[Pytorch] Efficient tokenization for recommendations and generative retrieval. Inspired by STATIC decoding from "Vectorizing the Trie"
-
Updated
Aug 24, 2026 - Python
[Pytorch] Efficient tokenization for recommendations and generative retrieval. Inspired by STATIC decoding from "Vectorizing the Trie"
Accept or Repair: Preserving Passage-Supported Answers in Contrastive Decoding
Reference implementation for "NRR-Phi: Text-to-State Mapping for Ambiguity Preservation in LLM Inference" (arXiv:2601.19933)
Reference implementation for "NRR-Core: Non-Resolution Reasoning as a Computational Framework for Contextual Identity and Ambiguity Preservation" (arXiv:2512.13478)
To associate your repository with the llm-decoding topic, visit your repo's landing page and select "manage topics."