https://duoan.github.io/tags/bf16/ Recent content in BF16 on Duo's Tech Blog https://duoan.github.io/images/papermod-cover.png https://duoan.github.io/images/papermod-cover.png Hugo -- 0.153.1 en-us Sun, 24 May 2026 00:00:00 +0000 https://duoan.github.io/posts/optimizing-vlm-training-on-one-gpu/ Sun, 24 May 2026 00:00:00 +0000 https://duoan.github.io/posts/optimizing-vlm-training-on-one-gpu/ <h1 id="optimizing-vlm-training-on-one-gpu-a-five-layer-recipe">Optimizing VLM Training on One GPU: A Five-Layer Recipe</h1> <p>How I got SiQ-VL from <code>14,713</code> to <code>100,923</code> real tokens per second on a single Blackwell GPU, and the four places that surprised me along the way.</p> <h2 id="tldr">TL;DR</h2> <p>I trained a small vision-language model (SigLIP-2 vision tower + Qwen2.5 LLM, projector-aligned) on one NVIDIA RTX PRO 6000 Blackwell, ran a 48-configuration sweep across two model sizes and both training stages, and ended up with a recipe that compounds five optimization layers.</p>