Table. 2 Screening explanations on cases with emotional disorders

Model	BERTScore	BLEU 2-gram	ROUGE-1	ROUGE-2	ROUGE-L
EmoScan	0.9408	0.0660	0.3951	0.1132	0.2086
Mistral-7B (zero-shot)	0.6897	0.0252	0.2204	0.0539	0.1214
Mistral-7B (few-shot)	0.6110	0.0103	0.1686	0.0341	0.1041
Mistral-7B (CoT)	0.7259	0.0227	0.2258	0.0535	0.1330
Mistral-7B (few-shot + CoT)	0.4471	0.0075	0.0964	0.0166	0.0591
GPT-4 (zero-shot)	0.8968	0.0391	0.3248	0.0752	0.1674
GPT-4 (few-shot)	0.9188	0.0364	0.3191	0.0762	0.1661
GPT-4 (CoT)	0.9259	0.0451	0.3301	0.0860	0.1857
GPT-4 (few-shot + CoT)	0.9268	0.0414	0.3276	0.0871	0.1696
Llama3 (zero-shot)	0.8775	0.0555	0.3737	0.1135	0.2072
Llama3 (few-shot)	0.9321	0.0571	0.3655	0.1079	0.2101
Llama3 (CoT)	0.9309	0.0608	0.3724	0.1085	0.2086
Llama3 (fewshot + CoT)	0.9218	0.0543	0.3484	0.1051	0.2090

CoT chain of thought. Bold numbers indicate the highest performance in the respective category.

Quick links

Search