Quantization and Hardware Architecture Co-Design for Matrix-Vector Multiplications of Large Language Models.
Wenjie LiAokun HuNingyi XuGuanghui HePublished in: IEEE Trans. Circuits Syst. I Regul. Pap. (2024)
Keyphrases
- language model
- hardware architecture
- language modeling
- hardware implementation
- document retrieval
- n gram
- speech recognition
- information retrieval
- retrieval model
- probabilistic model
- language modelling
- hardware architectures
- query expansion
- associative memory
- vector space model
- test collection
- context sensitive
- relevance model
- pseudo relevance feedback
- singular value decomposition
- statistical language models
- field programmable gate array
- document ranking
- smoothing methods
- ad hoc information retrieval
- language models for information retrieval
- spoken term detection
- xilinx virtex
- feature vectors