https://jaehun.me/en/categories/deepseek/Recent content in DeepSeek on Jaehun's BlogHugoen-USTue, 08 Sep 2026 03:26:23 +0000https://jaehun.me/en/posts/janus-pro-unifiedmultimodalunderstanding-and-generation-with-data-and-model-scaling/Sun, 06 Jul 2025 00:00:00 +0900https://jaehun.me/en/posts/janus-pro-unifiedmultimodalunderstanding-and-generation-with-data-and-model-scaling/<p><a href="https://arxiv.org/abs/2501.17811v1"target="_blank" class="inline-flex items-center gap-1" >Paper<svg class="h-3 w-3 flex-shrink-0" id="external-link" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M15 3h6v6m-11 5L21 3m-3 10v6a2 2 0 0 1-2 2H5a2 2 0 0 1-2-2V8a2 2 0 0 1 2-2h6"/></svg> </a></p> <h2 id="janus-pro-7b-dual-encoder-multimodal-llm-that-outsmarts-bigger-models"><strong>Janus-Pro 7B: Dual-Encoder Multimodal LLM That Outsmarts Bigger Models</strong><a href="#janus-pro-7b-dual-encoder-multimodal-llm-that-outsmarts-bigger-models" class="heading-anchor" aria-label="Link to this section"><svg class="h-4 w-4" aria-hidden="true" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><g fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2"><path d="M10 13a5 5 0 0 0 7.54.54l3-3a5 5 0 0 0-7.07-7.07l-1.72 1.71"/><path d="M14 11a5 5 0 0 0-7.54-.54l-3 3a5 5 0 0 0 7.07 7.07l1.71-1.71"/></g></svg></a></h2><h2 id="one-line-summary-tldr">One-line summary (TL;DR)<a href="#one-line-summary-tldr" class="heading-anchor" aria-label="Link to this section"><svg class="h-4 w-4" aria-hidden="true" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><g fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2"><path d="M10 13a5 5 0 0 0 7.54.54l3-3a5 5 0 0 0-7.07-7.07l-1.72 1.71"/><path d="M14 11a5 5 0 0 0-7.54-.54l-3 3a5 5 0 0 0 7.07 7.07l1.71-1.71"/></g></svg></a></h2><p>By <em>fully separating the SigLIP understanding encoder from the VQ generation encoder</em> and attaching them to a <strong>7 B-parameter LLM</strong> with a &lsquo;Dual-Encoder + Adapter&rsquo; design, it becomes the first case to <strong>beat a 13 B unified model (TokenFlow-XL) on both understanding and generation at the same time</strong>. – <strong>MMBench 79.2 (+10.3 pt) / GenEval 0.80 (+45 %)</strong></p>