MultiTok: Variable-Length Tokenization for Efficient LLMs Adapted from LZW Compression
This addresses the problem of resource-intensive training for LLMs, offering a practical efficiency improvement, though it is incremental as it builds on existing tokenization and compression techniques.
The paper tackles the high resource demands of training large language models by introducing MultiTok, a variable-length tokenization method based on LZW compression that compresses repetitive phrases. It achieves comparable performance to BERT and GPT-2 standards while enabling about 2.5x faster training with over 30% less data.
Large language models have drastically changed the prospects of AI by introducing technologies for more complex natural language processing. However, current methodologies to train such LLMs require extensive resources including but not limited to large amounts of data, expensive machinery, and lengthy training. To solve this problem, this paper proposes a new tokenization method inspired by universal Lempel-Ziv-Welch data compression that compresses repetitive phrases into multi-word tokens. With MultiTok as a new tokenizing tool, we show that language models are able to be trained notably more efficiently while offering a similar accuracy on more succinct and compressed training data. In fact, our results demonstrate that MultiTok achieves a comparable performance to the BERT and GPT-2 standards as both a stand-alone tokenizer and an add-on to existing tokenizers while also providing close to 2.5x faster training with more than 30% less training data.