Browse by topic
Topics
Follow a thread through the archive. The same questions keep meeting in unexpected places.
ai9 essays
- Transformer Attention, Backwards
- Pipeline Parallelism: Surgery on Models Too Big for the Operating Table
- Data Parallelism: Scaling LLM Training Through Parallel Processing
- Move Fast or Die Slow
- The Operating Room Setup
- A quick incision: ten minutes to RAG
- Performing Kernel Surgery: Profiling CUDA Kernels with NVIDIA Nsight Compute
- A Machine Learning Surgeon’s Toolkit: Advanced Matrix Multiplication in CUDA
- An Introduction to Sparsity for Efficient Neural Network Inference
attention1 essays
compilers1 essays
cuda6 essays
- A fused Lion optimizer kernel with Hugging Face kernels
- The Operating Room Setup
- Performing Kernel Surgery: Profiling CUDA Kernels with NVIDIA Nsight Compute
- A Machine Learning Surgeon’s Toolkit: Advanced Matrix Multiplication in CUDA
- Cerebral Cortex and Hippocampus: Understanding the Computational and Memory Design of GPUs
- Hello CUDA: A Surgical Dissection
efficiency11 essays
- GGUF quantization, bit by bit
- TurboQuant: What 3-Bit KV Caches Actually Mean for Your Inference Stack
- Pipeline Parallelism: Surgery on Models Too Big for the Operating Table
- Data Parallelism: Scaling LLM Training Through Parallel Processing
- The Operating Room Setup
- A quick incision: ten minutes to RAG
- Performing Kernel Surgery: Profiling CUDA Kernels with NVIDIA Nsight Compute
- A Machine Learning Surgeon’s Toolkit: Advanced Matrix Multiplication in CUDA
- Cerebral Cortex and Hippocampus: Understanding the Computational and Memory Design of GPUs
- Hello CUDA: A Surgical Dissection
- An Introduction to Sparsity for Efficient Neural Network Inference
gguf1 essays
gpu1 essays
grpo1 essays
hardware1 essays
inference7 essays
- What Kimi K3 asks of the hardware
- GGUF quantization, bit by bit
- TurboQuant: What 3-Bit KV Caches Actually Mean for Your Inference Stack
- Transformer Attention, Backwards
- Pipeline Parallelism: Surgery on Models Too Big for the Operating Table
- A quick incision: ten minutes to RAG
- An Introduction to Sparsity for Efficient Neural Network Inference
kernels1 essays
metal1 essays
ml20 essays
- What Kimi K3 asks of the hardware
- GGUF quantization, bit by bit
- GRPO is mostly a systems problem
- A fused Lion optimizer kernel with Hugging Face kernels
- TurboQuant: What 3-Bit KV Caches Actually Mean for Your Inference Stack
- Transformer Attention, Backwards
- Pipeline Parallelism: Surgery on Models Too Big for the Operating Table
- Data Parallelism: Scaling LLM Training Through Parallel Processing
- The Memory Anatomy of Large Language Models: A Surgeon's Guide
- From Sequential to Parallel: Your Journey into GPU Programming with Triton
- The Transformer's Anatomy: A Deep Dive into the Architecture that Revolutionized Machine Learning
- The Machine Learning Surgeon's Guide to Quantization: Precision Cuts for Smarter Models
- The Operating Room Setup
- Dissecting torch.compile: Surgical Precision in PyTorch Optimization
- A quick incision: ten minutes to RAG
- Performing Kernel Surgery: Profiling CUDA Kernels with NVIDIA Nsight Compute
- A Machine Learning Surgeon’s Toolkit: Advanced Matrix Multiplication in CUDA
- Cerebral Cortex and Hippocampus: Understanding the Computational and Memory Design of GPUs
- Hello CUDA: A Surgical Dissection
- An Introduction to Sparsity for Efficient Neural Network Inference