{
  "abstract": "Adaptive Signed Transformations for Budgeted Transformer Memory proposes a Transformer memory architecture that combines learned routing over multiple mathematical transformations, sign-preserving CReLU representations, and learned exponential retention within a bounded working-memory system. The proposed pipeline is\n\nWx+b→adaptive transformation→CReLU→α-based retention→BWM.Wx+b \\rightarrow \\text{adaptive transformation} \\rightarrow \\text{CReLU} \\rightarrow \\alpha\\text{-based retention} \\rightarrow \\text{BWM}.\n\nThe central hypothesis is that preserving both positive and negative activation evidence before learning a decay coefficient may provide a richer signal for deciding what information should be retained, compressed, or forgotten under a fixed memory budget.\n\nThe work defines parameter- and compute-matched ablations for evaluating routing behavior, signed delayed recall, retention versus delay, compression ratio, OOD performance, and memory-efficiency tradeoffs. It deliberately separates the proposed architecture from existing experimental evidence: the complete adaptive/CReLU/learned-retention pipeline has not yet been experimentally validated, while previously executed BWM experiments are reported only as supporting background.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": null,
  "author_total": 1,
  "authors": [
    {
      "name": "Chaman Kanth"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/cy80nt2yvq3x1.bib",
  "cited_by": 0,
  "doi": "10.5281/zenodo.23621092",
  "fulltext_url": null,
  "id": "cy80nt2yvq3x1",
  "issns": [],
  "landing_url": "https://zenodo.org/records/23621092",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [],
  "openalex": null,
  "pdf_url": "https://zenodo.org/api/records/23621092/files/Adaptive_Signed_Transformations_for_Budgeted_Transformer_Memory_v0.1.0.pdf/content",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-24",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/cy80nt2yvq3x1.ris",
  "sources": [
    {
      "first_seen": "2026-11-07T12:20:39Z",
      "native_id": "23621092",
      "source": "zenodo",
      "url": "https://zenodo.org/records/23621092"
    }
  ],
  "subjects": [],
  "title": "Adaptive Signed Transformations for Budgeted Transformer Memory: Learned Activation Routing, CReLU Sign Preservation, and Exponential Retention",
  "title_text": "Adaptive Signed Transformations for Budgeted Transformer Memory: Learned Activation Routing, CReLU Sign Preservation, and Exponential Retention",
  "type": "preprint",
  "updated_at": "2026-11-07T12:20:39Z",
  "url": "https://research.draconitic.com/paper/cy80nt2yvq3x1",
  "venue": "Zenodo",
  "versions": []
}