| ---
|
| 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
|
|
|
