Shot-Language Understanding on 12 In-Domain (ID) Tasks benchmark leaderboard · SOTA2 Research

Shot-Language Understanding on 12 In-Domain (ID) Tasks

79Composition Rule ScoreUNISHOT

Evaluation Results

Method	Links
UNISHOT2026.03		79	29	90	94	87.5	78.5	94	69	90.5	74	85	40	75.9	7
AGENTSHOTS2026.03		78.5	25.5	89.5	92.5	87.5	76	95.5	70.5	90	77.5	84	42.5	75.8	3
12 task-specific SFT Qwen3-VL-8B2026.03		78.4	32	85.5	87.5	86.5	68.5	96	71	87.5	72	84	39	74	3
ShotVL-7B2026.03		77.5	11.5	74	46	42.5	29	24.5	26	65	60	60	21	44.8	0
Gemini-3.0-Flash2026.03		4	12.5	85	65	59.5	28	44	45	70.5	55.5	71	29.5	47.5	0
Gemini-3.0-Pro2026.03		4	17	82.5	66	55.5	29.5	48.5	54.5	67	58	58.5	27	47.3	0
Qwen3-VL-8B2026.03		2	14.5	67	71.5	65	43	33	35	63	54	74.5	23	45.5	0
Gemini-2.5-Flash2026.03		0.5	16.5	77	67.5	46.5	16.5	42	51.5	69	38.5	65	21.5	42.7	0