List your company
ResearchFormerly WizWand

Emotion Detection on DailyDialog (test)

0.6034Micro-F1CoMPM
0.2199520.3195010.419050.518599Sep 24, 2019Oct 13, 2020Nov 2, 2021Nov 22, 2022Dec 12, 2023Dec 31, 2024Jan 20, 2026
Updated 8mo ago

Evaluation results

MethodLinks
CoMPMEvaluation Protocol=Average of 3 runs2021.080.6034—53.15
PsychologicalExternal Data Usage=Leverages structured external data2021.080.5975—51.95
CoMPM(k)Evaluation Protocol=Average of 3 runs2021.080.5909—52.33
CoMPM(f)Evaluation Protocol=Average of 3 runs2021.080.5902—51.67
CoMEvaluation Protocol=Average of 3 runs2021.080.5863—51.17
COSMICBase=RoBERTa2020.100.5848—51.05
COSMICExternal Data Usage=Leverages structured external data2021.080.5848—51.05
ToDKATExternal Data Usage=Leverages structured external data2021.080.5847—52.56
COSMIC w/o Listener CSKBase=RoBERTa, Ablation=w/o Listener CSK2020.100.5828—48.67
COSMIC w/o Speaker CSKBase=RoBERTa, Ablation=w/o Speaker CSK2020.100.5745—50.18
RoBERTa DialogueRNNBase=RoBERTa2020.100.5732—49.65
RoBERTa DialogueRNN2021.080.5732—49.65
PMEvaluation Protocol=Average of 3 runs2021.080.5623—49.08
COSMIC w/o Speaker, Listener CSKBase=RoBERTa, Ablation=w/o Speaker, Listener CSK2020.100.5616—48.68
DialogueRNNBase=GloVe2020.100.5595—41.8
CoMPM(s)Evaluation Protocol=Average of 3 runs2021.080.5581—48.86
RoBERTaBase=RoBERTa2020.100.5516—48.2
RGAT+P2021.080.5431——
KET2019.090.5337——
KETBase=GloVe2020.100.5337——
KETExternal Data Usage=Leverages structured external data2021.080.5337——
BERT_BASEMax Tokens=1002019.090.5312——
KET_StdAttn2019.090.5254——
KET_SingleSelfAttn2019.090.5192——
DialogueRNNNumber of Speakers=22019.090.5065——
CNNBase=GloVe2020.100.5032—36.87
CNN+cLSTM2019.090.5024——
cLSTM2019.090.499——
CNNContext=Utterance-level without context2019.090.4934——
APOLOBackbone=Qwen3-32B, Prompt Optimization Method=APOLO2026.010.3524—83.51
APOLOBackbone=GPT-5-mini, Prompt Optimization Method=APOLO2026.010.3411—82.34
OPROBackbone=Qwen3-32B, Prompt Optimization Method=Optimization by PROmpting2026.010.3233—80.21
PE2Backbone=Qwen3-32B, Prompt Optimization Method=Prompt Engineer2026.010.3184—79.75
OPROBackbone=GPT-5-mini, Prompt Optimization Method=Optimization by PROmpting2026.010.3125—79.12
ProTeGiBackbone=Qwen3-32B, Prompt Optimization Method=Prompt Optimization2026.010.3113—79.01
APOLOBackbone=DeepSeek-V3, Prompt Optimization Method=APOLO2026.010.3102—57.21
PE2Backbone=GPT-5-mini, Prompt Optimization Method=Prompt Engineer2026.010.3078—78.68
APEBackbone=Qwen3-32B, Prompt Optimization Method=Automatic Prompt Engineer2026.010.3055—78.4
ProTeGiBackbone=GPT-5-mini, Prompt Optimization Method=Prompt Optimization2026.010.3011—77.98
CoT(FS)Backbone=Qwen3-32B, Prompt Optimization Method=Chain-of-Thought (Few-Shot)2026.010.2998—77.8
APEBackbone=GPT-5-mini, Prompt Optimization Method=Automatic Prompt Engineer2026.010.2956—77.42
CoT(FS)Backbone=GPT-5-mini, Prompt Optimization Method=Chain-of-Thought (Few-Shot)2026.010.2902—76.85
CoT(ZS)Backbone=Qwen3-32B, Prompt Optimization Method=Chain-of-Thought (Zero-Shot)2026.010.2902—76.89
OPROBackbone=DeepSeek-V3, Prompt Optimization Method=Optimization by PROmpting2026.010.2831—53.19
CoT(ZS)Backbone=GPT-5-mini, Prompt Optimization Method=Chain-of-Thought (Zero-Shot)2026.010.2815—75.98
OriginBackbone=Qwen3-32B, Prompt Optimization Method=None2026.010.2811—76.01
PE2Backbone=DeepSeek-V3, Prompt Optimization Method=Prompt Engineer2026.010.2782—51.55
OriginBackbone=GPT-5-mini, Prompt Optimization Method=None2026.010.2737—75.26
ProTeGiBackbone=DeepSeek-V3, Prompt Optimization Method=Prompt Optimization2026.010.2701—48.81
APEBackbone=DeepSeek-V3, Prompt Optimization Method=Automatic Prompt Engineer2026.010.2634—46.21
CoT(FS)Backbone=DeepSeek-V3, Prompt Optimization Method=Chain-of-Thought (Few-Shot)2026.010.2568—43.71
CoT(ZS)Backbone=DeepSeek-V3, Prompt Optimization Method=Chain-of-Thought (Zero-Shot)2026.010.2455—39.87
OriginBackbone=DeepSeek-V3, Prompt Optimization Method=None2026.010.2347—36.16
BERT DCR-NetBase=BERT2020.10——48.9