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

General Reasoning on OOD Performance Suite (ARC-c, GPQA*, MMLU-Pro) (test)

91.4ARC-c Score

On-Policy

Updated 1mo ago

Evaluation Results

Method
On-Policy 2025.10	91.4	19.2	57.6	56
ExGRPO 2025.10	91.1	36.9	58.6	62.2
ExGRPO 2025.10	88.9	15.2	52.1	52
OLR 2026.04	88.9	37.9	59.5	62.1
GRPO 2026.04	88.1	29.8	56.2	58
OLR 2026.04	88.1	36.4	57.6	60.7
OLR 2026.04	87.4	29.8	57.3	58.2
GRPO w/ Confidence Penalty 2026.04	87.4	31.3	56.6	58.4
TTRL 2026.04	86.9	30.3	56	57.7
OLR 2026.04	86.8	38.4	57.8	61
OLR 2026.04	86.8	35.4	59.1	60.4
GRPO w/ OLR 2026.04	86.8	38.4	57.8	61
On-Policy 2025.10	86.7	0	48.1	44.9
GRPO 2026.04	86.7	28.8	56	57.2
GRPO 2026.04	86.7	28.8	56	57.2
GRPO w/ Label Smoothing 2026.04	86.5	29.8	57.4	57.9
GRPO 2026.04	86.3	37.4	59	60.9
Co-reward 2026.04	86.1	29.3	55.8	57.1
OLR 2026.04	85.9	30.3	55.3	57.2
OLR 2026.04	85.4	33.3	56.7	58.5
DAPO + EchoRL 2026.05	85.2	42.9	54.7	60.9
Instruct 2025.10	84.7	24.7	54.7	54.7
GRPO + EchoRL 2026.05	84.7	37.4	53	58.4
GRPO 2026.04	84.6	30.3	55.5	56.8
LUFFY + EchoRL 2026.05	83.6	45.3	54.1	61
GRPO 2026.04	83.4	26.3	54.5	54.7
GRPO w/ Confidence Penalty 2026.04	82.9	32.3	53.3	56.2
UFT + EchoRL 2026.05	82.7	43.4	53.5	59.9
UFT 2026.05	82.2	38.9	49.6	56.9
DAPO 2026.05	82.1	38.9	51.8	57.6
GRPO 2026.05	81.8	39.9	45.2	55.6
GRPO 2026.04	80.8	25.3	51.7	52.6
LUFFY 2026.05	80.5	39.9	53	57.8
OLR 2026.04	80.1	31.8	52.4	54.8
OLR 2026.04	78.5	28.3	54.8	53.9
GRPO w/ OLR 2026.04	78.5	28.3	54.8	53.9
GRPO w/ Random Select 2026.04	78.1	31.8	55	55
GRPO 2026.04	76.5	27.3	51.5	51.8
SFT 2026.05	75.2	24.7	42.7	47.5
GRPO 2026.04	73.9	24.7	48.6	49.1
GRPO 2026.04	73.9	24.7	48.6	49.1
GRPO w/ Samll-loss Select 2026.04	73.9	30.3	50.7	51.6
PRIME-Zero 2026.05	73.3	18.2	32.7	41.4
Qwen2.5-Math-7B-Instruct 2026.05	70.3	24.7	34.1	43
GRPO 2026.04	66.2	24.7	49.2	46.7
OpenReasoner-Zero 2026.05	66.2	29.8	58.7	51.6
ExGRPO 2025.10	59.3	30.8	30.4	40.2
ExGRPO 2025.10	59	7.1	26.2	30.8
On-Policy 2025.10	57.5	28.3	30.7	38.9
GRPO w/ Label Smoothing 2026.04	55.5	24.2	47.4	42.4
GRPO w/ Random Select 2026.04	45.6	14.6	46.3	35.5
Instruct 2025.10	42.4	0	40.5	27.6
Qwen3-4B-Base 2026.04	35.8	14.1	33.3	27.7
Qwen3-4B-Base 2026.04	35.8	14.1	33.3	27.7
GRPO 2026.04	34.2	12.6	30.5	25.8
SFT-KL 2026.05	34.1	23.2	31.5	29.6
GRPO w/ Samll-loss Select 2026.04	32.3	9.6	26.7	22.9
OLR 2026.04	31.1	16.2	32.1	26.5
SimpleRL-Zero 2026.05	30.2	23.2	34.5	29.3
PivotTrace 2026.06	26.9	23.4	46.3	32.2
Random 2026.06	25.5	22.2	43.2	30.3
Self-certainty 2026.04	24.3	12.1	30.6	22.3
Consistency 2026.06	23.3	20.3	40.7	28.1
Seq-entropy 2026.04	18.6	6.6	28.4	17.9
Qwen2.5-Math-7B 2026.05	18.2	11.1	16.9	15.4
Token-entropy 2026.04	17	7.6	28.9	17.8
Entropy 2026.06	16.5	23.1	47.5	29
CoT-Kinetics 2026.06	15.6	18.9	42.1	25.5
Self-Certainty 2026.06	15.6	20.2	42.9	26.2
CoE 2026.06	8.7	18.4	46.8	24.6
Base 2025.10	3.1	3	2.3	2.8
On-Policy 2025.10	0.6	0	3	1.2
Base 2025.10	0.2	0	3.6	1.3