On the Emergence of Induction Heads for In-Context Learning
By: Tiberiu Musat , Tiago Pimentel , Lorenzo Noci and more
Potential Business Impact:
Helps computers learn new things from examples.
Transformers have become the dominant architecture for natural language processing. Part of their success is owed to a remarkable capability known as in-context learning (ICL): they can acquire and apply novel associations solely from their input context, without any updates to their weights. In this work, we study the emergence of induction heads, a previously identified mechanism in two-layer transformers that is particularly important for in-context learning. We uncover a relatively simple and interpretable structure of the weight matrices implementing the induction head. We theoretically explain the origin of this structure using a minimal ICL task formulation and a modified transformer architecture. We give a formal proof that the training dynamics remain constrained to a 19-dimensional subspace of the parameter space. Empirically, we validate this constraint while observing that only 3 dimensions account for the emergence of an induction head. By further studying the training dynamics inside this 3-dimensional subspace, we find that the time until the emergence of an induction head follows a tight asymptotic bound that is quadratic in the input context length.
Similar Papers
Predicting the Formation of Induction Heads
Computation and Language
Teaches AI to learn from examples faster.
What One Cannot, Two Can: Two-Layer Transformers Provably Represent Induction Heads on Any-Order Markov Chains
Machine Learning (CS)
Makes computers learn new things with fewer layers.
Selective Induction Heads: How Transformers Select Causal Structures In Context
Machine Learning (CS)
Helps computers choose the right past facts.