QwenGenesis-4B

Artificial Life as Neural Computation
Created by Şuayp Talha Kocabay · @suayptalha

QwenGenesis explores deterministic artificial-life dynamics computed by compiled neural weights. The project demonstrates birth, predation, mutation, ecological collapse and evolutionary selection on a spatial cellular world.

Watch the three worlds

Evolution	Ecological Collapse	Predator Apocalypse
Golden Evolution	Ecological Collapse	Predator Apocalypse

What is included

• model.safetensors: single container for the original Qwen3.5-4B tensor keys plus compiled Genesis circuit tensors under genesis_circuit.*.
• qwengenesis/: deterministic world, compiled circuit, CUDA inference and cinematic renderer.
• qwengenesis_native/: Qwen-style SwiGLU implementation and comparison tests (when included).
• scripts/, tests/: reproducibility and scenario comparisons.
• videos/: three demonstration MP4s.

The original Qwen3.5 tensor names and values are preserved without replacement. Genesis tensors use separate names to avoid key collisions. The Genesis ecosystem is run by the supplied Python code. The base model is not itself a trained or verified QwenGenesis-native decoder. Text-only AutoModel inference does not itself advance the ecosystem.

Reproduce

pip install torch numpy safetensors pillow imageio imageio-ffmpeg transformers
# Run from the cloned/downloaded repo directory:
python run_checkpoint.py --checkpoint model.safetensors --steps 10 --seed 93 --size 24
python -m scripts.run_native_scenarios --device cuda --only all --world-size 96 --output-root outputs_native

The comparison scripts and GPU outputs allow independent verification. The full pretrained decoder was not parity-verified.

Verified results from the supplied run

• SwiGLU circuit vs reference: bit-state agreement over 3 × 100 steps (24×24), maximum reported absolute floating-point difference ~7.13×10⁻¹⁵.
• Full-sized Qwen3.5 FFN comparison on A100: 3 × 10 ticks with binary-state agreement.
• 96×96 demonstration scenarios: predator_apocalypse (70 steps), ecological_collapse (150 steps), golden_evolution (180 steps); all reported output frames matched the reference binary states.
• Illustrative predator trajectory (apocalypse): 2,795 at step 10, 1,523 at step 20, 91 at step 30, 0 at step 40.

Visual motion between simulated states is interpolated by the renderer. Video appearance is illustrative and does not constitute a measurement of independent individual movement.

Attribution

Qwen3.5-4B: Qwen, Apache-2.0.
QwenGenesis ecosystem, experiments and demos: Şuayp Talha Kocabay.