Trust Center · Large Quantitative Models
LQM model evidence and supply-chain controls
How FinanceGPT separates quantitative calculation from language, verifies exact checkpoints, evaluates model behaviour and records public model-supply-chain evidence.
7
required evaluation gates
0/0
public models with complete disclosure evidence
0
public models separately promoted inside FinanceGPT
Evidence chain
✓Sealed training lineage
✓Exact checkpoint SHA-256
✓Seven-gate quantitative evaluation
✓Versioned Evidence Packet
✓Model card and governance addendum
✓ML-BOM and publication manifest
✓Remote revision and artifact hashes
Authority separation
✓LQM does not generate language
✓Language runtime receives typed evidence, not latent state
✓Publication does not activate a model
✓Promotion requires a separate attributable human decision
✓Financial Actions authority stays outside model control
✓No automatic QLM rebinding
Public verification