SONAR-LLM: Autoregressive Transformer that Thinks in Sentence Embeddings and Speaks in Tokens

Documenti analoghi

• LLM-Microscope: Uncovering the Hidden Role of Punctuation in Context Memory of Transformers
  di: Razzhigaev, Anton, et al.
  Pubblicazione: (2025)
• Universal Adversarial Attack on Aligned Multimodal LLMs
  di: Rahmatullaev, Temurbek, et al.
  Pubblicazione: (2025)
• Logit-KL Flow Matching: Non-Autoregressive Text Generation via Sampling-Hybrid Inference
  di: Sevriugov, Egor, et al.
  Pubblicazione: (2024)
• Addressing Hallucinations in Language Models with Knowledge Graph Embeddings as an Additional Modality
  di: Chekalina, Viktoriia, et al.
  Pubblicazione: (2024)
• Sentence-Anchored Gist Compression for Long-Context LLMs
  di: Tarasov, Dmitrii, et al.
  Pubblicazione: (2025)