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

Multi-modal Long-context Benchmarking on MileBench

57.23Task T Score

FlashCache

Updated 1mo ago

Evaluation Results

Method	Links
FlashCache 2025.11		57.23	59.08	16.1	20.17
SnapKV 2025.11		57.06	58.71	16.06	18.83
StreamingLLM 2025.11		57	56.15	2.5	20
H2O 2025.11		56.94	58.33	2.5	21
Full Cache 2025.11		56.51	65.21	27.19	18.17
FlashCache 2025.11		56.25	53.11	8.29	26.33
SnapKV 2025.11		55.69	49.07	6.25	19.83
FlashCache 2025.11		55.65	60.93	18.45	14.5
Full Cache 2025.11		55.59	69.17	27.35	14.17
SnapKV 2025.11		55.58	61.3	7.19	15.5
LOOK-M 2025.11		55.08	58.54	5.78	13.67
StreamingLLM 2025.11		54.71	59.23	6.56	15.83
H2O 2025.11		54.58	58.65	6.25	14.83
FlashCache 2025.11		54.55	56.19	10.63	14.83
MEDA 2025.11		54.24	57.05	5.47	13.67
H2O 2025.11		52.31	48.44	1.25	23.67
StreamingLLM 2025.11		51.94	42.34	1.25	26.83
LOOK-M 2025.11		51.91	58.99	2.81	24.67
MEDA 2025.11		51.86	50.01	2.65	26.33
MEDA 2025.11		51.74	52.95	5.16	15.33
LOOK-M 2025.11		51.69	53.67	5.31	16.33
SnapKV 2025.11		51.52	56.33	6.25	15.33
MEDA 2025.11		51.31	48.45	2.97	28.67
H2O 2025.11		50.47	52.77	5.31	11.33
LOOK-M 2025.11		49.97	45.77	2.5	27.83
StreamingLLM 2025.11		49.46	53.64	5.63	15
LOOK-M 2025.11		48.01	56.99	12.5	11.5
FlashCache 2025.11		47.09	58.56	20	12.83
H2O 2025.11		47.01	56.99	12.5	11.5
Full Cache 2025.11		46.97	65.33	27.03	11.67
SnapKV 2025.11		46.97	57.62	10.47	12.67
H2O 2025.11		46.84	58.51	15.94	11.83
LOOK-M 2025.11		46.84	58.51	15.94	11.83
StreamingLLM 2025.11		46.81	58.4	6.72	12.5
StreamingLLM 2025.11		46.52	57.05	5.16	12
SnapKV 2025.11		45.97	57.82	7.97	13
FlashCache 2025.11		45.51	51.12	15.56	13.33
MEDA 2025.11		44.23	52.76	14.06	12.17
MEDA 2025.11		38.4	48.25	10.31	13