|
|
---
|
|
|
license: mit
|
|
|
tags:
|
|
|
- pytorch
|
|
|
- safetensors
|
|
|
- threshold-logic
|
|
|
- neuromorphic
|
|
|
- decoder
|
|
|
---
|
|
|
|
|
|
# threshold-4to16decoder
|
|
|
|
|
|
4-to-16 binary decoder. Converts 4-bit binary input to one-hot 16-bit output.
|
|
|
|
|
|
## Function
|
|
|
|
|
|
decode(a3, a2, a1, a0) -> [y0..y15] where yi=1 iff input=i
|
|
|
|
|
|
## One-Hot Encoding
|
|
|
|
|
|
| Input | a3a2a1a0 | Output |
|
|
|
|------:|:--------:|--------|
|
|
|
| 0 | 0000 | 1000000000000000 |
|
|
|
| 1 | 0001 | 0100000000000000 |
|
|
|
| 5 | 0101 | 0000010000000000 |
|
|
|
| 10 | 1010 | 0000000000100000 |
|
|
|
| 15 | 1111 | 0000000000000001 |
|
|
|
|
|
|
## Architecture
|
|
|
|
|
|
Single layer with 16 neurons. Each neuron yi is a pattern matcher for i:
|
|
|
- Weight +1 for bit positions that should be 1
|
|
|
- Weight -1 for bit positions that should be 0
|
|
|
- Bias = -(number of 1 bits in i)
|
|
|
|
|
|
All neurons run in parallel - no dependencies.
|
|
|
|
|
|
## Parameters
|
|
|
|
|
|
| | |
|
|
|
|---|---|
|
|
|
| Inputs | 4 |
|
|
|
| Outputs | 16 |
|
|
|
| Neurons | 16 |
|
|
|
| Layers | 1 |
|
|
|
| Parameters | 80 |
|
|
|
| Magnitude | 96 |
|
|
|
|
|
|
## Usage
|
|
|
|
|
|
```python
|
|
|
from safetensors.torch import load_file
|
|
|
import torch
|
|
|
|
|
|
w = load_file('model.safetensors')
|
|
|
|
|
|
def decode_4to16(a3, a2, a1, a0):
|
|
|
inp = torch.tensor([float(a3), float(a2), float(a1), float(a0)])
|
|
|
return [int((inp * w[f'y{i}.weight']).sum() + w[f'y{i}.bias'] >= 0)
|
|
|
for i in range(16)]
|
|
|
|
|
|
# Input 10 -> output 10 is hot
|
|
|
outputs = decode_4to16(1, 0, 1, 0)
|
|
|
print(outputs) # [0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0]
|
|
|
```
|
|
|
|
|
|
## License
|
|
|
|
|
|
MIT
|
|
|
|
