Search papers, labs, and topics across Lattice.
Chinese University of Hong Kong, Shenzhen {yanguang.liu, zh296}@njit.edu, hzhang931@gatech.edu, 123090629@link.cuhk.edu.cn, mengnandu@cuhk.edu.cn *Co鈥揻irst authors Abstract Financial prediction from long documents involves significant challenges, as actionable signals are often sparse and obscured by noise, and the optimal LLM for generating embeddings varies across tasks and time periods. In this paper, we propose FinAnchor (Financial Anchored Representations), a lightweight framework that integrates embeddings from multiple LLMs without fine-tuning the underlying models. FinAnchor addresses the incompatibility of feature spaces by selecting an anchor embedding space and learning linear mappings to align representations from other models into this anchor. These aligned features are then aggregated to form a unified representation for downstream prediction. Across multiple financial NLP tasks, FinAnchor consistently outperforms strong single-model baselines and standard ensemble methods, demonstrating the effectiveness of anchoring heterogeneous representations for robust financial prediction. FinAnchor: Aligned Multi-Model Representations for Financial Prediction Zirui He1,* Huopu Zhang2,* Yanguang Liu1 Sirui Wu3 Mengnan Du3
2
0
6
LLMs respond to increasingly difficult out-of-distribution inputs by activating sparser representations in their last hidden states, revealing a quantifiable relationship between task difficulty and neural activity.
Forget complex fine-tuning: FinAnchor unlocks surprisingly robust financial predictions by simply aligning embeddings from multiple off-the-shelf LLMs.
