DavidVivancos/NeuraxonLife2-1M

---
license: cc-by-4.0
task_categories:
- tabular-classification
- tabular-regression
tags:
- neuraxon
- neuroscience
- artificial-life
- neural-networks
- simulation
- biology
- AI
- AGI
- Qubic-Aigarth
- evolutionary-algorithms
pretty_name: 'Neuraxon Artificial Life 2.0 Simulation Dataset 1M '
size_categories:
- 1M<n<10M
language:
- en
---

# 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)

```python
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

```python
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
```python
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
```python
# 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
```python
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:

```bibtex
@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](https://creativecommons.org/licenses/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 [[e-mail]].