List your company

General Instruction Following Evaluation on Aggregated Multi-Task Suite

8.57Average Normalized ScoreEPI
7.277.60757.9458.2825Apr 15, 2026

Evaluation results

MethodLinks
EPIBase Model=Gemma-2-9B, Number of runs=32026.048.57
Static IsolationBase Model=Gemma-2-9B, Number of runs=32026.048.21
EPIBase Model=LLaMA-3-8B, Number of runs=32026.047.98
Full SFTBase Model=Gemma-2-9B, Number of runs=32026.047.78
Static IsolationBase Model=LLaMA-3-8B, Number of runs=32026.047.68
Full SFTBase Model=LLaMA-3-8B, Number of runs=32026.047.32