Use ES5 properties with Backbone. Goodbye get() and set()!

Self-improving workflow skills for AI agents: dev, ML experiments, PR review, debugging, ideation, roadmapping, and backprop optimization.

Reflect2Evolve memory plugin: layered L1/L2/L3 memory, reflection-weighted value backprop, cross-task policy induction, skill crystallization, three-tier retrieval. Adapters for OpenClaw and Hermes Agent via a shared algorithm core.

Deterministic 26-rule verifier for neural-network training steps. Re-derives gradients + optimizer state from named factors; emits canonical JSONL. SGD/Adam/AdamW/SGD-momentum (+coupled-L2 weight decay); live PyTorch + JAX helpers; CPU-only.

Creates a neural network and can handle backprop

An optimized implementation of the backpropagation algorithm in Typescript

Forward + reverse-mode automatic differentiation for MathTS rank-N Tensor

Cavekit SPEC.md workflow packaged for Pi

Offline AI assistant for AfterLink — 100% local RAG pipeline with FAISS vector search, BM25, cross-encoder reranking, and an interactive terminal UI

Cavekit spec-driven development skills and spec tracker widget for pi

Bayesian probability transforms for BM25 retrieval scores

Minimal neural network from scratch — neuron, layer, network, backpropagation, classical ML, unsupervised, generative models, autograd. No dependencies.

Simple script tree sequencer in javascript. Allows creating fairly complex trees of sync/async functions or requestAnimationFrame functions with optional delays, as well as subscribing to particular outputs in the tree at any time.

GPU-accelerated Deep Recurrent Neural Networks and LSTMs in Typescript. Ported, object-oriented and refactored version of Andrej Karpathy's recurrent-js (https://github.com/karpathy/recurrentjs)

Neural network library for NodeJS and browser

Rudimentary Neural Network in Typescript