AI engineering glossary · LLM training alignment
What is superficial alignment hypothesis?
Superficial alignment hypothesis: The claim, from the LIMA paper, that almost all knowledge and ability is learned in pretraining, and alignment mainly teaches the style and format of interaction, so a small set of high-quality examples can be enough.
Superficial alignment hypothesis: an example
LIMA fine-tuned a large model on only about a thousand carefully chosen examples and still obtained strong chat behaviour.
Related terms
More LLM training alignment terms
- Supervised fine-tuning
- Sycophancy
- Synthetic data
- Task arithmetic
- Teacher-student learning
- Tensor parallelism
- Test-time compute scaling
- Trust region policy optimisation
Where you learn it
Superficial alignment hypothesis is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.