Syntactic Evaluation on SLING Chinese-language benchmark leaderboard · SOTA2 Research

Syntactic Evaluation on SLING Chinese-language

100Alt. Quest.gpt-3.5

Evaluation Results

Method	Links
gpt-3.52025.06		100	83.1	75.1	98.7	96.7	83	89.4
gpt-3.52025.06		100	94.1	81	98.7	99.1	91.7	94.1
Haiku2025.06		100	95.5	68.6	96	100	100	93.3
Haiku2025.06		100	96.1	63.7	97.1	99.6	99.7	92.7
Haiku2025.06		100	98.9	81.1	98	100	100	96.3
Haiku2025.06		100	98.9	85.4	99.1	100	100	97.2
gpt-4o2025.06		100	99.2	89.7	99.8	100	100	98.1
gpt-4o2025.06		100	99.7	88.4	100	99.6	100	98
gpt-4o2025.06		100	99.7	89.4	99.6	100	100	98.1
gpt-4o2025.06		100	99.7	89.6	100	100	100	98.2
Sonnet2025.06		100	98.8	90	96.9	95.3	100	96.8
Sonnet2025.06		100	99.5	84.7	100	100	100	97.4
Sonnet2025.06		100	99.3	86.1	100	100	100	97.6
Sonnet2025.06		100	99.6	89.4	99.3	100	100	98.1
Sonnet2025.06		100	99.9	90.6	100	100	100	98.4
Llama (GPb)2025.06		100	95	83	100	100	100	96.3
Llama (GPb+CT)2025.06		100	97	88.1	100	100	100	97.5
gpt-3.52025.06		99.3	88.7	72.1	94.2	98.7	83	89.3
gpt-4o2025.06		92.7	94.1	88.1	94.7	89.6	93.3	92.1
gpt-3.52025.06		88.7	93.9	82.8	86	99.8	88.7	90
gpt-4o2025.06		82.7	98.1	90	95.8	95.1	98	93.3
Sonnet2025.06		60	96.5	87.4	95.6	94.2	95.7	88.2
Haiku2025.06		46	77.9	84	88.7	75.1	84	75.9
Llama (Base)2025.06		40.7	83.7	87.3	92.2	92.9	97.7	82.4
Haiku2025.06		31.3	89.2	86.3	95.8	79.8	99	80.2
gpt-3.52025.06		18.7	73.2	82.7	97.6	73.8	98.3	74
gpt-3.52025.06		16.7	76.9	81.9	89.8	83.1	90.3	73.1
Llama (CT)2025.06		8.5	88.4	87	94.1	92.4	97	77.9