NeuraxonLife2-1M: Artificial Life Neuraxon Neural Network Simulation Dataset

Dataset Description

The NeuraxonLife 2.0 1M Dataset contains detailed simulation data from an artificial life environment where autonomous agents ("NxErs") evolve biologically-plausible Neuraxon neural networks. This dataset captures the complete neural architecture, synaptic connectivity, neuromodulation states, and behavioral performance metrics of evolved artificial organisms.

Update 12/10/25: Added A Full version to include also each full game info captured check NeuraxonLife2-1MFull_manifest.json for details

Dataset Summary

This dataset provides a unique window into how neural networks evolve under survival pressure in a simulated ecosystem. Each NxEr (Neuraxon Entity) is an autonomous agent with:

• A Neuraxon neural network (https://www.researchgate.net/publication/397331336_Neuraxon ) with dendritic computation
• Multi-timescale synaptic plasticity (fast, slow, meta)
• Four neuromodulatory systems (dopamine, serotonin, acetylcholine, norepinephrine)
• Behavioral capabilities (movement, foraging, mating)
• Evolutionary fitness tracking

Supported Tasks

• Neural Architecture Analysis: Study evolved network topologies
• Synaptic Weight Distribution: Analyze learned connection patterns
• Neuromodulation Research: Investigate modulator dynamics
• Fitness Prediction: Predict agent fitness from neural parameters
• Evolutionary Dynamics: Track neural evolution across generations

Dataset Structure

The dataset consists of four interconnected tables stored as separate Parquet files:

/
├── neuraxonLife2-1M_nxers.parquet      # Agent-level data
├── neuraxonLife2-1M_neurons.parquet    # Neuron-level data
├── neuraxonLife2-1M_synapses.parquet   # Synapse-level data
├── neuraxonLife2-1M_branches.parquet   # Dendritic branch data
├── neuraxonLife2-1M_manifest.json      # Dataset metadata
└── README.md                   # This file

Data Tables

1. NxErs Table (neuraxonLife2-1M_nxers.parquet)

Agent-level data containing identity, attributes, neural network parameters, and performance metrics.

