Multimodal Sentiment Analysis on SIMS V2
85.62Accuracy (2-class)HyperEmo-ChatGLM3-6B
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| HyperEmo-ChatGLM3-6BBackbone=ChatGLM3-6B2026.05 | 85.62 | 85.87 | 77.78 | 0.279 | 0.748 | — | — | — | |
| HyperEmo-Qwen-1.8BBackbone=Qwen-1.8B2026.05 | 82.89 | 82.36 | 75.9 | 0.293 | 0.702 | — | — | — | |
| EGMFBackbone=GLM4-9B2026.01 | 82.57 | 82.43 | 74.7 | 0.284 | 0.72 | — | — | — | |
| AV-MC2026.05 | 82.5 | 82.55 | 74.54 | 0.297 | 0.732 | — | — | — | |
| EGMFBackbone=GLM3-6B2026.01 | 81.56 | 81.13 | 73.09 | 0.284 | 0.733 | — | — | — | |
| MMIM2026.05 | 80.95 | 80.97 | 72.28 | 0.316 | 0.707 | — | — | — | |
| MISA2026.05 | 80.53 | 80.63 | 70.5 | 0.314 | 0.725 | — | — | — | |
| MAG-BERT2026.01 | 79.79 | 79.78 | 71.87 | 0.334 | 0.691 | — | — | — | |
| MAG-BERT2026.05 | 79.79 | 79.78 | 71.87 | 0.334 | 0.691 | — | — | — | |
| ALMTimplementation=reproduced from open-source codebase2025.11 | 79.59 | 79.51 | — | 0.308 | 0.7 | 71.86 | 52.9 | — | |
| MODS2025.11 | 79.59 | 79.53 | — | 0.297 | 0.712 | 73.69 | 55.51 | — | |
| CENetsource=Mao et al. 20222025.11 | 79.56 | 79.63 | — | 0.31 | 0.699 | 73.1 | 53.04 | — | |
| MulT2026.01 | 79.5 | 79.59 | 69.61 | 0.317 | 0.703 | — | — | — | |
| MulT2026.05 | 79.5 | 79.59 | 69.61 | 0.317 | 0.703 | — | — | — | |
| MulTimplementation=reproduced from open-source codebase2025.11 | 79.3 | 79.43 | — | 0.304 | 0.705 | 72.63 | 53.29 | — | |
| Self-MMsource=Mao et al. 20222025.11 | 79.11 | 79.05 | — | 0.322 | 0.678 | 72.34 | 53 | — | |
| Self-MM2026.01 | 79.01 | 78.89 | 71.87 | 0.335 | 0.64 | — | — | — | |
| Self-MM2026.05 | 79.01 | 78.89 | 71.87 | 0.335 | 0.64 | — | — | — | |
| DMDimplementation=reproduced from open-source codebase2025.11 | 78.87 | 79.21 | — | 0.305 | 0.702 | 72.01 | 53.18 | — | |
| DTNimplementation=reproduced from open-source codebase2025.11 | 78.29 | 78.12 | — | 0.302 | 0.701 | 72.56 | 53.71 | — | |
| HyperEmo-LLaMA2-7BBackbone=LLaMA2-7B2026.05 | 78.2 | 78.32 | 71.73 | 0.353 | 0.574 | — | — | — | |
| TFNsource=Mao et al. 20222025.11 | 77.95 | 77.74 | — | 0.329 | 0.64 | 70.21 | 51.93 | — | |
| MIMimplementation=reproduced from open-source codebase2025.11 | 77.56 | 78.56 | — | 0.31 | 0.694 | 71.45 | 52.87 | — | |
| LMF2026.05 | 77.05 | 77.02 | 69.34 | 0.343 | 0.638 | — | — | — | |
| EGMFBackbone=llama2-7B2026.01 | 77.04 | 76.93 | 70.85 | 0.364 | 0.579 | — | — | — | |
| TFN2026.05 | 76.51 | 76.31 | 66.27 | 0.323 | 0.667 | — | — | — | |
| LMFsource=Mao et al. 20222025.11 | 74.18 | 73.88 | — | 0.367 | 0.557 | 64.9 | 47.79 | — | |
| EGMFBackbone=llama3-8B2026.01 | 57.74 | 42.27 | 63.35 | 0.398 | 0.64 | — | — | — | |
| AffectAgentModality=A, V, T, Backbone=Video-LLaMA2026.04 | — | — | — | — | — | — | — | 60.4 | |
| AffectAgentModality=A, V, T, Backbone=VideoChat2026.04 | — | — | — | — | — | — | — | 61.53 | |
| AffectAgentModality=A, V, T, Backbone=ChatBridge2026.04 | — | — | — | — | — | — | — | 62.29 | |
| AffectAgentModality=A, V, T, Backbone=Video-LLaMA 22026.04 | — | — | — | — | — | — | — | 63.6 | |
| AffectAgentModality=A, V, T, Backbone=PandaGPT2026.04 | — | — | — | — | — | — | — | 70.31 | |
| AffectAgentModality=A, V, T, Backbone=Emotion-LLaMA2026.04 | — | — | — | — | — | — | — | 79.35 | |
| AffectAgentModality=A, V, T, Backbone=AffectGPT2026.04 | — | — | — | — | — | — | — | 87.33 | |
| AffectGPTModality=A, V, T2026.04 | — | — | — | — | — | — | — | 86.18 | |
| ChatBridgeModality=A, V, T2026.04 | — | — | — | — | — | — | — | 56.55 | |
| Emotion-LLaMAModality=A, V, T2026.04 | — | — | — | — | — | — | — | 77.23 | |
| PandaGPTModality=A, V, T2026.04 | — | — | — | — | — | — | — | 67.23 | |
| Video-LLaMAModality=A, V, T2026.04 | — | — | — | — | — | — | — | 56.32 | |
| Video-LLaMA 2Modality=A, V, T2026.04 | — | — | — | — | — | — | — | 58.69 | |
| VideoChatModality=A, V, T2026.04 | — | — | — | — | — | — | — | 57.65 |