Share your thoughts, 1 month free Claude Pro on usSee more

Reasoning on BBH, MMLU, ARC-C, and ThmQA (test)

64.66BBH

Teacher

Updated 7d ago

Evaluation Results

Method
Teacher 2026.05	64.66	78.22	89.9	32.47	66.31
Teacher 2026.05	57.72	70.9	83.58	18.1	57.58
Teacher 2026.05	52.57	66.17	81.23	26.12	56.52
HybKD 2026.05	46.53	69.05	81.23	23.82	55.16
HybKD 2026.05	46.53	69.05	81.23	23.82	55.16
Hybrid KD 2026.05	45.7	66.8	80	24	54.1
JS divergence 2026.05	45.5	64.68	78.85	22.27	52.83
JS divergence 2026.05	45.5	64.68	78.85	22.27	52.83
α-β divergence 2026.05	45.12	64.95	79.81	22.94	53.21
α-β divergence 2026.05	45.12	64.95	79.81	22.94	53.21
Adaptive KL 2026.05	44.71	64.69	79.23	22.25	52.72
Adaptive KL 2026.05	44.71	64.69	79.23	22.25	52.72
Reverse KL 2026.05	44.07	65.67	77.68	24.2	52.9
Reverse KL 2026.05	44.07	65.67	77.68	24.2	52.9
Soft KD 2026.05	44	65.6	78.1	22.9	52.6
Forward KL 2026.05	41.65	64.45	78.33	23.02	51.87
Hard KD 2026.05	41.52	65.76	78.75	23.75	52.45
Skew FKL 2026.05	41.39	64.67	77.75	23.77	51.89
Skew FKL 2026.05	41.39	64.67	77.75	23.77	51.89
Skew RKL 2026.05	41.22	63.95	76.91	23.67	51.44
Skew RKL 2026.05	41.22	63.95	76.91	23.67	51.44
Total Variation 2026.05	40.5	64.52	78.11	22.83	51.49
Total Variation 2026.05	40.5	64.52	78.11	22.83	51.49
Hard KD 2026.05	34.4	67.2	79.5	23.1	51
Skew RKL 2026.05	28.16	45.03	47.51	11.37	33.02
Skew RKL 2026.05	28.16	45.03	47.51	11.37	33.02
HybKD 2026.05	27.44	35.64	37.01	5	26.27
Total Variation 2026.05	26.74	44.35	46.76	10.2	32.01
Total Variation 2026.05	26.74	44.35	46.76	10.2	32.01
HybKD 2026.05	26.58	49.08	51.69	10.5	34.46
HybKD 2026.05	26.58	49.08	51.69	10.5	34.46
α-β divergence 2026.05	26.18	42.59	46.7	11.07	31.63
α-β divergence 2026.05	26.18	42.59	46.7	11.07	31.63
Adaptive KL 2026.05	26.04	41.85	45.89	11.33	31.28
Adaptive KL 2026.05	26.04	41.85	45.89	11.33	31.28
Skew FKL 2026.05	25.87	44.12	47.12	10.65	31.94
Skew FKL 2026.05	25.87	44.12	47.12	10.65	31.94
Skew RKL 2026.05	25.24	33	32.24	4.27	23.69
Adaptive KL 2026.05	25.18	33.86	33.74	4.85	24.41
JS divergence 2026.05	25.16	34.08	34.42	5.7	24.84
α-β divergence 2026.05	25.07	34.78	34.9	5.27	25.01
Skew FKL 2026.05	25.05	31.39	32.34	4.73	23.37
Reverse KL 2026.05	24.91	44.72	47.44	11	32.02
Reverse KL 2026.05	24.91	44.72	47.44	11	32.02
JS divergence 2026.05	24.55	43.31	44.78	11.82	31.12
JS divergence 2026.05	24.55	43.31	44.78	11.82	31.12
Forward KL 2026.05	24.42	43.19	46.84	10.33	31.2
Reverse KL 2026.05	23.69	31.63	32.08	3.6	22.75
Total Variation 2026.05	23.55	27.41	28.99	2.68	20.66
Student (No Distill) 2026.05	22.34	64.61	78.4	12.22	44.39
Forward KL 2026.05	22.07	33.13	33.41	4.37	23.25
Hard KD 2026.05	15.29	22.54	23.98	3.88	16.42
Student (No Distill) 2026.05	14.01	19.78	21.57	2.22	14.4
Hard KD 2026.05	10.27	36.83	40.58	11.35	24.76
Reverse KL 2026.05	9.52	24.77	27.58	1.26	15.78
HybKD 2026.05	7.78	25.94	26.07	5.24	16.26
α-β divergence 2026.05	7.26	25.49	26.07	4.34	15.79
JS divergence 2026.05	7.14	25.27	26.03	4.57	15.75
Skew FKL 2026.05	7.01	25.08	26.14	5.01	15.81
Student (No Distill) 2026.05	6.47	15.94	15.7	0.95	9.76
Total Variation 2026.05	6.29	25.07	25.83	1.29	14.62
Forward KL 2026.05	5.91	24.05	25	4.25	14.8
Skew RKL 2026.05	5.83	25.17	25.93	3.38	15.08
Adaptive KL 2026.05	5.47	24.76	24.07	4.23	14.63
Hard KD 2026.05	5.28	18.11	18.09	3.15	11.16
Student (No Distill) 2026.05	3.81	44.67	43.72	7.78	24.99