File size: 2,005 Bytes
5dd640d 818e32c 5dd640d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 |
---
license: mit
tags:
- pytorch
- safetensors
- threshold-logic
- neuromorphic
---
# threshold-majority
Strict majority detector for 8 inputs. Fires when more than half are active.
## Circuit
```
x₀ x₁ x₂ x₃ x₄ x₅ x₆ x₇
│ │ │ │ │ │ │ │
└──┴──┴──┴──┼──┴──┴──┴──┘
▼
┌─────────┐
│ w: all 1│
│ b: -5 │
└─────────┘
│
▼
HW ≥ 5?
```
## Mechanism
- Sum = (number of 1s) - 5
- Fires when Hamming weight ≥ 5 (more 1s than 0s)
A tie (4-4) doesn't count as majority. You need strictly more than half.
Functionally identical to threshold-5outof8.
## Duality with Minority
| Circuit | Weights | Bias | Fires when |
|---------|---------|------|------------|
| **Majority** | all +1 | -5 | HW ≥ 5 |
| Minority | all -1 | +3 | HW ≤ 3 |
Majority: "enough votes to pass"
Minority: "not enough votes to block"
These are not complements (they don't sum to 1). The gap at HW=4 belongs to neither.
## Parameters
| | |
|---|---|
| Weights | [1, 1, 1, 1, 1, 1, 1, 1] |
| Bias | -5 |
| Total | 9 parameters |
## Optimality
Exhaustive enumeration of all 27,298,155 weight configurations at magnitudes 0-13 confirms this circuit is **already at minimum magnitude (13)**. There is exactly one valid configuration at magnitude 13, and no valid configurations exist below it.
## Usage
```python
from safetensors.torch import load_file
import torch
w = load_file('model.safetensors')
def majority(bits):
inputs = torch.tensor([float(b) for b in bits])
return int((inputs * w['weight']).sum() + w['bias'] >= 0)
```
## Files
```
threshold-majority/
├── model.safetensors
├── model.py
├── config.json
└── README.md
```
## License
MIT
|