2026-10-11 17:12 UTC

Independent evaluations will determine whether Model Genome can reliably distinguish scratch-trained LLMs from derived models and identify their training lineage.

state: expiredheat: lowuncertainty: highconvergesscott: mediummodel-provenance model-fingerprinting open-modelsmayafreeHugging Face

What is this?

Model Genome is presented as a model-fingerprinting method intended to determine whether an LLM was trained from scratch or derived from another model, potentially tracing its training lineage. The case associates it with mayafree and Hugging Face, but the supplied results do not establish their exact roles, explain the technique, or provide independent validation. Most search results concern training-data provenance or biological genomics rather than this tool, so its reliability and practical capabilities remain ungrounded here.

Why it matters to Scott

Model Genome brings Scott’s provenance and independent capability-audit principles into post-hoc model-lineage detection. If independently validated, it could provide a new verifier for the radar’s open Kimi distillation allegations and complement existing model-identification fingerprinting, but the supplied evidence does not yet establish the technique or its reliability.
ip:framework.provenance-coupled-workip:concept.capability-auditip:concept.mechanically-different-verifiersdev:concept.claim-bounded-adversarial-verificationradar:one-token-api-model-fingerprintingradar:kimi-k3-anthropic-distillation-allegationradar:moonshot-fable-kimi-distillationradar:concept.model-distillationradar:concept.model-evaluation
queries asked of Scott's wikis
  • model provenance and derivative-model detection
  • open-weight model lineage and attribution
  • independent evaluation of model fingerprints
  • model ancestry claims and reproducibility
  • provenance requirements for AI compliance
  • fingerprinting fine-tunes, merges, and distilled models

Measured heat

no measured readings yet — the hourly heat pass fills this in

How the heat travelled

no chain yet — the hourly chain pass fills this in

Evidence (2) — ⭐ canonical anchor

sourceobjectauthorscorecomments
🟧 hnModel Genome: Fingerprinting Whether an LLM Was Trained from Scratch or Derivedprof_ai_cn10
🟧 echo.blog ⭐Introduces Model Genome as a method for fingerprinting whether an LLM was trained from scratch or derived from another model.mayafree——

Interpretation history

Decision trace