{
  "id": "base-model",
  "code": "PTL-0104",
  "term": "Base Model",
  "aliases": [
    "pretrained model",
    "base LLM",
    "completion model"
  ],
  "category": "foundations",
  "definition": "A base model is a language model after pretraining but before instruction tuning or preference tuning, so it continues text in the style of its training data rather than following instructions.",
  "description": "Asked a question, a base model may answer it, continue with more questions, or produce something unrelated, because it only predicts plausible continuations. Prompting one effectively means framing the task as a document to complete, for example with few-shot examples or the start of the desired output. Chat assistants are derived from base models through instruction tuning and RLHF, which is why they are often called instruct or chat models.",
  "example": "A completion-style prompt for a base model: \"Q: What is the capital of France?\\nA: Paris\\nQ: What is the capital of Japan?\\nA:\"",
  "broader": [
    "large-language-model"
  ],
  "narrower": [],
  "related": [
    "instruction-tuning",
    "rlhf",
    "few-shot-prompting",
    "chat-template"
  ],
  "introduced": null,
  "added": "2026-10-10",
  "sources": [
    {
      "title": "Training language models to follow instructions with human feedback",
      "authors": "Ouyang et al.",
      "year": 2022,
      "url": "https://arxiv.org/abs/2203.02155"
    },
    {
      "title": "Language Models are Few-Shot Learners",
      "authors": "Brown et al.",
      "year": 2020,
      "url": "https://arxiv.org/abs/2005.14165"
    }
  ],
  "url": "https://protologue.com/t/base-model/",
  "citation": "Protologue. (2026). Base Model. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0104). https://protologue.com/t/base-model/"
}