GPT-3.5, Llama2-7b, SeaLLM-7b-v2, Gemma-7b, OpenThaiGPT tokenizer.This tokenizer with calculate the number of tokens used by each model. (Lower is better)