{
 "slug": "T5",
 "title": "T5",
 "type": "model",
 "short_desc": "Google's 2019 Text-to-Text Transfer Transformer, an open-weight encoder-decoder family that cast every NLP task as text-to-text.",
 "categories": [
  "Google models",
  "Models",
  "Open-weight models",
  "2019 model releases"
 ],
 "infobox": {
  "Developer": "Google Research (Brain team)",
  "Announced": "October 2019",
  "Architecture": "Encoder-decoder Transformer trained with a span-corruption denoising objective",
  "Parameters": "60 million to 11 billion across five released sizes",
  "Context window": "512 tokens (pre-training sequence length; relative position embeddings)",
  "Post-training": "Supervised fine-tuning on downstream tasks; base release had no instruction tuning",
  "License / access": "Apache 2.0; open weights and code",
  "Paper/report": "Raffel et al., \"Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer\" (arXiv:1910.10683)"
 },
 "infobox_links": {
  "Architecture": [
   "Transformer_(architecture)"
  ],
  "Post-training": [
   "Instruction_tuning"
  ],
  "Paper/report": [
   "Transformer_(architecture)"
  ]
 },
 "aliases": [
  "Text-to-Text Transfer Transformer"
 ],
 "family": "",
 "org": "Google",
 "registry_status": "stub",
 "words": 453,
 "references": 4,
 "outbound": [
  "BERT",
  "Common_Crawl",
  "Flan-T5",
  "GPT-2",
  "GPT-3",
  "Hugging_Face",
  "Imagen",
  "Instruction_tuning",
  "Large_language_model",
  "PaLM",
  "RoBERTa",
  "Transformer_(architecture)",
  "XLNet"
 ],
 "inbound": [
  "Adam_(optimizer)",
  "Aya",
  "Common_Crawl",
  "DALL-E_3",
  "FLUX",
  "Flan-T5",
  "HunyuanVideo",
  "Imagen",
  "Large_language_model",
  "Masked_language_modeling",
  "Meena",
  "Multi-head_attention",
  "Positional_encoding",
  "RMSNorm",
  "Self-attention",
  "Seq2seq",
  "Tokenization",
  "Transformer_(architecture)",
  "XLNet"
 ],
 "url": "wiki/T5.html",
 "built": "2026-07-24"
}