2026

• posts Jul 2 Common Probability Distributions: Variance And Standard Deviation
• posts Jun 29 Optimizers: From SGD To AdamW
• posts Jun 23 vLLM Scheduler: How Request Queues Become SchedulerOutput
• posts Jun 23 vLLM ModelRunner: How SchedulerOutput Becomes a GPU Forward
• posts Jun 23 Numeric Types in Neural Networks: FP32, BF16, FP8, INT8, and INT4
• posts Jun 23 Loss Functions: What a Model Is Really Optimizing
• posts Jun 18 LLM Inference Sampling: What Temperature, Top-p, and Top-k Actually Control
• posts Jun 18 Three Routes For Embodied Models: VLA, World Models, And WAM
• posts Jun 18 Activation Functions: The Small Nonlinearity That Shapes a Network
• posts Jun 17 Streaming Design: Why The Application Layer Still Matters
• posts Jun 7 vLLM Request Lifecycle: From OpenAI API to One Forward Pass
• posts Jun 5 Prefill vs Decode: Why One Model Has Two Very Different Bottlenecks
• posts Jun 5 LLM Attention Kernels and GPU Primitives
• posts Jun 5 LLM Quantization and Low-Precision Serving
• posts Jun 5 LLM Inference Lab Reports: Experiments and Benchmarks for Serving Systems
• posts Jun 4 vLLM / SGLang Source Reading: From Request to Forward Pass
• posts Jun 4 LLM Inference Internals: Core Mechanisms for Serving Engines
• posts Jun 1 A Survey of LLM Quantization: From Linear Quantization to Codebooks
• posts May 28 From Absolute Positional Encoding to RoPE: Why Position Can Be a Rotation
• posts May 27 Estimating Compute and Memory Requirements for LLM Training and Inference

• 1
• 2
• 3
• 4