List your company

Safety Classification on OpenAI-moderation (test)

74.88AccuracyTPC
59.612863.576467.5471.5036Sep 30, 2025Nov 8, 2025Dec 18, 2025Jan 26, 2026Mar 7, 2026Apr 15, 2026May 25, 2026

Evaluation results

MethodLinks
TPCOrder=5th order, Backbone=gemma-3-27b-it, Layer=40, Training Set=WildGuardTrain2025.0974.8869.0255.9590.0431.95
EE-MLPOrder=5th order, Backbone=gemma-3-27b-it, Layer=40, Training Set=WildGuardTrain2025.0973.0467.3953.9889.6634.46
Linear probeBackbone=gemma-3-27b-it, Layer=40, Training Set=WildGuardTrain2025.0972.9866.2754.1385.4432.64
D2-TimeAttnBackbone=LLaDA-8B-Instruct, E[P]=≤0.85M2026.0569.669.1———
D2-MLPBackbone=LLaDA-8B-Instruct, E[P]=≤0.25M2026.0568.868.2———
LSTMBackbone=LLaDA-8B-Instruct, E[P]=2.57M2026.0568.567.7———
LP (Last Step)Backbone=LLaDA-8B-Instruct, E[P]=4×10−3M2026.056867.4———
MLP (Last Step)Backbone=LLaDA-8B-Instruct, E[P]=1.05M2026.0566.966.4———
D2-MLPBackbone=LLaDA-8B-Base, E[P]=≤0.25M2026.056665.6———
D2-TimeAttnBackbone=LLaDA-8B-Base, E[P]=≤0.85M2026.056665.7———
TimeAttnBackbone=LLaDA-8B-Base, E[P]=1.59M2026.0565.865.3———
LSTMBackbone=LLaDA-8B-Base, E[P]=2.57M2026.0565.565———
TimeAttnBackbone=LLaDA-8B-Instruct, E[P]=1.59M2026.0565.465.4———
LP (MV)Backbone=LLaDA-8B-Instruct, E[P]=4×10−3M2026.0565.164.7———
LP (Mean)Backbone=LLaDA-8B-Instruct, E[P]=4×10−3M2026.0564.964.5———
MLP (Mean)Backbone=LLaDA-8B-Base, E[P]=1.05M2026.0564.864.6———
LP (Mean)Backbone=LLaDA-8B-Base, E[P]=4×10−3M2026.0563.963.8———
MLP (MV)Backbone=LLaDA-8B-Base, E[P]=1.05M2026.0563.963.7———
MLP (Last Step)Backbone=LLaDA-8B-Base, E[P]=1.05M2026.0563.363.2———
LP (MV)Backbone=LLaDA-8B-Base, E[P]=4×10−3M2026.0562.962.7———
LP (Last Step)Backbone=LLaDA-8B-Base, E[P]=4×10−3M2026.0562.662.4———
MLP (Mean)Backbone=LLaDA-8B-Instruct, E[P]=1.05M2026.0560.560.4———
MLP (MV)Backbone=LLaDA-8B-Instruct, E[P]=1.05M2026.0560.260.2———