2025-09-29

Models

• [ ]

Papers

• [2510.02228] xLSTM Scaling Laws: Competitive Performance with Linear Time-Complexity

Code

• [ ]

Articles

• LoRA Without Regret - Thinking Machines Lab
• Inside NVIDIA GPUs: Anatomy of high performance matmul kernels - Aleksa Gordić
• Inside vLLM: Anatomy of a High-Throughput LLM Inference System - Aleksa Gordić
• Writing high-performance matrix multiplication kernels for Blackwell — JAX documentation

Videos

• Lecture 80: How FlashAttention 4 Works - YouTube

Other

• [ ]

Tweets