorch, featuring RoPE, RMSNorm, SwiGLU, and SDPA

Custom APRL machine-learning algorithm + a curated ML knowledge base of 923 papers, lectures, and explainers. NumPy classifier/regressor, te

Generative image and video runtimes for Apple Silicon (MLX): text/image-to-image, instruction editing, masked edit, text/image/reference-to-

ML model ports (ASR · TTS · LLM · VLM · vision · audio codecs) on the rlx multi-backend compiler + runtime — CPU/Metal/MLX/CUDA/wgpu

Companion code for the Manning book 'LLM Customization and Fine-Tuning.' Adapt open-weights LLMs end to end: prompting and RAG, LoRA/QLoRA,

Tiny-MoE is a lightweight Mixture-of-Experts language model built entirely from scratch in native PyTorch and trained end-to-end on Kaggle u

Single-GPU LLM fine-tuning preflight: memory estimates, runnable LoRA/QLoRA recipes, and local measurements.