List your company

Natural Language Generation and Understanding on S_NLG and S_NLU (test)

84.17Average PerformanceTeacher
23.610839.332955.05570.7771Feb 19, 2024

Evaluation results

MethodLinks
TeacherStudent Model=LLaMA-68M, Teacher Model=LLaMA-7B2024.0284.17
TeacherStudent Model=LLaMA-68M, Teacher Model=LLaMA-1.1B2024.0275.23
TeacherStudent Model=Pythia-410M, Teacher Model=Pythia-2.8B2024.0273.5
TeacherStudent Model=Pythia-410M, Teacher Model=Pythia-1.4B2024.0267.86
Supervised KD + ATKDStudent Model=Pythia-410M, Teacher Model=Pythia-1.4B2024.0261.81
Supervised KD + ATKDStudent Model=Pythia-410M, Teacher Model=Pythia-2.8B2024.0261.22
Supervised KDStudent Model=Pythia-410M, Teacher Model=Pythia-1.4B2024.0260.66
Supervised KDStudent Model=Pythia-410M, Teacher Model=Pythia-2.8B2024.0259.91
Reverse KD + ATKDStudent Model=Pythia-410M, Teacher Model=Pythia-2.8B2024.0257.94
Reverse KD + ATKDStudent Model=Pythia-410M, Teacher Model=Pythia-1.4B2024.0257.05
Reverse KDStudent Model=Pythia-410M, Teacher Model=Pythia-1.4B2024.0255.92
Reverse KDStudent Model=Pythia-410M, Teacher Model=Pythia-2.8B2024.0254.67
Supervised KD + ATKDStudent Model=LLaMA-68M, Teacher Model=LLaMA-1.1B2024.0231.19
Supervised KD + ATKDStudent Model=LLaMA-68M, Teacher Model=LLaMA-7B2024.0230.19
Supervised KDStudent Model=LLaMA-68M, Teacher Model=LLaMA-1.1B2024.0230.06
Supervised KDStudent Model=LLaMA-68M, Teacher Model=LLaMA-7B2024.0227.94
Reverse KD + ATKDStudent Model=LLaMA-68M, Teacher Model=LLaMA-7B2024.0226.99
Reverse KD + ATKDStudent Model=LLaMA-68M, Teacher Model=LLaMA-1.1B2024.0226.73
Reverse KDStudent Model=LLaMA-68M, Teacher Model=LLaMA-1.1B2024.0226.15
Reverse KDStudent Model=LLaMA-68M, Teacher Model=LLaMA-7B2024.0225.94