API Reference faber_tweann v#2.4.0

Actuator process for TWEANN networks.

Brain process for real-time neural network inference.

Brain learner process for weight adaptation via plasticity.

Internal pub/sub for brain subsystem communication using pg.

Supervisor for brain system components.

Brain system facade - unified API for neural network inference and learning.

Phenotype constructor for TWEANN networks.

Neural network coordinator for TWEANN networks.

Genetic crossover (recombination) for neural networks.

Exoself - Neural network lifecycle manager

High-performance Rust NIFs for faber_tweann.

Application behaviour for faber_tweann.

Supervisor for faber_tweann application.

Fitness postprocessing for multi-objective optimization.

Flatland: a minimal 2D artificial-life world engine (pure, process-free).

Activation and utility functions for neural computation.

NEAT-style crossover for variable topology neural networks.

Genetic mutation operators for neural network evolution.

Genotype representation for TWEANN networks.

Canonical, lossless serialisation of a genotype, so an evolved topology can leave the VM it was bred in.

The genotype layer's own random generator, so a run can be a function of its seed and a library draw can never perturb a caller's stream.

Convert an evolved genotype into the flat node list the DAG evaluator takes, so an arbitrary evolved topology can be flown at inference speed.

Convert an evolved genotype into a network_evaluator network, carrying the weights, or REFUSE when the topology cannot be represented.

Innovation number tracking for NEAT-style evolution.

Liquid Time-Constant (LTC) neural dynamics.

LTC (Liquid Time-Constant) mutation operators for neural network evolution.

Morphology module for sensor/actuator specifications.

Behaviour for implementing morphologies (sensor/actuator specifications).

Registry for morphology modules.

Self-adaptive (mu, lambda) Evolution Strategy over a real-valued vector.

Multi-cue T-maze: a memory-CAPACITY benchmark (two held bits).

Helper functions for genome mutation operations.

Network compiler for NIF-accelerated evaluation.

Synchronous neural network evaluator for inference.

ONNX export for neural networks.

Internal pub/sub for neural network component communication using pg.

Neural processing unit for TWEANN networks.

Neuron Introspection API.

Liquid Time-Constant (LTC) neural processing unit for TWEANN networks.

Parametric mutation operators for neural network evolution.

Cart-pole balancing scape: single and double pole, with or without velocity.

Weight perturbation utilities for neural network evolution.

Plasticity behavior module - defines the interface for learning rules.

Basic Hebbian plasticity rule implementation.

Reward-modulated Hebbian plasticity rule.

No-op plasticity rule (static weights).

Population-level evolutionary process manager.

Probabilistic reversal bandit: the FAIR lifetime-learning contest.

Reversal bandit: a lifetime-LEARNING benchmark (learning to learn).

Robo Rumble: the five scripted opponents that fix the competence floor.

Robo Rumble: a genome on the wire, and the limits a host enforces on one.

Robo Rumble: the match runner, the scoring rule and the start ensemble.

Robo Rumble: the deterministic integer forward pass (pure, table-driven).

Robo Rumble: the controller contract. 17 sensor channels in, 5 intents out.

Robo Rumble: a battle between N entrants, on one machine, publishable.

Robo Rumble: a deterministic 2D tank-arena engine (pure, process-free).

Robo Rumble: where a match begins. The start set is a rule of the game.

The scape: the environment a neural network is evaluated in.

Selection algorithms for evolutionary processes.

Selection utilities for evolutionary algorithms.

Sensor process for TWEANN networks.

Separable (diagonal) CMA-ES over a real-valued vector.

Signal aggregation functions for neural computation.

Species identification and behavioral fingerprinting.

Cue-memory T-maze: a clean memory benchmark solvable in few generations.

Topological mutation operators for neural network evolution.

Tuning duration: how many memetic attempts an agent gets per evaluation.

Tuning selection: which neurons to perturb each memetic attempt.

Logging infrastructure for faber-tweann.

Native Implemented Functions for high-performance network evaluation.

Pure Erlang fallback implementations for TWEANN NIFs.

XOR scape. The module examples/xor/src/morphology_xor.erl has declared scape = {private, xor_sim} since the repository was created, and until now no such module existed.