Daniel Y. Fu
3 papers · Latest:
Machine Learning
Search Your Block Floating Point Scales!
ScaleSearch optimizes Block Floating Point quantization scales by searching for minimal error, significantly improving generative model performance.
2605.12464
Machine LearningParcae: Scaling Laws For Stable Looped Language Models
Parcae is a stable looped language model architecture that resolves instability, improves quality, and provides scaling laws for efficiency.
2604.12946
Machine LearningFlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness
FlashAttention is an IO-aware exact attention algorithm that significantly speeds up Transformer training and enables longer context lengths by optimizing GPU memory access patterns.
2205.14135
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.