What “Small” LLMs Are Actually Good For
Practical uses, trade‑offs, and benchmarks that matter for teams without unlimited GPUs.
Evidence‑based notes from building and shipping with machine learning. News I actually read, reviews I test, and essays from the messy edge. By Jen.
Practical uses, trade‑offs, and benchmarks that matter for teams without unlimited GPUs.
From demos to durable services: prompts, tools, and guardrails that don’t crumble.
Latency under load, cost cliffs, and the operational gotchas I wish I knew earlier.
Hardware, models, and a dumb‑simple pipeline your future self won’t hate.
LoRA still wins for most teams. The surprise is where QLoRA breaks (and how to spot it).