Author: Carlo Young
License: Apache-2.0
Source of truth: https://github.com/carloyoung01/Delta-Neural-Operator-DNO-
DNO is an experimental trainable neural architecture investigating evolving effective states, autonomous computational carriers, and history-dependent interactions through shared learnable dynamics.
This Hugging Face repository mirrors a curated subset of the GitHub source, including core code, synthetic training and verification scripts, research documentation, and four synthetic-task checkpoints. It is not a pretrained general-purpose language model.
Architecture research, defined synthetic-task reproduction, and independent experimental extensions.
The documented GitHub research materials report passing automated tests and successful checkpoint evaluations in their specified synthetic settings. See docs/EXPERIMENTS.md and docs/VERIFICATION_REPORT.md for details, controls, and scope. These results do not establish general-purpose language capabilities, broad model superiority, or energy savings.
Refer to the complete GitHub repository for the full source, test suite, dependencies, and provenance.
Checkpoint files use PyTorch serialization; load only artifacts from trusted sources and review loading practices before use.
evidence/history_17.pt 122.6 kBevidence/history_31.pt 122.6 kBevidence/pair_13.pt 121.5 kBevidence/pair_29.pt 121.5 kB