Isbn: 9798189081163 - fine-tuning open models without regret: practical lora, qlora, and preference tuning for llama, qwen, and mistral models (applied llm engineering series, band 7) (2 Ergebnisse)

PAP. Zustand: New. New Book. Shipped from UK. Established seller since 2000.

Taschenbuch. Zustand: Neu. Neuware - Stop Burning GPU Hours on Fine-Tuning You Don't NeedMost AI teams fine-tune their open-weight Large Language Models (LLMs) when they shouldn't-and they regret it by month three. Fine-Tuning Open Models Without Regret is your opinionated, production-tested roadmap to making the right architectural decisions and executing them with surgical precision.Written for ML engineers, AI architects, and data scientists shipping systems in 2026, this book cuts through the hype to deliver hard numbers, real-world recipes, and honest cost-quality accounting. You will master the transition from foundation models to specialized, high-performing domain experts.What You Will Master Inside: - The Five-Question Framework: Filter out unnecessary fine-tuning projects before spending a single dollar on GPUs.- State-of-the-Art Architecture: Build and deploy with Llama 4, Qwen 3, Mistral, and DeepSeek families.- Data Engineering: High-leverage curation, deduplication, and synthetic preference data generation.- Parameter-Efficient Training: Production-grade LoRA and QLoRA recipes with stable hyperparameters.- Alignment & Preference Tuning: Implement DPO, IPO, KTO, and ORPO to eliminate reward hacking.- Deployment & Operations: Run high-throughput setups using vLLM, multi-LoRA serving, and quantization.Every recipe in this guide has been run in real-world production environments, and every cost estimate has been cross-checked against actual enterprise cloud bills. Spanning twelve practical chapters, this book is designed to be an indispensable reference on your desk.…