HS
HongJin Song
cs.AIcs.SDeess.AScs.CLcs.LG
On Valency
published · living versionsW_kruxywyh·v1 · currentpublished
The Right Source Is Not Enough: Understanding History Interference in Speech-LLM Memory
with Weiqiao Shan, Runwu Shi, Yifei Wu, Yujin Wang +9
1 version
Preprints & journals
6 papers in the corpus · 2024–2026Rethinking Length-Based Training: Batch Composition and Loss Normalization in Speech Token Language Models2609.25890v1 · Hongjin Song, Runwu Shi, Weiqiao Shan et al.2026 · 0 citationsarXiv
REVE: Efficient Hallucination Correction for Large Audio-Language Models via Reused Encoder States2609.26028v1 · Hongjin Song, Jiasheng Kuang, Xinyu Yang et al.2026 · 0 citationsarXiv
Likelihood-Constrained Acoustic Reranking for Training-Free Hallucination Mitigation in LLM-Based ASR2608.30776v1 · Jiasheng Kuang, Linru Zheng, Hongjin Song et al.2026 · 0 citationsarXiv
Joint Residual Reweighting for Classifier Free Guidance in Flow-Matching Zero-Shot TTS2606.25672v3 · Runwu Shi, Yujin Wang, Hongjin Song et al.2026 · 0 citationsarXiv
Applying Ensemble Models based on Graph Neural Network and Reinforcement Learning for Wind Power Forecasting2501.16591v1 · Hongjin Song, Qianrun Chen, Tianqi Jiang et al.2025 · 1 citationarXiv
DST-GTN: Dynamic Spatio-Temporal Graph Transformer Network for Traffic Forecasting2404.11996v1 · Songtao Huang, Hongjin Song, Tianqi Jiang et al.2024 · 1 citationarXiv
Career total: 6 works. 6 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.