Historical Development Checkpoint: E4B DPO V2 represents an early research iteration. The canonical reference checkpoint remains SFT, and latest empirical findings reflect the completed H8 double-blind evaluation. View live telemetry on the Current Research Status page →

Experiment: E4B DPO V2 (Historical)

Narrative experiment timeline detailing the progression from Base Model to SFT to DPO V2 and Clean Evaluation.

Base Model (Gemma 4)

SFT Reference

DPO V6 Diagnostic

H8 Human Eval (No Promotion)

SHAE Assurance (Sealed)

Technical Details

Technical data temporarily unavailable.

Read the canonical E4B DPO V2 Corrective document on GitHub →