threshold-atmost6outof8

At-most-6-out-of-8 detector. Fires when 6 or fewer inputs are active. The near-unanimity blocker.

Circuit

  x₀ x₁ x₂ x₃ x₄ x₅ x₆ x₇
   │  │  │  │  │  │  │  │
   └──┴──┴──┴──┼──┴──┴──┴──┘
               ▼
          ┌─────────┐
          │ w: -1×8 │
          │ b:  +6  │
          └─────────┘
               │
               ▼
           HW ≤ 6?

Detecting Non-Consensus

This circuit fires unless near-unanimity or unanimity is achieved:

HW	AtMost6	Status
0-6	1	At least 2 dissenters
7	0	Single holdout
8	0	Unanimous

It identifies states where meaningful opposition exists.

The Two-Dissenter Minimum

For something to pass AtMost6:

• At least 2 inputs must be inactive
• At most 6 inputs can be active

This is the threshold for "blocking minority" in many systems.

Coverage

Nearly everything passes:

HW	C(8,k)	AtMost6?
0-6	247	Yes
7	8	No
8	1	No

Total: 247 of 256 inputs (96.5%).

Dual of AtLeast2

Circuit	Condition	Fires on
AtLeast2	HW ≥ 2	247 inputs
AtMost6	HW ≤ 6	247 inputs

Both fire on almost everything. AtLeast2 misses empty/singleton; AtMost6 misses near-unanimous/unanimous.

Parameters

Component	Value
Weights	all -1
Bias	+6
Total	9 parameters

Usage

from safetensors.torch import load_file
import torch

w = load_file('model.safetensors')

def atmost6(bits):
    inp = torch.tensor([float(b) for b in bits])
    return int((inp * w['weight']).sum() + w['bias'] >= 0)

# Supermajority with 2 dissenters
print(atmost6([1,1,1,1,1,1,0,0]))  # 1

# Single holdout
print(atmost6([1,1,1,1,1,1,1,0]))  # 0

Files

threshold-atmost6outof8/
├── model.safetensors
├── model.py
├── config.json
└── README.md

License

MIT