Napačna izbira? Nič za to! Izdelke lahko vrnete do 30 dni
Z darilnim bonom ne morete zgrešiti. Obdarovanec lahko v zameno za darilni bon izbere karkoli iz naše ponudbe.
Do 30 dni za vračilo
The Transformer Principles Series is a three-volume graduate-level treatise that builds a complete mathematical and engineering understanding of modern AI systems, from the foundational attention mechanism to large language models and multimodal architectures.
Volume I - Mathematical Foundations and Transformer Principles begins with the historical evolution from symbolic AI to deep learning, then develops the essential mathematics: linear algebra, probability, optimization, neural network backpropagation, and information theory. These tools are applied through a systematic construction of the Transformer - self-attention, multi-head projections, positional encodings, feed-forward networks, residual connections, and normalization - culminating in the complete encoder-decoder architecture and an exploration of efficient attention variants, mixture-of-experts, and state-space models.
Pozdravljeni! Sem Libroamiko, vaš knjižni svetovalec.
Kako vam lahko pomagam?