8-bit bit reversal, magnitude 16

README.md

@@ -0,0 +1,72 @@
+---
+license: mit
+tags:
+- pytorch
+- safetensors
+- threshold-logic
+- neuromorphic
+---
+
+# threshold-reverse8
+
+8-bit bit reversal. Reverses the order of bits.
+
+## Function
+
+reverse8(a7, a6, a5, a4, a3, a2, a1, a0) = [a0, a1, a2, a3, a4, a5, a6, a7]
+
+## Examples
+
+| Input | Output |
+|-------|--------|
+| 10000000 | 00000001 |
+| 00000001 | 10000000 |
+| 10101010 | 01010101 |
+| 11110000 | 00001111 |
+
+## Architecture
+
+Single layer with 8 neurons, each copying one input bit to its reversed position.
+
+| Output | Copies from | Weights | Bias |
+|--------|-------------|---------|------|
+| y0 | a7 | [1,0,0,0,0,0,0,0] | -1 |
+| y1 | a6 | [0,1,0,0,0,0,0,0] | -1 |
+| y2 | a5 | [0,0,1,0,0,0,0,0] | -1 |
+| y3 | a4 | [0,0,0,1,0,0,0,0] | -1 |
+| y4 | a3 | [0,0,0,0,1,0,0,0] | -1 |
+| y5 | a2 | [0,0,0,0,0,1,0,0] | -1 |
+| y6 | a1 | [0,0,0,0,0,0,1,0] | -1 |
+| y7 | a0 | [0,0,0,0,0,0,0,1] | -1 |
+
+## Parameters
+
+| | |
+|---|---|
+| Inputs | 8 |
+| Outputs | 8 |
+| Neurons | 8 |
+| Layers | 1 |
+| Parameters | 16 |
+| Magnitude | 16 |
+
+## Usage
+
+```python
+from safetensors.torch import load_file
+import torch
+
+w = load_file('model.safetensors')
+
+def reverse8(a7, a6, a5, a4, a3, a2, a1, a0):
+    inp = torch.tensor([float(a7), float(a6), float(a5), float(a4),
+                        float(a3), float(a2), float(a1), float(a0)])
+    return [int((inp @ w[f'y{i}.weight'].T + w[f'y{i}.bias'] >= 0).item())
+            for i in range(8)]
+
+print(reverse8(1, 0, 0, 0, 0, 0, 0, 0))  # [0, 0, 0, 0, 0, 0, 0, 1]
+```
+
+## License
+
+MIT

config.json

@@ -0,0 +1,9 @@
+{
+  "name": "threshold-reverse8",
+  "description": "8-bit bit reversal",
+  "inputs": 8,
+  "outputs": 8,
+  "neurons": 8,
+  "layers": 1,
+  "parameters": 16
+}

create_safetensors.py

@@ -0,0 +1,46 @@
+import torch
+from safetensors.torch import save_file
+
+weights = {}
+
+# Input order: [a7, a6, a5, a4, a3, a2, a1, a0]
+# Output: [a0, a1, a2, a3, a4, a5, a6, a7]
+
+# Each output yi copies from input a(7-i)
+for i in range(8):
+    w = [0.0] * 8
+    w[7 - i] = 1.0  # y0 copies a7, y1 copies a6, ..., y7 copies a0
+    weights[f'y{i}.weight'] = torch.tensor([w], dtype=torch.float32)
+    weights[f'y{i}.bias'] = torch.tensor([-1.0], dtype=torch.float32)
+
+save_file(weights, 'model.safetensors')
+
+# Verify
+def reverse8(a7, a6, a5, a4, a3, a2, a1, a0):
+    inp = torch.tensor([float(a7), float(a6), float(a5), float(a4),
+                        float(a3), float(a2), float(a1), float(a0)])
+    outputs = []
+    for i in range(8):
+        y = int((inp @ weights[f'y{i}.weight'].T + weights[f'y{i}.bias'] >= 0).item())
+        outputs.append(y)
+    return outputs
+
+print("Verifying reverse8...")
+errors = 0
+for i in range(256):
+    bits = [(i >> j) & 1 for j in range(7, -1, -1)]
+    a7, a6, a5, a4, a3, a2, a1, a0 = bits
+    result = reverse8(a7, a6, a5, a4, a3, a2, a1, a0)
+    expected = [a0, a1, a2, a3, a4, a5, a6, a7]
+    if result != expected:
+        errors += 1
+        if errors <= 5:
+            print(f"ERROR: {bits} -> {result}, expected {expected}")
+
+if errors == 0:
+    print("All 256 test cases passed!")
+else:
+    print(f"FAILED: {errors} errors")
+
+mag = sum(t.abs().sum().item() for t in weights.values())
+print(f"Magnitude: {mag:.0f}")

model.py

@@ -0,0 +1,27 @@
+import torch
+from safetensors.torch import load_file
+
+def load_model(path='model.safetensors'):
+    return load_file(path)
+
+def reverse8(a7, a6, a5, a4, a3, a2, a1, a0, weights):
+    """8-bit bit reversal."""
+    inp = torch.tensor([float(a7), float(a6), float(a5), float(a4),
+                        float(a3), float(a2), float(a1), float(a0)])
+    outputs = []
+    for i in range(8):
+        y = int((inp @ weights[f'y{i}.weight'].T + weights[f'y{i}.bias'] >= 0).item())
+        outputs.append(y)
+    return outputs
+
+if __name__ == '__main__':
+    w = load_model()
+    print('reverse8 examples:')
+    test_cases = [
+        (1, 0, 0, 0, 0, 0, 0, 0),
+        (0, 0, 0, 0, 0, 0, 0, 1),
+        (1, 0, 1, 0, 0, 1, 0, 1),
+    ]
+    for bits in test_cases:
+        result = reverse8(*bits, w)
+        print(f'  {list(bits)} -> {result}')