@inproceedings{modi-szymanski-2026-rpi,
    title = "{RPI} Team at {S}em{E}val-2026 Task 3: An {LLM}-Encoder Ensemble for Coarse-to-Fine Valence-Arousal Sentiment Prediction",
    author = "Modi, Mohammed Shahid  and
      Szymanski, Boleslaw",
    editor = "Kochmar, Ekaterina  and
      Ghosh, Debanjan  and
      North, Kai  and
      Komachi, Mamoru",
    booktitle = "Proceedings of the 20th {I}nternational {W}orkshop on {S}emantic {E}valuation (2026)",
    month = jul,
    year = "2026",
    address = "San Diego, California, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.semeval-1.403/",
    pages = "3218--3224",
    ISBN = "979-8-89176-414-9",
    abstract = "We present our coarse-to-fine Valence-Arousal (VA) ensemble system for subtask 1 of task 3 (DimABSA) which covers aspect-level VA prediction. We use a pair of trained Qwen 3 8B LoRA-tuned LLMs to predict coarse bins between 1 and 8, providing ordinal VA guidance signals along with distributional features. We then train an instruction-style, multilingual E5 encoder model with a multitask head using these LLM-derived guidance features to produce continuous VA predictions. At inference time, the same guidance signals are generated for the test set by the trained LLMs and fed into the trained encoder. This approach leverages the LLM as a high-level prior while relying on the encoder for precise calibration across languages and domains. Our system achieves an RMSEVA of 1.20 across six languages and five domains. We compare the joint VA model to separated valence and arousal models trained on coarsened ground truth data, showing that it outperforms them, particularly on arousal correlations."
}