It doesn't generate new tokens.
Train an LLM on text that only uses lowercase, and it will never output an uppercase letter.
Train an LLM on text that only uses lowercase, and it will never output an uppercase letter.
its as stupid as that. some try to get around it by indeed only having the 10 different digits and glue them together, but its a hallucination that that works.
an important point in generalization is for example that you teach it something. This is literally important
'ycombinator is a website' is a prompt that is almost impossible of ycombinator is not in your training set
10 01 = 1001?