An efficient sparse LSTM accelerator on embedded FPGAs with bandwidth-oriented pruning

• Main Authors: Li, Shiqing, Zhu, Shien, Luo, Xiangzhong, Luo, Tao, Liu, Weichen
• Other Authors: School of Computer Science and Engineering
• Format: Conference or Workshop Item
• Language: English
• Published: 2023
• Subjects: Engineering::Computer science and engineering; Engineering::Computer science and engineering::Hardware; Sparse LSTM; Pruning; Bandwidth
• Online Access: https://hdl.handle.net/10356/172603
• Institution: Nanyang Technological University

Long short-term memory (LSTM) networks have been widely used in natural language processing applications. Although over 80% weights can be pruned to reduce the memory requirement with little accuracy loss, the pruned model still cannot be buffered on-chip for small embedded FPGAs. Considering that w...