An open-source terminal coding agent runtime and CLI for real project work.
7-layer RAG framework that eliminates semantic drift + context poisoning. Faithfulness 0.94 vs 0.71 Naive RAG. +13.6 EM on multi-hop QA.
Simulation code for Ambiguity-Bearing Outputs (ABO) across Interconnected Systems Environment (ISE). Validates Inter-Systems Coherence & Integrity Layer (ISCIL) containment architecture. Paper: Ayada (2026).
PromptGuard is a pragmatic, opinionated framework for establishing continuous integration for LLM behavior. It operates on a simple, verifiable principle: run the same prompts across multiple model configurations, compare outputs against defined expectations, and flag semantic regressions.
ModelPulse helps maintain model reliability and performance by providing early warning signals for these issues, allowing teams to address them before they impact users significantly.
Research proposal and pilot study on detecting temporal behavioral drift in Model Context Protocol (MCP) tool descriptions.
Code and data manifest for semantic drift in Portuguese financial disclosures
Preprint + pseudocode for the PUTMAN Model (relational meaning graphs, activation, constrained reconstruction, rigidity).
Predicting Semantic Drift from Polysemy Density A cognitive modeling project for APLN-552 (Spring 2025) exploring the relationship between a word's polysemy (number of noun senses) and its tendency to change meaning over time. Combines WordNet sense counts with diachronic word embeddings (HistWords) to compute polysemy density and semantic drift.
Public app demo showing LLOYD working with GPT-2.
Lexical contour protection layer of RET preserving canonical terminology, anti-mapping boundaries, and semantic inversion resistance.
Semantic contracts for AI-modified TypeScript. Auto-extracts project-specific intent and catches what tests and linters miss.
Statistical monitoring for legal document revision — detect semantic drift in contracts and policies
AI code governance engine that detects scope drift in AI‑generated diffs.
A meta-governance architecture for AI engines, human-origin traces, value circulation, dynamic equilibrium, validation, and recovery.
This project quantifies the geometric and structural "break-point" of LLM representations under data corruption. Using Intrinsic Dimension and CKA mapping, it identifies the exact transformer layers where noise destroys semantic alignment and triggers Uncertainty Calibration collapse.
A protocol for relaying conceptual traces across dialogues, memory layers, and derivative structures.
Quote-to-Cash LLMOps Control Plane — token cost attribution, quality scoring, latency SLOs, prompt versioning, and semantic drift detection for production LLM revenue workflows. Local-first with Ollama. No API key required.
A structural tuning layer for AI governance, designed to align AI actions, recovery decisions, self-improvement loops, and human review with constitutional principles.
Add a description, image, and links to the semantic-drift topic page so that developers can more easily learn about it.
To associate your repository with the semantic-drift topic, visit your repo's landing page and select "manage topics."