arXiv Computation and Language By Xixian Yong, Siyuan Chang, Yingying Zhang, Xian Wu, Xiao Zhou

Controllable Affective Generation via Latent Vector Steering

Read the original on arXiv Computation and Language →

The paper introduces EmoVec, a lightweight framework that enables controllable affective generation in large language models by steering latent vectors. EmoVec identifies emotion-specific directions from paired neutral and emotion-conditioned responses using contrastive activation addition, then refines these directions through task-specific debiasing and principal subspace removal. During inference, the refined vectors are injected into the final residual stream with static or scenario-adaptive scaling, allowing continuous control over emotional intensity without updating model weights, and experiments across three LLMs and eight emotions demonstrate improved emotional salience while preserving semantic content, fluency, and coherence.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

Hugging Face Trending Papers
Aug 11

E$^3$mo-Bench: A Scalable Benchmark for Multimodal Evoked and Expressed Emotion Understanding via Bayesian Pairwise Alignment

Understanding both expressed and evoked emotions is critical for multimodal large language models (MLLMs) to achieve comprehensive affect-aware interactions. However, existing benchmarks typically examine expressed and evoked emotions in isolation or are constrained to coarse-grained and incomplete affective characterizations.

arXiv AI
1d ago

EmoStance: Response-Side Affective-Orientation Control for Empathetic Response Generation via Emoji Weak Supervision

The paper introduces EmoStance, a method for controlling the affective orientation of empathetic responses by leveraging weak supervision from emoji distributions. It builds the EmojiDialogue dataset, extending EmpatheticDialogues with emoji votes and confidence scores, and uses a frozen instruction‑tuned LLM steered by continuous prefix embeddings to generate responses that align with the listener’s stance. In blind pairwise evaluations, EmoStance achieves a 62.2% decisive win rate, notably improving contextual specificity and perceived responsiveness compared to baselines.

By Ziyuan Jin, Yuxuan Ge, Zheng Tian