2026-10-11 16:37 UTC

model-fingerprinting

band: coolmomentum: stable score: 0.0
temperature history

Episodes (2)

Independent evaluations will determine whether Model Genome can reliably distinguish scratch-trained LLMs from derived models and identify their training lineage.
expiredconvergesscott: medium
Independent testing will determine whether roughly 200 one-token random-number queries can reliably identify LLMs and expose silent model substitution by third-party API relays.
expiredconvergesscott: high

Trajectory notes