Manning Transformers in Action

Transformer models power the chatbots, coders, and translators reshaping every industry today. Yet their architecture, math, and tuning often remain an intimidating black box. Stop copy pasting tutorials and start truly understanding what happens under the hood. Transformers in Action walks you through every layer with practical Python and clear analogies. Master small, large, and multimodal models, then optimize them for speed and cost. Build solutions that translate, summarize, and generate with confidence, efficiency, and rigor. Layer-by-layer walkthrough: See how attention, embeddings, and positional encodings produce fluent output. Task adaptation recipes: Fine-tune models for summarization, classification, or translation in minutes. Optimization strategies: Reduce latency, shrink memory, and cut cloud bills without sacrificing accuracy. Reinforcement learning techniques: Refine text generation quality using reward models and policy gradients. Multimodal expansion: Combine text and vision to build next-generation, cross-media applications. Complete code repository: Experiment instantly, tweak hyperparameters, and validate concepts on real datasets. Transformers in Action, by Quantmate CEO and Chief AI Officer Nicole Koenigstein, has clear math walkthroughs, annotated Python, and production-ready patterns that you can trust. The journey starts with encoder-only, decoder-only, and encoder-decoder variants, then moves to small language models for constrained environments. Each chapter couples theory with runnable notebooks, visual explanations, and performance benchmarks. Finish knowing exactly when to deploy a lightweight model, how to tune hyperparameters, and how to monitor costs. You will ship faster, safer, and leaner LLM solutions that impress users and stakeholders. Ideal for software engineers and data scientists comfortable with Python and basic machine learning, eager to unlock transformer power.