# Posts

Field notes from local inference experiments. Results, failed ideas, and enough detail to repeat the run.

- [Why this blog exists](https://andreabor.io/posts/why-this-blog-exists/index.md) — 2026-08-21: The useful part of an experiment usually happens after the screenshot.
- [I tried to run GLM-5.2 on a 64GB Mac](https://andreabor.io/posts/i-tried-to-run-glm-52-on-a-64gb-mac/index.md) — 2026-06-23: Field notes from an experimental ds4 fork, a 244GB GGUF, and the small horror of sparse models that are sparse in compute but not very friendly to filesystems.
- [Re-quantizing a local model, 14× faster](https://andreabor.io/posts/re-quantizing-a-local-model-14-faster/index.md) — 2026-06-10: Where a 2-bit model spends its bits, and why trying answers used to cost eighty minutes