Llm-Inference on socratic notes

Llm-Inference on socratic noteshttps://socraticagent.dev/tags/llm-inference/Recent content in Llm-Inference on socratic notesHugoenMon, 07 Sep 2026 00:00:00 +0000Why Your LLM Is Slow for Three Completely Different Reasonshttps://socraticagent.dev/blog/why-your-llm-is-slow-for-three-completely-different-reasons/Mon, 07 Sep 2026 00:00:00 +0000https://socraticagent.dev/blog/why-your-llm-is-slow-for-three-completely-different-reasons/Time-to-first-token and tokens-per-second aren't the same problem wearing different names - they're governed by opposite physical bottlenecks, plus a third one nobody names until requests start queueing for no obvious reason.