{
  "slug": "gpt-2-xl-1-5b",
  "order": 0,
  "name": "GPT-2 XL (1.5B)",
  "baseTitle": "GPT-2 XL 1.5B",
  "company": "OpenAI",
  "country": "United States",
  "flag": "🇺🇸",
  "date": "2019-11-05",
  "decoderType": "Dense",
  "attention": "MHA with learned absolute positional embeddings",
  "layerMix": "48 MHA",
  "scale": "1.5B parameters",
  "totalParams": 1500000000,
  "activeParams": 1500000000,
  "context": "1,024",
  "contextTokens": 1024,
  "license": "OpenAI \"Modified MIT\" license",
  "kvCacheText": "300 KiB",
  "kvCacheKiB": 300,
  "kvCacheLevel": "High",
  "summary": "Late-2019 dense baseline included here as a reference point for how much decoder stacks have changed since GPT-2.",
  "keyDetail": "Classic GPT-2 recipe with dropout, GELU, LayerNorm, and full multi-head attention.",
  "image": {
    "full": "https://sebastianraschka.com/llm-architecture-gallery/images/architectures/gpt-2-xl.webp",
    "thumb": "https://sebastianraschka.com/llm-architecture-gallery/images/architectures/thumbnails/gpt-2-xl.webp",
    "alt": "Figure 2: GPT-2 XL 1.5B",
    "width": 882,
    "height": 960
  },
  "articleUrl": "https://magazine.sebastianraschka.com/p/from-gpt-2-to-gpt-oss-analyzing-the#%C2%A72-coming-from-gpt-2",
  "links": [
    {
      "label": "View in article",
      "href": "https://magazine.sebastianraschka.com/p/from-gpt-2-to-gpt-oss-analyzing-the#%C2%A72-coming-from-gpt-2",
      "title": null
    },
    {
      "label": "config.json",
      "href": "https://huggingface.co/openai-community/gpt2-xl/blob/main/config.json",
      "title": "openai-community/gpt2-xl"
    },
    {
      "label": "Tech report",
      "href": "https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf",
      "title": null
    }
  ],
  "concepts": [
    {
      "slug": "mha",
      "label": "MHA",
      "name": "Multi-Head Attention",
      "url": "https://llm-atlas-7or.pages.dev/concepts/mha/"
    }
  ],
  "aai": null,
  "sourceUrl": "https://sebastianraschka.com/llm-architecture-gallery/#card-gpt-2-xl-1-5b",
  "family": "dense",
  "url": "https://llm-atlas-7or.pages.dev/models/gpt-2-xl-1-5b/",
  "markdownUrl": "https://llm-atlas-7or.pages.dev/models/gpt-2-xl-1-5b.md"
}