A Practical Guide to Training a Small Language Model: Tokenizers, Training, and Real-World PitfallsSimilarity score = 0.58 More