NeuroLink

NeuroLink contains nesylink, RL smoke training helpers, and DreamerV3 experiments.

Project Framework
    nesylink/     - game env and core mechanics (world, collision, rendering, rewards)
    rl/           - training and evaluation pipeline (algorithms, runner, config loading)
    world_model/  - DreamerV3 experiments and scripts
    envs/         - environment variants
    docs/     - project documentation

Quick Start

python -m venv .venv
source .venv/bin/activate
pip install -r requirements/requirements.txt

NesyLink

from nesylink.env import make_env

env = make_env(
    map_id="dungeon",
    reward_id="sparse_exit",
    max_steps=500,
)

This creates a dungeon environment where the agent receives a reward only upon exiting the dungeon. You can customize the map and reward by changing the map_id and reward_id parameters.

for more details on available maps and rewards, see the nesylink directory.