Tool Response Generation on GTM Multi-turn benchmark leaderboard · SOTA2 Research

Tool Response Generation on GTM Multi-turn

97.2Format ScoreGTM-1.5B

Evaluation Results

Method	Links
GTM-1.5B2025.12		97.2	90.1	94	98.5	99	86.7
Qwen2.5-7B-Instruct2025.12		91.4	94.9	95	97.2	99.5	85.6
Qwen2.5-14B-Instruct2025.12		88	96.4	96.7	98.6	99.7	84
Llama-3.2-3B-Instruct2025.12		82.4	74.5	72.9	83.5	97.9	58.3
Qwen2.5-3B-Instruct2025.12		80	85.4	83.4	93.9	99.2	65.8
InternLM2.5-20B2025.12		79.8	80.4	78.7	83.3	99.6	68.5
Qwen2.5-1.5B-Instruct2025.12		74.1	74.5	68.9	84.8	98.4	53
Qwen2.5-0.5B-Instruct2025.12		66.7	47	43.2	70.4	93.4	33.4
InternLM2.5-7B2025.12		63.6	48.9	40.9	51.2	98.9	32.4
Llama-3.2-1B-Instruct2025.12		38.3	30.7	21.9	45.8	95.3	13.4
InternLM2.5-1.8B2025.12		34.8	19.3	12.4	24.8	94.8	5.2