Category: Seminar
-
Introduction to Transformers
An introduction to the transformer architecture and the attention mechanism lying at its heart, following the famous paper “Attention Is All You Need” by Vaswani et. al. Go to article…
-
Custom CUDA and Python
Just-in-time (JIT) compilation and Python bindings for interfacing with NVIDIA CUDA. Go to article…
-
Analog AI
A look at the status of analog deep learning inference and training technologies. Go to article…
-
Narrow Bit-Width Formats for Deep Learning
New number formats with precision less than FP32 allow for faster and more power-efficient deep learning. Go to article…
-
Proxmox VE
Proxmox Virtual Environment (Proxmox VE) is an open-source virtualization platform that integrates multiple open-source technologies. Go to article…
-
Software Pipelining in the NVIDIA Hopper Architecture
C++ Software Pipelining Template for overlapping TMA and GEMM operations on the NVIDIA Hopper architecture. Go to article…
-
JSON Web Token Standard (JWT)
JSON Web Tokens are an open, industry standard RFC 7519 method for representing claims securely between two parties. Go to article…
-
SPEC Benchmarks
Industry-standardized, CPU intensive suites for measuring and comparing compute intensive performance, stressing a system’s processor, memory subsystem and compiler. Go to article…