Image Quality Evaluation on Storyeval (test) benchmark leaderboard · SOTA2 Research

Image Quality Evaluation on Storyeval (test)

69.9Image ScorePRQ INT4

Evaluation Results

Method
PRQ INT42026.03	69.9	72.4	18.8	—	69.9
PRQ INT22026.03	69.8	73.3	17.9	20.66	69.8
QuaRot KV INT42026.03	68.7	68.5	21.7	19.25	68.9
FlowCache Prune INT42026.03	68.2	46.5	49	—	68
FlowCache Soft-Prune INT42026.03	68	51.8	41.6	—	67.9
RTN INT4 Recent22026.03	68	72.1	18.7	—	68.4
RTN INT4 Refresh2026.03	67.8	65.4	25.2	18.55	67.9
RTN INT42026.03	67.4	66.1	24.5	18.66	67.5
Age-Tier INT42026.03	67.4	66.1	24.5	18.66	67.6
QuaRot KV INT4 Recent22026.03	66.6	70.7	22.2	—	67
FlowCache Native Soft-Prune INT42026.03	65.7	43.6	52.3	11.94	65.7
TPTQ INT22026.03	65.4	61.5	30.1	17.03	65.8
KIVI INT4 Refresh2026.03	64.5	41.4	56.9	12.49	64.1
KIVI INT42026.03	63.5	42.4	57.5	12.41	63.5
SpatialMixed fg RTN INT 4 / bg RTN INT 42026.03	60.6	58.7	34.9	—	60.7
QAQ INT22026.03	57.9	32.4	63.5	11.79	58.5
QAQ INT42026.03	56.7	23.8	71.9	11.03	57.1
FlowCache Soft-Prune INT22026.03	53.2	49.7	49.5	14.18	53.6
KIVI INT22026.03	53.1	24.3	73.5	—	52.7
FlowCache Prune INT22026.03	51.6	49.2	50.6	14.26	51.6
FlowCache Adaptive INT22026.03	49.8	45.1	52.8	—	49.6
FlowCache Hybrid INT22026.03	49.2	47.9	51.2	13.95	49.4
QuaRot KV INT22026.03	47.7	44.3	53.2	13.01	48
Age-Tier INT22026.03	46.9	46.3	52.3	13.67	47.3
RTN INT22026.03	46.4	45.9	52.6	13.57	47.1
SpatialMixed fg KIVI INT 4 / bg KIVI INT 22026.03	45.4	45.3	65.4	13.2	45.1
SpatialMixed fg RTN INT 4 / bg RTN INT 22026.03	42.1	43	58.7	12.64	41.9
Spatial Mixed (QuaRot fg, RTN bg)2026.03	40	44.4	59.9	12.97	39.8