Search Results for author:
Found 3 papers, 0 papers with code
The Developmental Landscape of In-Context Learning
We show that in-context learning emerges in transformers in discrete developmental stages, when they are trained on either language modeling or linear regression tasks.
Proximity to Losslessly Compressible Parameters
In the setting of single-hidden-layer hyperbolic tangent networks, we define the rank of a parameter as the minimum number of hidden units required to implement the same function.
Invariance in Policy Optimisation and Partial Identifiability in Reward Learning
It is often very challenging to manually design reward functions for complex, real-world tasks.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