Column	Type	Description
Identifiers
game_id	string	Unique game/simulation identifier
nxer_id	int	Agent ID within the game
nxer_name	string	Agent name
Game Context
game_step	int	Current simulation tick
game_births	int	Total births in game
game_deaths	int	Total deaths in game
game_index	int	Game sequence index
World Configuration
NxWorldSize	int	World grid size
NxWorldSea	float	Sea proportion (0-1)
NxWorldRocks	float	Rock proportion (0-1)
MaxFood	int	Maximum food items
MaxNeurons	int	Maximum neurons per agent
Basic Attributes
is_male	int	Gender (1=male, 0=female)
gender	string	"Male" or "Female"
can_land	int	Can traverse land (0/1)
can_sea	int	Can traverse sea (0/1)
terrain	string	"Land", "Sea", or "Amphibious"
alive	int	Alive status (0/1)
food	float	Current food/energy level
Color
color_r	int	Red component (0-255)
color_g	int	Green component (0-255)
color_b	int	Blue component (0-255)
Sensory
vision_range	int	Vision distance in tiles
smell_radius	int	Smell detection radius
heading	int	Current heading direction
clan_id	int	Clan affiliation (-1 if none)
Position
pos_x	int	Current X position
pos_y	int	Current Y position
last_pos_x	int	Previous X position
last_pos_y	int	Previous Y position
Lifecycle
born_ts	float	Birth timestamp
died_ts	float	Death timestamp (0 if alive)
ticks_per_action	int	Action frequency
visited_count	int	Unique positions visited
Behavioral State
is_harvesting	int	Currently harvesting (0/1)
is_mating	int	Currently mating (0/1)
dopamine_boost_ticks	int	Dopamine boost duration
Lineage
has_parents	int	Has known parents (0/1)
parent_count	int	Number of parents
Neural Inputs
last_input_0 to last_input_5	float	Last sensory inputs
last_output_o4	int	Last O4 output
Performance Stats
food_found	float	Total food discovered
food_taken	float	Total food consumed
explored	int	Tiles explored
time_lived	float	Lifetime in seconds
mates	int	Successful matings
energy_eff	float	Energy efficiency score
temporal_sync	float	Temporal synchronization
fitness	float	Overall fitness score
Network Topology
n_input	int	Input neuron count
n_hidden	int	Hidden neuron count
n_output	int	Output neuron count
n_total	int	Total neuron count
n_synapses	int	Total synapse count
conn_density	float	Connection density
conn_prob	float	Connection probability
small_world_k	int	Small-world k parameter
rewire_prob	float	Rewiring probability
pref_attach	int	Preferential attachment (0/1)
max_axon_delay	float	Maximum axonal delay
Network Time
net_dt	float	Simulation timestep
net_min_dt	float	Minimum timestep
net_max_dt	float	Maximum timestep
activity_threshold	float	Activity threshold
Neuron Parameters
membrane_tau	float	Membrane time constant
thresh_exc	float	Excitatory threshold
thresh_inh	float	Inhibitory threshold
adaptation	float	Adaptation rate
spont_rate	float	Spontaneous firing rate
health_decay	float	Health decay rate
Dendritic Parameters
n_branches	int	Branches per neuron
branch_thresh	float	Branch threshold
plateau_decay	float	Plateau decay constant
Synaptic Time Constants
tau_fast	float	Fast synapse tau
tau_slow	float	Slow synapse tau
tau_meta	float	Metaplasticity tau
tau_ltp	float	LTP time constant
tau_ltd	float	LTD time constant
Weight Initialization
w_fast_min/max	float	Fast weight bounds
w_slow_min/max	float	Slow weight bounds
w_meta_min/max	float	Meta weight bounds
Learning & Plasticity
learn_rate	float	Base learning rate
stdp_window	float	STDP window size
plast_thresh	float	Plasticity threshold
assoc_strength	float	Associativity strength
Structural Plasticity
syn_integrity	float	Integrity threshold
syn_form_prob	float	Synapse formation prob
syn_death_prob	float	Synapse death prob
neuron_death	float	Neuron death threshold
Neuromodulation Baselines
da_base	float	Dopamine baseline
ser_base	float	Serotonin baseline
ach_base	float	Acetylcholine baseline
ne_base	float	Norepinephrine baseline
Neuromodulation Thresholds
da_high/low	float	Dopamine thresholds
ser_high/low	float	Serotonin thresholds
ach_high/low	float	Acetylcholine thresholds
ne_high/low	float	Norepinephrine thresholds
neuromod_decay	float	Modulator decay rate
diffusion	float	Diffusion rate
Oscillators
osc_low/mid/high	float	Oscillator frequencies
osc_strength	float	Oscillator strength
phase_coupling	float	Phase coupling strength
Energy Metabolism
energy_base	float	Baseline energy
firing_cost	float	Firing energy cost
plast_cost	float	Plasticity energy cost
metabolic_rate	float	Metabolic rate
recovery_rate	float	Energy recovery rate
Homeostasis
target_fire_rate	float	Target firing rate
homeo_plast_rate	float	Homeostatic plasticity
AIGarth/ITU
itu_radius	int	ITU circle radius
evol_interval	int	Evolution interval
fit_temporal_w	float	Temporal fitness weight
fit_energy_w	float	Energy fitness weight
fit_pattern_w	float	Pattern fitness weight
Current Neuromodulators
curr_da	float	Current dopamine
curr_ser	float	Current serotonin
curr_ach	float	Current acetylcholine
curr_ne	float	Current norepinephrine
Network State
net_time	float	Network simulation time
net_steps	int	Network step count
branching_ratio	float	Criticality measure
energy_consumed	float	Total energy consumed
itu_circle_count	int	ITU circle count

---

2. Neurons Table (neuraxonLife2-1M_neurons.parquet)

Individual neuron data within each agent's neural network.

