{
  "apiVersion": "v1",
  "methodology": "https://sourcescore.org/methodology/",
  "canonical": "https://sourcescore.org/claims/f5b422e3255fd7c0/",
  "claim": {
    "vertical": "ai-ml",
    "subject": "BART",
    "predicate": "introduced_in",
    "object": "Lewis et al. 2019 — denoising sequence-to-sequence pretraining",
    "confidence": 1,
    "sources": [
      {
        "url": "https://arxiv.org/abs/1910.13461",
        "title": "BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension",
        "publisher": "arXiv (Lewis, Liu, Goyal, Ghazvininejad, Mohamed, Levy, Stoyanov, Zettlemoyer / Facebook AI)",
        "publishedDate": "2019-10-29",
        "accessedDate": "2026-05-16",
        "type": "preprint",
        "excerpt": "We present BART, a denoising autoencoder for pretraining sequence-to-sequence models. BART is trained by (1) corrupting text with an arbitrary noising function, and (2) learning a model to reconstruct the original text. It uses a standard Tranformer-based neural machine translation architecture which, despite its simplicity, can be seen as generalizing BERT (due to the bidirectional encoder), GPT (with the left-to-right decoder), and many other more recent pretraining schemes."
      },
      {
        "url": "https://huggingface.co/docs/transformers/model_doc/bart",
        "title": "BART — Hugging Face Transformers documentation",
        "publisher": "Hugging Face",
        "publishedDate": "2019-10-29",
        "accessedDate": "2026-05-16",
        "type": "official-blog"
      }
    ],
    "publishedAt": "2026-05-16T00:00:00Z",
    "lastVerified": "2026-05-16",
    "methodologyVersion": "veritas-v0.1",
    "tags": [
      "bart",
      "facebook-ai",
      "denoising",
      "seq2seq",
      "encoder-decoder",
      "foundational",
      "2019",
      "introduced_in"
    ],
    "id": "f5b422e3255fd7c0",
    "statement": "BART introduced in: Lewis et al. 2019 — denoising sequence-to-sequence pretraining."
  },
  "signature": {
    "algorithm": "HMAC-SHA256",
    "signedBy": "did:web:sourcescore.org",
    "signedAt": "2026-05-16T00:00:00.000Z",
    "signature": "f9be37b52136a1f1f539e83bb567b9ca6825f04c28460a57080a86b5c5fd8f84"
  },
  "citedAs": "BART introduced in: Lewis et al. 2019 — denoising sequence-to-sequence pretraining. — SourceScore Claim f5b422e3255fd7c0 (verified 2026-05-16, signed f9be37b5…). https://sourcescore.org/claims/f5b422e3255fd7c0/"
}