List your company

General Language Model Evaluation on MMLU, BBH, TruthfulQA, TriviaQA, and AlpacaEval

68.93MMLUELUDe
38.520446.415254.3162.2048Apr 15, 2026

Evaluation results

MethodLinks
ELUDeFinetune=GA+GD2026.0468.9322.3839.7528.596.1453
RWKUFinetune=GA2026.0468.7931.333840.776.0383
ELUDeFinetune=GA+KL2026.0468.7940.939.527.36.1676
RWKUFinetune=GA+GD2026.0468.7134.4137.542.456.0307
Before2026.0468.6440.937.7539.856.0196
MAGEFinetune=GA+KL2026.0468.6443.0638.529.696.1099
ELUDeFinetune=GA2026.0468.5731.4839.2526.76.1704
MAGEFinetune=GA+GD2026.0468.5743.0638.2544.085.9923
DirectQAFinetune=GA+KL2026.0468.5734.2638.537.526.0251
DirectQAFinetune=NPO2026.0468.4942.1338.7537.716.0951
RWKUFinetune=GA+KL2026.0468.4939.9737.541.136.0506
MAGEFinetune=GA2026.0468.4243.8339.7524.66.1535
ELUDeFinetune=NPO2026.0468.4238.5844.7515.466.4073
MAGEFinetune=NPO2026.0468.3542.943.2516.426.3429
RWKUFinetune=NPO2026.0468.2740.123922.536.2853
DirectQAFinetune=GA+GD2026.0468.234.4138.2545.155.8489
DirectQAFinetune=GA2026.0467.8443.3637.7535.066.0679
BeforeBackbone=Llama-2-7b-chat2026.0442.426.5428.7540.566.2993
DirectQAFinetuning Strategy=NPO, Backbone=Llama-2-7b-chat2026.0442.1120.992946.586.2765
DirectQAFinetuning Strategy=GA, Backbone=Llama-2-7b-chat2026.0442.0319.752946.46.2686
MAGEFinetuning Strategy=GA+KL, Backbone=Llama-2-7b-chat2026.0441.9623.583032.916.3436
DirectQAFinetuning Strategy=GA+GD, Backbone=Llama-2-7b-chat2026.0441.7428.0927.7549.16.2556
ELUDeFinetuning Strategy=GA, Backbone=Llama-2-7b-chat2026.0441.6724.3831.7532.326.3341
ELUDeFinetuning Strategy=GA+KL, Backbone=Llama-2-7b-chat2026.0441.6710.0331.7531.516.3354
MAGEFinetuning Strategy=GA+GD, Backbone=Llama-2-7b-chat2026.0441.5927.4130.524.696.3499
ELUDeFinetuning Strategy=GA+GD, Backbone=Llama-2-7b-chat2026.0441.5214.6633.528.26.3393
DirectQAFinetuning Strategy=GA+KL, Backbone=Llama-2-7b-chat2026.0441.5220.8328.545.176.2774
RWKUFinetuning Strategy=NPO, Backbone=Llama-2-7b-chat2026.0441.3720.2229.2540.536.3267
MAGEFinetuning Strategy=GA, Backbone=Llama-2-7b-chat2026.0441.325.1530.529.556.3446
MAGEFinetuning Strategy=NPO, Backbone=Llama-2-7b-chat2026.0441.2324.383316.566.3849
RWKUFinetuning Strategy=GA+KL, Backbone=Llama-2-7b-chat2026.0440.521.630.537.186.3051
RWKUFinetuning Strategy=GA+GD, Backbone=Llama-2-7b-chat2026.0439.8421.9130.537.086.2991
ELUDeFinetuning Strategy=NPO, Backbone=Llama-2-7b-chat2026.0439.7718.6735.2515.266.393
RWKUFinetuning Strategy=GA, Backbone=Llama-2-7b-chat2026.0439.6923.4630.2536.756.3109