File size: 1,175 Bytes
b10ec68 |
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 |
import torch
from safetensors.torch import save_file
weights = {}
# Single digit BCD (4 bits) to Binary (4 bits)
# BCD: 0-9 encoded in 4 bits
# Output: same value (identity for single digit)
def add_neuron(name, w_list, bias):
weights[f'{name}.weight'] = torch.tensor([w_list], dtype=torch.float32)
weights[f'{name}.bias'] = torch.tensor([bias], dtype=torch.float32)
# Pass through (BCD 0-9 is same as binary 0-9)
for i in range(4):
w = [0.0] * 4
w[i] = 1.0
add_neuron(f'b{3-i}', w, -1.0)
save_file(weights, 'model.safetensors')
def bcd2binary(d3, d2, d1, d0):
return d3, d2, d1, d0
print("Verifying BCD to Binary...")
errors = 0
for d in range(10): # Valid BCD: 0-9
d3, d2, d1, d0 = (d>>3)&1, (d>>2)&1, (d>>1)&1, d&1
b3, b2, b1, b0 = bcd2binary(d3, d2, d1, d0)
result = b3*8 + b2*4 + b1*2 + b0
if result != d:
errors += 1
if errors == 0:
print("All 10 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}")
print(f"Parameters: {sum(t.numel() for t in weights.values())}")
|