Dimensional Aspect Sentiment Regression on SemEval Task 3 Track A Subtask 1 2026 (test)
1.2676RMSEVA (eng-res)ICT-NLP System
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ICT-NLP SystemLearning Protocol=Supervised Fine-Tuning2026.05 | 1.2676 | 1.3098 | 0.6289 | 0.8331 | 1.442 | 1.8596 | 1.475 | 0.9256 | 0.6553 | 0.4892 | 1.0886 | |
| GPT-OSS 120BLearning Protocol=Supervised Fine-Tuning, Model Parameters=120B2026.05 | 1.4605 | 1.5269 | 0.7188 | 1.0188 | 1.4775 | 1.7153 | 1.5166 | 1.0349 | 0.8032 | 0.6511 | 1.1924 | |
| Kimi K2 ThinkingLearning Protocol=One-Shot Learning2026.05 | 2.1461 | 2.1893 | 1.7553 | 1.6396 | 1.7768 | 1.938 | 1.7805 | 1.8959 | 1.644 | 1.9652 | 1.8731 | |
| Kimi K2 ThinkingLearning Protocol=Zero-Shot Learning2026.05 | 2.3432 | 2.6546 | 2.3294 | 2.3379 | 2.063 | 2.3636 | 2.0782 | 2.623 | 2.0426 | 2.9662 | 2.3802 | |
| Qwen-3 14BLearning Protocol=Supervised Fine-Tuning, Model Parameters=14B2026.05 | 2.6427 | 2.8089 | 2.2906 | 1.8964 | 2.1528 | 2.6367 | 2.2121 | 2.0073 | 1.7706 | 1.4707 | 2.1889 |