Creează un cont pentru a te abona la acest blog și a-i urmări articolele pe pagina Lectură.
-
Learn how speculative decoding accelerates LLM inference through predictable tokens, rejection sampling, parallel verification, and GPU roofline analysis.
-
Volume, equators, and near-orthogonality in high dimensions, and why embeddings and random projections work.
-
Writing an NVFP4 GEMV kernel for Blackwell B200.
-
A practical guide to CuTe's layout algebra — shapes, strides, composition, complement, division and product — with a tiled GEMM kernel walkthrough.
-
Deep dive into HNSW, an approximate nearest neighbor search algorithm.
-
A deep dive into building a Linux filesystem watcher.
-
An intuition based explanation around why random choice of 2 load balancing works so well.
-
An account of experience using Clojure for Advent of Code 2021.