Upload folder using huggingface_hub

README.md

@@ -0,0 +1,71 @@
+---
+license: mit
+tags:
+- pytorch
+- safetensors
+- threshold-logic
+- neuromorphic
+---
+
+# threshold-exactly2outof8
+
+Exactly-2-out-of-8 detector. Fires when exactly two inputs are active.
+
+## Circuit
+
+```
+  x₀ x₁ x₂ x₃ x₄ x₅ x₆ x₇
+   │  │  │  │  │  │  │  │
+   └──┴──┴──┴──┼──┴──┴──┴──┘
+               │
+       ┌───────┴───────┐
+       ▼               ▼
+  ┌─────────┐     ┌─────────┐
+  │ AtLeast2│     │ AtMost2 │
+  │ w: +1×8 │     │ w: -1×8 │
+  │ b:  -2  │     │ b:  +2  │
+  └─────────┘     └─────────┘
+       │               │
+       └───────┬───────┘
+               ▼
+          ┌─────────┐
+          │   AND   │
+          └─────────┘
+               │
+               ▼
+          (HW = 2?)
+```
+
+## Mechanism
+
+- **AtLeast2**: Fires when HW ≥ 2 (bias -2)
+- **AtMost2**: Fires when HW ≤ 2 (weights -1, bias +2)
+- **AND**: Exactly2 = AtLeast2 AND AtMost2
+
+## Truth Table
+
+| HW | AtLeast2 | AtMost2 | Exactly2 |
+|----|----------|---------|----------|
+| 0 | 0 | 1 | 0 |
+| 1 | 0 | 1 | 0 |
+| 2 | 1 | 1 | **1** |
+| 3 | 1 | 0 | 0 |
+| ... | 1 | 0 | 0 |
+
+## Architecture
+
+**3 neurons, 21 parameters, 2 layers**
+
+## Files
+
+```
+threshold-exactly2outof8/
+├── model.safetensors
+├── model.py
+├── config.json
+└── README.md
+```
+
+## License
+
+MIT

config.json

@@ -0,0 +1,9 @@
+{
+  "name": "threshold-exactly2outof8",
+  "description": "Exactly-2-out-of-8 detector as threshold circuit",
+  "inputs": 8,
+  "outputs": 1,
+  "neurons": 3,
+  "layers": 2,
+  "parameters": 21
+}

model.py

@@ -0,0 +1,23 @@
+import torch
+from safetensors.torch import load_file
+
+def load_model(path='model.safetensors'):
+    return load_file(path)
+
+def exactlyK(bits, weights):
+    """Exactly-K-out-of-8 detector.
+    bits: list of 8 binary values
+    Returns: 1 if exactly K bits are set, 0 otherwise
+    """
+    inp = torch.tensor([float(b) for b in bits])
+    atleast = int((inp * weights['atleast.weight']).sum() + weights['atleast.bias'] >= 0)
+    atmost = int((inp * weights['atmost.weight']).sum() + weights['atmost.bias'] >= 0)
+    return int((torch.tensor([float(atleast), float(atmost)]) * weights['and.weight']).sum() + weights['and.bias'] >= 0)
+
+if __name__ == '__main__':
+    w = load_model()
+    print('ExactlyKOutOf8 Detector')
+    for hw in range(9):
+        bits = [1] * hw + [0] * (8 - hw)
+        result = exactlyK(bits, w)
+        print(f'HW={hw}: {result}')

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:140924bcb6ff3dd9b269169c5652694c92023c60617169c1a799e3b4ab0baa86
+size 484