Fast Vocabulary Transfer for Language Model Compression

arXiv:2402.09977v1 Announce Type: cross
Abstract: Real-world business applications require a trade-off between language model performance and size. We propose a new method for model compression that relies on vocabulary transfer. We evaluate the method on various vertical domains and downstream tasks. Our results indicate that vocabulary transfer can be effectively used in combination with other compression techniques, yielding a significant reduction in model size and inference time while marginally compromising on performance.

abstract applications arxiv business business applications combination compression cs.ai cs.cl cs.lg domains language language model performance tasks trade trade-off transfer type world