Topics

Efficient AI

Algorithms and systems that reduce memory, compute, or latency for large models.

FlashAttention: The Attention Speedup That Came From Reading GPU Memory

FlashAttention keeps attention exact but makes it IO-aware, using tiling to reduce slow GPU memory traffic and make long-sequence Transformers faster and cheaper.

Sequence Modeling · Carnegie Mellon University

Mamba: A Serious Attempt to Challenge Attention on Long Sequences

Mamba makes state space models selective, letting them decide what to remember or forget from the input while scaling linearly with sequence length.