The KV Cache: Why LLM Serving Is a Memory Problem
Every LLM serving stack lives or dies by one data structure most developers never see: the KV cache. It is
Continue readingThe KV Cache: Why LLM Serving Is a Memory Problem
Tag: LLM
Inside Mixture-of-Experts: Routing, Load Balancing, and Why 671B Parameters Cost 37B
Every modern large language model you use is almost certainly a Mixture-of-Experts model. Mixtral, DeepSeek-V3, GPT-OSS, the latest Kimi and
Continue readingInside Mixture-of-Experts: Routing, Load Balancing, and Why 671B Parameters Cost 37B
Thinking Inertia: Why LLMs Don’t Stop Reasoning When You Tell Them To
Most teams treat “thinking mode” as a switch. Turn it on and the model reasons through the problem step by
Continue readingThinking Inertia: Why LLMs Don’t Stop Reasoning When You Tell Them To
RAG Chunking Is Not Settled: What Token-Level Evaluations Actually Show
Every RAG pipeline starts with the same piece of code: split documents into chunks, embed them, store them. It looks
Continue readingRAG Chunking Is Not Settled: What Token-Level Evaluations Actually Show
The Definitive Guide to Local LLM Inference on Mac: Why Your Machine Is Slow and the Six Fixes That Work
Every Mac owner who has run a local LLM knows the feeling: the model downloads fine, the first answer looks
Continuous Batching: The Scheduling Trick Behind Every Fast LLM Server
Every team that serves language models eventually hits the same wall: the GPU has plenty of memory and the model
Continue readingContinuous Batching: The Scheduling Trick Behind Every Fast LLM Server
Temperature, Top-p, Top-k, and Min-p: How the LLM Sampler Actually Works
Every LLM application has two knobs that get blamed for everything: temperature and top-p. When outputs are boring, someone raises
Continue readingTemperature, Top-p, Top-k, and Min-p: How the LLM Sampler Actually Works
4-Bit by Default: Choosing Between GPTQ, AWQ, GGUF, and MXFP4 in 2026
Four-bit weights used to be a hobbyist compromise — the thing you did to squeeze a 70B model onto a
Continue reading4-Bit by Default: Choosing Between GPTQ, AWQ, GGUF, and MXFP4 in 2026
MHA, MQA, GQA, MLA: The KV-Cache Trade-offs Behind Every Modern Transformer
Ask someone to sketch a transformer on a whiteboard and they’ll draw the 2017 version: softmax attention over every pair
Continue readingMHA, MQA, GQA, MLA: The KV-Cache Trade-offs Behind Every Modern Transformer
Stop Vibing Your RAG Pipeline: Recall@k, MRR, and NDCG in an Afternoon
Most teams that ship a retrieval-augmented generation feature never measure it. The demo looks good, a few hand-picked questions return
Continue readingStop Vibing Your RAG Pipeline: Recall@k, MRR, and NDCG in an Afternoon