Distillation
A technique for training a smaller, faster model to mimic the behavior of a larger, more capable one — trading some performance for dramatically lower cost and latency.
Loading...
Related terms
Read more
- Fine-Tuning a Small Model for One Specific JobOne task, 800 examples, a single 4090. A walkthrough of fine-tuning Qwen3 8B for structured extraction. What worked, what didn't, what it cost.
- Self-Hosting Your AI Stack: The 2026 SetupTailscale, Hetzner, Postgres, n8n, OpenClaw, and Ollama. The exact stack I run, what each piece does, and what it costs.
Last updated 2026-05-12