Column	Type	Description
game_id	string	Game identifier
nxer_name	string	Parent agent name
neuron_id	int	Neuron ID
type	string	Neuron type ("input", "hidden", "output")
type_from_data	string	Type from raw data
Core State
membrane_pot	float	Membrane potential
trinary	int	Trinary state (-1, 0, 1)
trinary_label	string	"Inhibitory", "Neutral", "Excitatory"
adaptation	float	Adaptation level
health	float	Neuron health (0-1)
is_active	int	Active status (0/1)
energy	float	Energy level
Oscillation
phase	float	Current phase
nat_freq	float	Natural frequency
intrinsic_ts	float	Intrinsic timescale
ITU
circle_id	int	ITU circle ID (-1 if none)
neuron_fitness	float	Neuron fitness score
Individual Parameters
ind_membrane_tau	float	Individual membrane tau
ind_thresh_exc	float	Individual excitatory threshold
ind_thresh_inh	float	Individual inhibitory threshold
ind_adaptation	float	Individual adaptation rate
ind_spont_rate	float	Individual spontaneous rate
ind_health_decay	float	Individual health decay
ind_energy_base	float	Individual energy baseline
ind_firing_cost	float	Individual firing cost
ind_plast_cost	float	Individual plasticity cost
ind_metabolic	float	Individual metabolic rate
ind_recovery	float	Individual recovery rate
Dendritic Statistics
n_branches	int	Number of dendritic branches
branch_pot_mean/std/min/max	float	Branch potential statistics
plateau_mean/max	float	Plateau potential statistics
branch_thresh_mean/std	float	Branch threshold statistics
plateau_decay_mean	float	Mean plateau decay

---

3. Synapses Table (neuraxonLife2-1M_synapses.parquet)

Synaptic connection data between neurons.

Column	Type	Description
game_id	string	Game identifier
nxer_name	string	Parent agent name
pre_id	int	Presynaptic neuron ID
post_id	int	Postsynaptic neuron ID
Weights
w_fast	float	Fast synaptic weight
w_slow	float	Slow synaptic weight
w_meta	float	Meta-plasticity weight
w_total	float	w_fast + w_slow
w_abs	float	|w_fast| + |w_slow|
w_fast_abs	float	|w_fast|
w_slow_abs	float	|w_slow|
w_meta_abs	float	|w_meta|
Flags
is_silent	int	Silent synapse (0/1)
is_modulatory	int	Modulatory synapse (0/1)
syn_type	string	Synapse type string
is_ionotropic_fast	int	Fast ionotropic (0/1)
is_ionotropic_slow	int	Slow ionotropic (0/1)
is_metabotropic	int	Metabotropic (0/1)
Properties
integrity	float	Synapse integrity (0-1)
axon_delay	float	Axonal delay
learn_mod	float	Learning rate modifier
delta_w	float	Potential weight change
Individual Time Constants
ind_tau_fast	float	Individual tau fast
ind_tau_slow	float	Individual tau slow
ind_tau_meta	float	Individual tau meta
ind_tau_ltp	float	Individual tau LTP
ind_tau_ltd	float	Individual tau LTD
ind_learn_rate	float	Individual learning rate
ind_plast_thresh	float	Individual plasticity threshold
Derived Metrics
tau_ratio_fast_slow	float	tau_fast / tau_slow
tau_ratio_ltp_ltd	float	tau_ltp / tau_ltd

---

4. Branches Table (neuraxonLife2-1M_branches.parquet)

Dendritic branch data for detailed dendritic computation.

Column	Type	Description
game_id	string	Game identifier
nxer_name	string	Parent agent name
neuron_id	int	Parent neuron ID
branch_id	int	Branch ID
branch_pot	float	Branch potential
branch_pot_abs	float	|branch_pot|
plateau_pot	float	Plateau potential
branch_thresh	float	Branch threshold
plateau_decay	float	Plateau decay constant
above_threshold	int	Above threshold (0/1)
has_plateau	int	Has plateau (0/1)

---

Relationships Between Tables

NxErs (1) ──────┬───────── (N) Neurons
                │
                └───────── (N) Synapses

