Can a generator write new network weights?
A diffusion generator trained on 300 checkpoints of a small MLP's last layer was asked to produce new weights at equal accuracy. It copied: for every one of 600 generated samples, some training checkpoint predicts the same class on all 450 test inputs.
Key numbers
generated samples with zero disagreement novelty
600 of 600Source file:
Source hash:
weightgen-audit/runs/2026-10-13/results.jsonSource hash:
2bffc7f62ddbdaf25d759b32677a49f93f81dda53f762d6f758f8051722397f2Caveats
- One 330-parameter head on 8x8 digits
- Replicates Zeng et al. 2025 (copying)
Watch the roundup
What we tested, and what we learned (90 s)
Video hash: 3de55bc330bba5f8
Request the full write-up
Full write-ups are not public yet. You can request one; requests are reviewed by hand.
Requests are stored privately for up to twelve months, reviewed by hand, never shared, and deleted on request. Contact QMAI@quantummindaillc.com.