Does Visual Rendering Bypass Tokenization? Investigating Script-Tokenizer Misalignment in Pixel-Based Language Models

Documenti analoghi

• NusaAksara: A Multimodal and Multilingual Benchmark for Preserving Indonesian Indigenous Scripts
  di: Adilazuarda, Muhammad Farid, et al.
  Pubblicazione: (2025)
• Beyond Transfer Accuracy: Faithful Circuits for Controlled Low-Resource Adaptation
  di: Nur'aini, Khumaisa, et al.
  Pubblicazione: (2026)
• IndoToxic2024: A Demographically-Enriched Dataset of Hate Speech and Toxicity Types for Indonesian Language
  di: Susanto, Lucky, et al.
  Pubblicazione: (2024)
• A Multi-Labeled Dataset for Indonesian Discourse: Examining Toxicity, Polarization, and Demographics Information
  di: Susanto, Lucky, et al.
  Pubblicazione: (2025)
• Environmental carrying capacity and environmental capacity in the issuance of recommendation and borrow-use permit of forest area
  di: Wiwin, Nur'aini
  Pubblicazione: (2019)