Neurons (1) ────────────── (N) Branches

• NxErs → Neurons: One NxEr contains multiple neurons (join on game_id + nxer_name)
• NxErs → Synapses: One NxEr contains multiple synapses (join on game_id + nxer_name)
• Neurons → Branches: One neuron contains multiple dendritic branches (join on game_id + nxer_name + neuron_id)
• Synapses → Neurons: pre_id and post_id reference neuron_id within the same NxEr

Usage

Loading with Python (pandas)

import pandas as pd

# Load individual tables
nxers = pd.read_parquet('neuraxonLife2-1M_nxers.parquet')
neurons = pd.read_parquet('neuraxonLife2-1M_neurons.parquet')
synapses = pd.read_parquet('neuraxonLife2-1M_synapses.parquet')
branches = pd.read_parquet('neuraxonLife2-1M_branches.parquet')

# Example: Get all neurons for a specific agent
agent_neurons = neurons[neurons['nxer_name'] == 'NxEr_42']

# Example: Analyze fitness vs network topology
import matplotlib.pyplot as plt
plt.scatter(nxers['n_synapses'], nxers['fitness'])
plt.xlabel('Number of Synapses')
plt.ylabel('Fitness Score')
plt.show()

Loading with Hugging Face Datasets

from datasets import load_dataset

# Load from Hugging Face Hub
dataset = load_dataset("DavidVivancos/NeuraxonLife2-1M")

# Access tables
nxers = dataset['nxers']
neurons = dataset['neurons']

Example Analyses

1. Fitness Prediction

from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import train_test_split

features = ['n_synapses', 'conn_density', 'curr_da', 'curr_ser', 
            'membrane_tau', 'learn_rate', 'n_hidden']
X = nxers[features]
y = nxers['fitness']

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
model = RandomForestRegressor()
model.fit(X_train, y_train)
print(f"R² Score: {model.score(X_test, y_test):.3f}")

2. Synaptic Weight Analysis

# Weight distribution by synapse type
synapses.groupby('syn_type')['w_fast'].describe()

# Excitatory vs inhibitory balance
exc_weights = synapses[synapses['w_fast'] > 0]['w_fast'].sum()
inh_weights = synapses[synapses['w_fast'] < 0]['w_fast'].abs().sum()
print(f"E/I Ratio: {exc_weights / inh_weights:.2f}")

3. Network Topology

import networkx as nx

# Build graph for one agent
agent_synapses = synapses[synapses['nxer_name'] == 'NxEr_42']
G = nx.DiGraph()
for _, syn in agent_synapses.iterrows():
    G.add_edge(syn['pre_id'], syn['post_id'], weight=syn['w_fast'])

# Analyze topology
print(f"Clustering coefficient: {nx.average_clustering(G):.3f}")
print(f"Average path length: {nx.average_shortest_path_length(G):.3f}")

Dataset Creation

This dataset was generated using the Neuraxon Artificial Life simulation Research framework 2.0.

The extraction process:

1. 1000s of Test Games where performed, that saved 1000s of json files
2. Then Loading game state JSON files from simulation runs
3. Extracting hierarchical data (agents → neurons → synapses → branches)
4. Converting to columnar Parquet format with Snappy compression
5. Validating data integrity and relationships

Citation

If you use this dataset, please cite:

@dataset{NeuraxonLife2-1M,
  title={Neuraxon: Artificial Life 2.0 BioInspired Neural Network Simulation 1M Dataset},
  author={Vivancos, David and Sanchez, Jose},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/DavidVivancos/NeuraxonLife2-1M}
}

License

This dataset is released under the CC BY 4.0 license.

Additional Information

Authors

• David Vivancos / Artificiology Research https://artificiology.com/ - Qubic Science https://qubic.org/
• Dr. Jose Sanchez / UNIR - Qubic Science https://qubic.org/

Dataset Curators

• David Vivancos / Artificiology Research https://artificiology.com/ - Qubic Science https://qubic.org/
• Dr. Jose Sanchez / UNIR - Qubic Science https://qubic.org/

Version History

• v1.0.0 (2025): Initial release

Contact

For questions or issues, please open a GitHub issue here https://github.com/DavidVivancos/Neuraxon or contact [vivancos@vivancos.com].