BitCluster: Fine-Grained Weight Quantization for Load-Balanced Bit-Serial Neural Network Accelerators.
Ang LiHuiyu MoWenping ZhuQiang LiShouyi YinShaojun WeiLeibo LiuPublished in: IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. (2022)
Keyphrases
- fine grained
- neural network
- uniform quantization
- coarse grained
- artificial neural networks
- access control
- adaptive quantization
- tightly coupled
- computing systems
- back propagation
- neural network model
- massively parallel
- quantization error
- quantization scheme
- successive approximation
- databases
- machine translation
- operating system
- general purpose
- data lineage