Upload folder using huggingface_hub

README.md

@@ -0,0 +1,56 @@
+---
+license: mit
+tags:
+- pytorch
+- safetensors
+- threshold-logic
+- neuromorphic
+---
+
+# threshold-xnor3
+
+3-input XNOR gate. Outputs 1 when an even number of inputs are 1 (0 or 2).
+
+## Architecture
+
+Cascade: XNOR3(a,b,c) = XOR(XNOR(a,b), c)
+
+```
+a,b ──► XNOR ──► XOR ──► output
+            c ──┘
+```
+
+## Parameters
+
+| | |
+|---|---|
+| Neurons | 6 |
+| Layers | 4 |
+| Parameters | 18 |
+| Magnitude | 19 |
+
+## Truth Table
+
+| a | b | c | output |
+|---|---|---|--------|
+| 0 | 0 | 0 | 1 |
+| 0 | 0 | 1 | 0 |
+| 0 | 1 | 0 | 0 |
+| 0 | 1 | 1 | 1 |
+| 1 | 0 | 0 | 0 |
+| 1 | 0 | 1 | 1 |
+| 1 | 1 | 0 | 1 |
+| 1 | 1 | 1 | 0 |
+
+## Usage
+
+```python
+from safetensors.torch import load_file
+
+w = load_file('model.safetensors')
+# See model.py for forward pass
+```
+
+## License
+
+MIT

config.json

@@ -0,0 +1,9 @@
+{
+  "name": "threshold-xnor3",
+  "description": "3-input XNOR gate as threshold circuit",
+  "inputs": 3,
+  "outputs": 1,
+  "neurons": 6,
+  "layers": 4,
+  "parameters": 18
+}

create_safetensors.py

@@ -0,0 +1,40 @@
+import torch
+from safetensors.torch import save_file
+
+# XNOR3 = XOR(XNOR(a,b), c)
+# Using standard (non-optimized) weights
+
+# XNOR(a,b): NOR + AND -> OR structure
+# n1 (NOR): fires when both inputs 0
+# n2 (AND): fires when both inputs 1
+# out (OR): fires when either n1 or n2
+
+# XOR(x,y): OR + NAND -> AND structure
+# n1 (OR): fires when at least one input 1
+# n2 (NAND): fires when not both inputs 1
+# out (AND): fires when both n1 and n2
+
+weights = {
+    # First XNOR: a XNOR b
+    'xnor1.layer1.n1.weight': torch.tensor([-1.0, -1.0]),  # NOR
+    'xnor1.layer1.n1.bias': torch.tensor([0.0]),
+    'xnor1.layer1.n2.weight': torch.tensor([1.0, 1.0]),    # AND
+    'xnor1.layer1.n2.bias': torch.tensor([-2.0]),
+    'xnor1.layer2.weight': torch.tensor([1.0, 1.0]),       # OR
+    'xnor1.layer2.bias': torch.tensor([-1.0]),
+
+    # Second XOR: (a XNOR b) XOR c
+    'xor2.layer1.n1.weight': torch.tensor([1.0, 1.0]),     # OR
+    'xor2.layer1.n1.bias': torch.tensor([-1.0]),
+    'xor2.layer1.n2.weight': torch.tensor([-1.0, -1.0]),   # NAND
+    'xor2.layer1.n2.bias': torch.tensor([1.0]),
+    'xor2.layer2.weight': torch.tensor([1.0, 1.0]),        # AND
+    'xor2.layer2.bias': torch.tensor([-2.0]),
+}
+
+save_file(weights, 'model.safetensors')
+
+mag = sum(t.abs().sum().item() for t in weights.values())
+print(f'Created model.safetensors')
+print(f'  magnitude: {mag:.0f}')
+print(f'  parameters: {sum(t.numel() for t in weights.values())}')

model.py

@@ -0,0 +1,57 @@
+"""
+Threshold Network for 3-input XNOR Gate
+
+XNOR3(a,b,c) = 1 when even number of inputs are 1 (0 or 2)
+Built as: XOR(XNOR(a,b), c)
+"""
+
+import torch
+from safetensors.torch import load_file
+
+
+class ThresholdXNOR3:
+    def __init__(self, weights_dict):
+        self.w = weights_dict
+
+    def __call__(self, a, b, c):
+        # First XNOR: a XNOR b (1 when a=b)
+        inp1 = torch.tensor([float(a), float(b)])
+        n1 = int((inp1 * self.w['xnor1.layer1.n1.weight']).sum() + self.w['xnor1.layer1.n1.bias'] >= 0)
+        n2 = int((inp1 * self.w['xnor1.layer1.n2.weight']).sum() + self.w['xnor1.layer1.n2.bias'] >= 0)
+
+        h1 = torch.tensor([float(n1), float(n2)])
+        xnor_ab = int((h1 * self.w['xnor1.layer2.weight']).sum() + self.w['xnor1.layer2.bias'] >= 0)
+
+        # Second XOR: xnor_ab XOR c
+        inp2 = torch.tensor([float(xnor_ab), float(c)])
+        n3 = int((inp2 * self.w['xor2.layer1.n1.weight']).sum() + self.w['xor2.layer1.n1.bias'] >= 0)
+        n4 = int((inp2 * self.w['xor2.layer1.n2.weight']).sum() + self.w['xor2.layer1.n2.bias'] >= 0)
+
+        h2 = torch.tensor([float(n3), float(n4)])
+        out = int((h2 * self.w['xor2.layer2.weight']).sum() + self.w['xor2.layer2.bias'] >= 0)
+
+        return float(out)
+
+    @classmethod
+    def from_safetensors(cls, path="model.safetensors"):
+        return cls(load_file(path))
+
+
+if __name__ == "__main__":
+    weights = load_file("model.safetensors")
+    model = ThresholdXNOR3(weights)
+
+    print("3-input XNOR Gate Truth Table:")
+    print("-" * 30)
+    correct = 0
+    for a in [0, 1]:
+        for b in [0, 1]:
+            for c in [0, 1]:
+                out = int(model(a, b, c))
+                # XNOR3 = even parity = NOT XOR3
+                expected = 1 - (a ^ b ^ c)
+                status = "OK" if out == expected else "FAIL"
+                if out == expected:
+                    correct += 1
+                print(f"XNOR3({a}, {b}, {c}) = {out}  [{status}]")
+    print(f"\nTotal: {correct}/8 correct")

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f789ee1eaf958fa522d3df45978f9ce232709a45029ad20cc38270c5016b47ea
+size 960