AI RESEARCH
Peek2: Regex-free Byte-level Byte-Pair Encoding Pretokenizer for LLM Inference on Edge Devices
arXiv CS.CL
•
ArXi:2601.05833v2 Announce Type: replace Pretokenization is a crucial, sequential pass in Byte-level BPE tokenizers, yet little work has been done to optimize it for edge-side inference. Our proposed new implementation, Peek2, serves as a drop-in replacement for cl100k-like pretokenizers used in GPT-3, LLaMa-3, and Qwen-2.5. After breaking down and analyzing the logic of the original cl100k pretokenizer, we