README.md

---
license: mit
tags:
- pytorch
- safetensors
- threshold-logic
- neuromorphic
- encoding
---

# threshold-onehot-encoder

2-to-4 one-hot encoder. Converts a 2-bit binary value to a 4-bit one-hot representation.

## Function

onehot_encode(a1, a0) -> (y3, y2, y1, y0)

Exactly one output bit is set, corresponding to the input value.

## Truth Table

| a1 | a0 | y3 | y2 | y1 | y0 | Value |
|----|----|----|----|----|-----|-------|
| 0 | 0 | 0 | 0 | 0 | 1 | 0 |
| 0 | 1 | 0 | 0 | 1 | 0 | 1 |
| 1 | 0 | 0 | 1 | 0 | 0 | 2 |
| 1 | 1 | 1 | 0 | 0 | 0 | 3 |

## Architecture

Single-layer implementation using threshold logic:

```
        a1    a0
        │     │
    ┌───┴─────┴───┐
    │             │
    ▼   ▼   ▼   ▼
  ┌───┬───┬───┬───┐
  │y3 │y2 │y1 │y0 │  Layer 1
  │AND│A·B│A·B│NOR│
  └───┴───┴───┴───┘
    │   │   │   │
    ▼   ▼   ▼   ▼
```

Each output is a single threshold neuron:
- y0 = NOR(a1, a0): w=[-1,-1], b=0
- y1 = NOT(a1) AND a0: w=[-1,1], b=-1
- y2 = a1 AND NOT(a0): w=[1,-1], b=-1
- y3 = a1 AND a0: w=[1,1], b=-2

## Parameters

| | |
|---|---|
| Inputs | 2 |
| Outputs | 4 |
| Neurons | 4 |
| Layers | 1 |
| Parameters | 12 |
| Magnitude | 12 |

## Usage

```python
from safetensors.torch import load_file
import torch

w = load_file('model.safetensors')

def onehot(a1, a0):
    inp = torch.tensor([float(a1), float(a0)])
    y0 = int((inp @ w['y0.weight'].T + w['y0.bias'] >= 0).item())
    y1 = int((inp @ w['y1.weight'].T + w['y1.bias'] >= 0).item())
    y2 = int((inp @ w['y2.weight'].T + w['y2.bias'] >= 0).item())
    y3 = int((inp @ w['y3.weight'].T + w['y3.bias'] >= 0).item())
    return y3, y2, y1, y0

# onehot(1, 0) = (0, 1, 0, 0)  # value 2
```

## Applications

- Address decoding
- Memory select lines
- State machine encoding
- Neural network input preprocessing

## License

MIT