erracc

the question

Does weight-quantization error accumulate multiplicatively through network depth, or hold a plateau through the body?

what came out

Mixed. The core claim holds: absolute residual error is flat through the body (1.00 to 1.02x growth, layer 4 to 85% depth) in both 0.5B and 1.5B, not compounding, and the late-layer relative jump is largely residual-norm collapse. But matched-magnitude Gaussian noise propagates 1.4 to 2.3x less, so quantization structure matters, refuting a pure-noise model.

→ read the code and re-run it