arXiv Machine Learning By Junqing Lin, Jingwei Sun, Guangzhong Sun

DeltaLog: Deferred Materialization of Recurrent States for Linear Attention Decoding

Read the original on arXiv Machine Learning →

arXiv:2608. 15533v1 Announce Type: cross Abstract: Linear attention models eliminate the quadratic prefix computation and context-growing KV cache of softmax attention by replacing pairwise token interactions with recurrent state updates.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv Machine Learning.

arXiv Machine Learning
Jun 10

SinkRec: Mitigating Semantic State Sink in Long Sequence Recommendation with Memory-Conditioned Gated Delta Networks

arXiv:2606. 09888v1 Announce Type: new Abstract: Linear attention provides an efficient backbone for long-sequence recommendation by avoiding the quadratic cost of standard Transformers, but its compressed recurrent state can be dominated by repetitive behavior patterns.

By Zhuang Zhuang, Zhipeng Wei, Ji Dai, Jie Chen, Fei Pan, Peng Jiang, Kun Gai
arXiv Machine Learning
Aug 4

Bole: Efficient Tree Speculation for Hybrid-Attention Language Models

arXiv:2608. 01651v1 Announce Type: cross Abstract: Hybrid-attention large language models combine full attention with recurrent linear attention to reduce long-context inference costs, yet their autoregressive decoding remains memory-bound.

By Li Wang, Yi Su, Xiabao Wu, Chiran You, Yongchao Liu, Zhan Qiu, Juelu Zhang, Jiajun Zheng, Fangxin Liu, Jie Zhang, Chen Tian, Chengying Huan