alignment — The work of making AI systems actually pursue what their builders and users intend, rather than something subtly or dangerously different.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
fine-tuning — Taking a trained model and training it a bit more on your own examples so it gets better at one specific job.
Why it matters
If fine-tuningTaking a trained model and training it a bit more on your own examples so it gets better at one specific job.Full definition → leaves your model falling into repetition loops, a post-SFT DPO stage built from its own failures gives an objective preference signal with no human labeling, and the result held across five model families.
