{ "title": "Nexuss AI Training Tutorials", "description": "End-to-end guide for AI model training from blank slate to production", "version": "1.0.0", "tutorials": [ { "id": "00", "number": 0, "slug": "introduction-overview", "filename": "00-introduction-overview.md", "title": "Introduction & Overview", "category": "Core Training", "level": "Beginner", "description": "Framework overview, training lifecycle, and hardware requirements" }, { "id": "01", "number": 1, "slug": "blank-slate-models", "filename": "01-blank-slate-models.md", "title": "Blank Slate Models", "category": "Core Training", "level": "Intermediate", "description": "Transformer architecture from scratch and model initialization" }, { "id": "02", "number": 2, "slug": "first-training-run", "filename": "02-first-training-run.md", "title": "First Training Run", "category": "Core Training", "level": "Beginner", "description": "Complete pipeline setup, monitoring, and debugging" }, { "id": "03", "number": 3, "slug": "full-finetuning", "filename": "03-full-finetuning.md", "title": "Full Fine-Tuning", "category": "Core Training", "level": "Intermediate", "description": "DeepSpeed ZeRO, multi-GPU training, and memory optimization" }, { "id": "04", "number": 4, "slug": "advanced-finetuning", "filename": "04-advanced-finetuning.md", "title": "Advanced Fine-Tuning", "category": "Core Training", "level": "Advanced", "description": "Multi-task learning, DPO/SimPO, and instruction tuning" }, { "id": "05", "number": 5, "slug": "peft-lora", "filename": "05-peft-lora.md", "title": "PEFT & LoRA", "category": "Core Training", "level": "Intermediate", "description": "Parameter-efficient methods, QLoRA, and adapter techniques" }, { "id": "06", "number": 6, "slug": "rlhf", "filename": "06-rlhf.md", "title": "RLHF", "category": "Core Training", "level": "Advanced", "description": "Reward modeling, PPO, and preference optimization" }, { "id": "07", "number": 7, "slug": "validation-testing", "filename": "07-validation-testing.md", "title": "Validation & Testing", "category": "Production & Lifecycle", "level": "Intermediate", "description": "Statistical validation, bias detection, and adversarial testing" }, { "id": "08", "number": 8, "slug": "continual-learning-lifecycle", "filename": "08-continual-learning-lifecycle.md", "title": "Continual Learning & Lifecycle", "category": "Production & Lifecycle", "level": "Advanced", "description": "Forgetting prevention, deployment strategies, and drift detection" }, { "id": "09", "number": 9, "slug": "release-management", "filename": "09-release-management.md", "title": "Release Management", "category": "Production & Lifecycle", "level": "Advanced", "description": "Version control, freezing, rollback, and deprecation" }, { "id": "10", "number": 10, "slug": "distributed-training-scale", "filename": "10-distributed-training-scale.md", "title": "Distributed Training at Scale", "category": "Production & Lifecycle", "level": "Advanced", "description": "TP/PP/ZeRO, hybrid parallelism strategies" }, { "id": "11", "number": 11, "slug": "inference-optimization-deployment", "filename": "11-inference-optimization-deployment.md", "title": "Inference Optimization & Deployment", "category": "Production & Lifecycle", "level": "Advanced", "description": "Quantization, vLLM, TGI, and speculative decoding" }, { "id": "12", "number": 12, "slug": "mlops-automation-governance", "filename": "12-mlops-automation-governance.md", "title": "MLOps, Automation & Governance", "category": "Production & Lifecycle", "level": "Intermediate", "description": "CI/CD, registries, compliance, and monitoring" }, { "id": "13", "number": 13, "slug": "troubleshooting-performance-debugging", "filename": "13-troubleshooting-performance-debugging.md", "title": "Troubleshooting & Performance Debugging", "category": "Production & Lifecycle", "level": "Intermediate", "description": "NaNs, OOMs, convergence issues, and profiling" } ], "categories": [ "Core Training", "Production & Lifecycle" ], "levels": [ "Beginner", "Intermediate", "Advanced" ] }