Writing
- 2026-04-18Hello
What this site is, why it exists, what will live here.
- 2026-05-07Dense Bench, Part 0: Why I built a benchmark I didn't need.
Why existing MLX benchmark data didn't answer the question I had about dense 70B inference at long context on M3 Ultra 512GB — and the three properties I needed in a harness before I trusted any of it.
- 2026-05-03Dense Bench, Part 1: Prefill and decode curves from 4K to 128K context.
Real prefill, TTFT, decode, and memory curves for Llama-3.1-70B-Instruct-4bit and Qwen-2.5-72B-Instruct-4bit at 4K–128K context on Mac Studio M3 Ultra 512GB. Methodology, raw tables, and a reproducible harness.
- 2026-05-07Dense Bench, Part 2: The Llama 128K cliff.
At 128K context, Llama-3.1-70B-Instruct emits end-of-turn as its first generated token. Qwen-2.5-72B-Instruct doesn't. The 42 tokens past the nominal context limit, the harness flag that caught it, and what to take away if you're shipping long-context Llama at the edge.
- 2026-05-12Dense Bench, Part 3: Compressed KV at the retrieval boundary.
120 runs sweeping int8 / int4 / FP16 KV cache across 5 needle depths and 5–7 context lengths for dense Llama-3.1-70B and Qwen-2.5-72B on M3 Ultra 512GB. Bit width turned out not to be the variable that mattered. A cache-key bug in the 192K probe, caught after publish, is documented in §6.
- 2026-06-24Dense Bench, Part 4: Memory doesn't predict it — agentic reliability vs quantization.
640 graded multi-turn tool-calling episodes across Llama-3.1-70B and Qwen-2.5-72B at 4-bit and 8-bit on M3 Ultra 512GB. Peak memory does not predict task completion, and the quantization effect reverses sign between the two model families. Includes a Plain-English reading mode.
- 2026-08-04Fine-tuning a dictation cleanup model: designing the corpus, and the category 25% self-correction data still missed.
Why dictation cleanup needed a fine-tune rather than a longer prompt, how I designed a synthetic corpus to cover the space of things people dictate, and the failure category that survived a corpus where a quarter of all rows were already self-corrections. Includes a Plain-English reading mode.
- 2026-10-29The event tap that died after deep sleep
macOS stopped delivering events to Pomvox's hotkey tap after deep sleep while every API said it was fine. The first fix passed the easy repro and missed the real bug.