{
  "annotation_id": "67c52983-8cb7-45bb-bbcc-34d18b6cecf5",
  "date_created": "2026-02-17T05:53:19.964000Z",
  "date_modified": "2026-02-17T05:53:19.964000Z",
  "file_hash": "b3b7172aeb321f902d64ad5a8a34fa3235581a8a46bd3b0f851fb0a481c40fad",
  "private": false,
  "record": {
    "abstract": "Large language models typically represent Chinese characters as discrete index-based tokens, largely ignoring their visual form. For logographic scripts, visual structure carries semantic and phonetic information, which may aid prediction. We investigate whether low-resolution visual inputs can serve as an alternative for character-level modeling. Instead of token IDs, our decoder receives grayscale images of individual characters, with resolutions as low as $8 \\times 8$ pixels. Remarkably, these inputs achieve 39.2\\% accuracy, comparable to the index-based baseline of 39.1\\%. Such low-resource settings also exhibit a pronounced \\emph{hot-start} effect: by 0.4\\% of total training, accuracy reaches above 12\\%, while index-based models lag at below 6\\%. Overall, our results demonstrate that minimal visual structure can provide a robust and efficient signal for Chinese language modeling, offering an alternative perspective on character representation that complements traditional index-based approaches.",
    "arxiv_id": "2601.09566",
    "authors": [
      "Shuyang Xiang",
      "Hao Guan"
    ],
    "categories": [
      "cs.CV",
      "cs.AI"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "title": "Hot-Start from Pixels: Low-Resolution Visual Tokens for Chinese Language Modeling",
    "url": "https://arxiv.org/abs/2601.09566",
    "version": "v1"
  },
  "schema_id": "dorsal/arxiv",
  "source": {
    "execution_id": "88284f72-f286-4566-be9f-1e052c2d72fd",
    "id": "arXiv Dataset",
    "type": "Model",
    "variant": "snapshot-2026-01-17",
    "version": "0.1.0"
  },
  "user_id": 1000002
}