Adam Klein

Implicit Test-Time Training for infinite-context sequence modelling and continual learning. Currently a work in progress.

Train parallel and test serial for scalable latent reasoning. Currently a work in progress.

Compute Where It Counts. A new state-of-the-art method for creating sparse transformers that automatically decide when to use more or less compute.

Enabling both sparsity and low-bit quantization in neural networks using stochastic weights and the local reparameterization trick.

Efficiently pretraining energy-based language models that operate at the token level.

Improving diffusion model sample quality by searching for better random noise using metaheuristics (early work on test-time scaling).