|
|
--- |
|
|
license: apache-2.0 |
|
|
tags: |
|
|
- code |
|
|
--- |
|
|
|
|
|
boolean networks, each sampled over 8192 timesteps, ranging from network size 32-1024, with connectivities (sizes of functions) per item taken from random gaussian mixtures |
|
|
|
|
|
features: |
|
|
``` |
|
|
network_size # number of nodes in the network |
|
|
connectivity_modes # gaussians that connectivities are sampled from |
|
|
connectivity_mu |
|
|
connectivity_sigma |
|
|
connectivity_mode_weights |
|
|
initial_state |
|
|
function_keys # think left (input) side of truth table |
|
|
function_values # think right (output) side of truth table |
|
|
function_inputs # the node indices that go into this function |
|
|
``` |
|
|
|
|
|
for a tiny subset with 128 networks, for example if testing a pipeline, try: |
|
|
|
|
|
```python |
|
|
from datasets import load_dataset |
|
|
|
|
|
dataset = load_dataset( |
|
|
"parquet", data_files=[ |
|
|
"https://huggingface.co/datasets/midwestern-simulation/boolean-networks/resolve/main/chunk_gUKO6HBS_128_8192_32-1024.parquet" |
|
|
# "chunk_{id}_{num_samples}_{min_network_size}_{max_network_size}.parquet" |
|
|
] |
|
|
) |
