Share your thoughts, 1 month free Claude Pro on usSee more

Emotion Classification on Emotion (Accuracy)

87.68Accuracy

Multi-head self-attn

Updated 1mo ago

Evaluation Results

Method	Links
Multi-head self-attn 2026.01		87.68
DeBERTa V3 Large 2026.01		87.65
Scoring attention 2026.01		84.58
RoBERTa Large 2026.01		84.16
SentriLlama 3.2 (3B) Instruct 2026.01		82.2
Text Output 2026.02		75.2
Direct pooling 2026.01		69.18
Text Output 2026.02		68.4
MULI (logits, Llama-3.2-3B) 2026.01		64.05
Text Output 2026.02		62.2
MIPIC 2026.04		58.65
Llama 3.2 (3B) Chain-of-Thought 2026.01		56.05
Unsupervised Probe 2026.02		54.59
Unsupervised Probe 2026.02		53.52
Zero-Shot Probe 2026.02		52.43
Unsupervised Probe 2026.02		51.34
Zero-Shot Probe 2026.02		50.15
Text Output 2026.02		49.8
Zero-Shot Probe 2026.02		46.73
Llama 3.2 (3B) Zero-shot 2026.01		44.55
SAE Probe 2026.02		43.64
SAE Probe 2026.02		38.32
SAE Probe 2026.02		35.76
Llama 3.2 (3B) Few-shot 2026.01		33.4
Unsupervised Probe 2026.02		21.32
SAE Probe 2026.02		17.69
Zero-Shot Probe 2026.02		17.37