Large language models can readily produce diverse and unusual outputs, yet increased variation does not by itself explain when a system becomes more capable of creative thought. This paper develops a preliminary hypothesis that prompt design may alter the effective dynamical regime of inference and, under some conditio...
The VisDrone-derived subset is used only as an industrial low-altitude proxy stress test: its high precision coexists with very low recall, so the result exposes domain-shift limitations, rather than demonstrating deployment-level industrial performance.
Rong-Zuo Guo, Yu-Hang He· Journal of Imaging· 0 citations
✨ Highlights Review skill updates as a real diff. Update previews render through a shared source diff viewer — file headers with change counts, line-number gutter, dashed red deletion and solid green addition rails, optional syntax highlighting, and readable fallbacks for malformed or oversized patches. (#2632) Side c...
Ewen, wen2zhou, Roxi et al.· Zenodo (CERN European Organi...· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Classical software memory leaks are allocations that outlive their intended lifetime. Tool-using large language model (LLM) agents exhibit an analogous failure that current “forgetting” mechanisms do not measure. A user or policy issues a forget request for a memory m*; the underlying store correctly drops the record,...
Monish Kanungo, Seemant Kaushal· Zenodo (CERN European Organi...· 0 citations
Composed Image Retrieval (CIR) asks a system to retrieve a target image from a reference image and a natural-language modification. Although large vision-language models provide strong global image-text representations, CIR remains difficult because training data often contains verbose or weakly pragmatic modification...
Zihang Li· Zurich Open Repository and A...· 0 citations
First stable release of the toolkit that tests the structural hallucination thesis directly: that a language model's confidence tracks consistency with its own training corpus, not correspondence with the world. A small causal LM (Qwen2.5-0.5B by default) is fine-tuned on a corpus that consistently asserts a fabricated...
José Luciano Verçosa Marques, Frederico Jorge Heitmann, Daniel Omar Pérez et al.· Zenodo (CERN European Organi...· 0 citations
This paper reports the first executed computational proof-of-concept and cross-corpus replication of Bacterialized Gematria, a population-based symbol–number system in which an initially fixed A=1 through Z=26 correspondence table is transformed into a historically contingent numerical ecology capable of mutation, hori...
Kenji Siratori· Zenodo (CERN European Organi...· 0 citations
First stable release of the toolkit that tests the structural hallucination thesis directly: that a language model's confidence tracks consistency with its own training corpus, not correspondence with the world. A small causal LM (Qwen2.5-0.5B by default) is fine-tuned on a corpus that consistently asserts a fabricated...
José Luciano Verçosa Marques, Frederico Jorge Heitmann, Daniel Omar Pérez et al.· Zenodo (CERN European Organi...· 0 citations
