{"repo":"Adit-Jain-srm/NightmareNet","free":true,"listed":false,"github":"https://github.com/Adit-Jain-srm/NightmareNet","clone":"git clone https://github.com/Adit-Jain-srm/NightmareNet.git","description":"Biologically-grounded adversarial training platform: cyclic Wake/Dream/Nightmare/Compress phases that accumulate model robustness without catastrophic forgetting. Dockerized, EU AI Act compliance reporting.","language":"Python","stars":48,"topics":["adversarial-robustness","adversarial-training","deep-learning","fastapi","knowledge-distillation","machine-learning","nextjs","nlp","pytorch","text-classification"],"license":"Apache-2.0","category":"machine-learning","readme_excerpt":"NightmareNet Autonomous Adversarial Robustness Training Platform A cyclic adversarial training platform that continuously strengthens model robustness through the Wake → Dream → Nightmare → Compress learning cycle. --- NightmareNet Zero-Install Research Sandboxes The first platform that actively improves model robustness through biologically-grounded training cycles. Wake. Dream. Nightmare. Compress. Repeat. --- The Problem Production models silently degrade. Adversarial perturbations as small as a single token swap collapse model accuracy from 92% to 23% (Jin et al. 2020, TextFooler ). Conventional adversarial training trades clean accuracy for robustness — and worse, it suffers from \"robustness forgetting\" (AAAI 2025, ICCV 2025), where each new training run erodes previously-acquired defenses. The EU AI Act Article 15 (fully applicable August 2, 2026) now mandates demonstrable robustness for high-risk AI systems, but no existing tool combines adversarial generation, forgetting prevention, compression, and orchestration into a single coherent workflow. [!NOTE] NightmareNet is not a runtime guardrail (Lakera) or evaluation library (TextAttack). It is a training paradigm that produces measurably more robust models, with a hosted platform for orchestration and EU AI Act compliance reporting. --- The Solution — A 4-Phase Sleep Cycle NightmareNet implements a biologically-grounded cyclic training loop inspired by sleep-mediated memory consolidation. Each cycle decomposes robustne","default_branch":null,"files":null,"tree":[],"storefront":"/r/Adit-Jain-srm","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/Adit-Jain-srm/NightmareNet/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}