Add tokenizer

SFT-trained tiny-aya-global for tool calling

Include dataset information

Model save

Training in progress, step 100

initial commit