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

Instruction Following on FollowBench

79HSR

ImpRIF-32B

Updated 5d ago

Evaluation Results

Method
ImpRIF-32B 2026.02	79	82.39	-	-
ImpRIF-32B 2026.02	78.72	82.51	-	-
Qwen3-235B-A22B 2026.02	76.27	81.81	-	-
ImpRIF-8B 2026.02	74.94	81	-	-
ImpRIF-8B 2026.02	74.77	80.33	-	-
ImpRIF-4B 2026.02	72.7	78.93	-	-
ImpRIF-4B 2026.02	72.12	79.67	-	-
ImpRIF-32B 2026.02	71.73	78.64	-	-
Qwen3-32B 2026.02	71.58	78.58	-	-
Qwen2.5-72B 2026.02	71.15	79.39	-	-
Qwen3-14B 2026.02	70.74	78.98	-	-
Ministral3-14B 2026.02	70.52	76.9	-	-
GPT-4o 2026.05	70.4	-	-	-
ImpRIF-8B 2026.02	69.09	77.68	-	-
Claude-Opus-4.7 2026.05	68.9	-	-	-
Llama3.1-70B 2026.02	67.94	79.62	-	-
Qwen3-8B 2026.02	67.89	77	-	-
MuSC 2025.02	66.9	75.11	-	-
MuSC 2025.02	66.71	74.84	-	-
R1-0528-Qwen3-8B 2026.05	66.2	-	-	-
Qwen3-4B 2026.02	65.56	74.6	-	-
R1-0528-Qwen3-8B 2026.05	64.8	-	-	-
R1-0528-Qwen3-8B 2026.05	64.7	-	-	-
ImpRIF-4B 2026.02	64.32	76.47	-	-
Self-Reward w/ BSM 2025.02	64.3	73.84	-	-
Self-Reward w/ GPT-4 2025.02	64.02	73.26	-	-
Self-Reward w/ BSM 2025.02	63.96	73.78	-	-
ISHEEP 2025.02	63.54	73.21	-	-
ISHEEP 2025.02	62.77	72.86	-	-
MuSC 2025.02	62.73	73.09	-	-
MuSC 2025.02	62.6	72.57	-	-
LLaMA-3-8B-Instruct 2025.02	62.39	73.07	-	-
QwQ-32B 2026.05	62.2	-	-	-
Self-Reward w/ GPT-4 2025.02	62.18	73.34	-	-
Self-Reward 2025.02	61.2	72.22	-	-
Self-Reward 2025.02	60.88	72.17	-	-
R1-0528-Qwen3-8B 2026.05	60.4	-	-	-
Self-Correct 2025.02	60.11	70.94	-	-
Qwen2-7B-Instruct 2025.02	59.81	71.69	-	-
Qwen2.5-7B-Instruct 2026.05	59	-	-	-
Self-Reward w/ BSM 2025.02	58.02	70.62	-	-
Self-Reward w/ BSM 2025.02	57.83	70.53	-	-
Self-Supervised-7B 2026.05	57.5	-	-	-
Llama-3.1-8B-Instruct 2026.05	57.3	-	-	-
ISHEEP 2025.02	57.01	69.88	-	-
VERIF-8B 2026.05	56.9	-	-	-
Llama-3.1-8B-Instruct 2026.05	56.9	-	-	-
Qwen2.5-7B-Instruct 2026.05	56.8	-	-	-
Llama-3.1-8B-Instruct 2026.05	56.8	-	-	-
Qwen2.5-7B-Instruct 2026.05	56.6	-	-	-
Self-Reward 2025.02	56.45	70	-	-
RAIF-7B 2026.05	56.2	-	-	-
SPAR-8B-DPO 2026.05	56.1	-	-	-
ISHEEP 2025.02	55.52	69.62	-	-
Self-Reward 2025.02	55.36	69.71	-	-
Qwen2.5-7B-Instruct 2026.05	55.1	-	-	-
Self-Correct 2025.02	54.38	67.19	-	-
Distill-Qwen-14B 2026.05	54	-	-	-
Llama-3.1-8B-Instruct 2026.05	53.8	-	-	-
Distill-Qwen-14B 2026.05	53.2	-	-	-
Distill-Qwen-14B 2026.05	52.7	-	-	-
Self-Correct 2025.02	51.98	67.89	-	-
Distill-Qwen-14B 2026.05	51.2	-	-	-
RM-Distiller 2026.01	50.09	63.31	-	-
SFT 2025.02	50.06	66.48	-	-
Conifer-7B-DPO 2026.05	50	-	-	-
RM-Distiller 2026.01	49.95	63.14	-	-
RM-Distiller 2026.01	49.62	63.15	-	-
Self-Correct 2025.02	49.47	66.35	-	-
RM-Distiller 2026.01	49.45	62.9	-	-
Crab-7B-DPO 2026.05	49.4	-	-	-
BT Classifier 2026.01	48.59	62.48	-	-
RM-Distiller 2026.01	48.42	62.05	-	-
BT Classifier 2026.01	48.33	62.15	-	-
BT Classifier 2026.01	48.15	62.1	-	-
BT Classifier 2026.01	47.9	61.88	-	-
RM-Distiller 2026.01	47.78	61.77	-	-
SFT 2025.02	47.36	64.67	-	-
BT Classifier 2026.01	46.85	61.2	-	-
BT Classifier 2026.01	46.2	60.95	-	-
Baseline 2026.01	41.04	57.39	-	-
Qwen2.5-1.5B-Instruct 2026.05	36.6	-	-	-
Qwen2.5-1.5B-Instruct 2026.05	36.5	-	-	-
Qwen2.5-1.5B-Instruct 2026.05	36	-	-	-
Qwen2.5-1.5B-Instruct 2026.05	34.6	-	-	-
Instruct 2026.01	-	-	54.2	-
SFT 2026.01	-	-	53.2	-
SFT 2026.01	-	-	55.3	-
GRPO (Random) 2026.01	-	-	17.5	-
GRPO (BLEU) 2026.01	-	-	49.8	-
GRPO (RM) 2026.01	-	-	53.7	-
GRPO (GRM) 2026.01	-	-	54.3	-
GRPO (RLPR) 2026.01	-	-	54.4	-
DPO 2026.01	-	-	53.6	-
GRPO (RLVRR) 2026.01	-	-	56.2	-
Gemini-2.5-Flash 2026.05	-	-	-	86
API Prompting (Rubric+Judge pairwise) 2026.05	-	-	-	83.2
API Prompting (Rubric+Judge pointwise) 2026.05	-	-	-	60.5
API Prompting (direct Judge) 2026.05	-	-	-	81.7
RM-R1-14B (Qwen-2.5-Inst) 2026.05	-	-	-	84

Showing 100 of 118 rows