The Complete Guide to Transformers Just Dropped! 👉 New Course

DeepLearning.AI··2 min read
AI/MLTechnologyEngineering
Share𝕏in

AI Summary

DeepLearning.AI has launched 'Transformers in Practice', a new course taught by Sharon Zhou, VP of Engineering & AI at AMD. The course covers the internal mechanics of transformer-based LLMs, including attention mechanisms, hallucinations, RAG, and GPU inference optimizations like quantization, KV caching, and speculative decoding. It uses interactive visualizations to build practical intuition from both model and systems perspectives.

Key Facts

✓DeepLearning.AI launched Transformers in Practice, a new course taught by Sharon Zhou (VP of Engineering & AI at AMD) covering LLM internals from both model and systems perspectives.
✓The course teaches why hallucinations occur and how techniques like RAG and constrained generation mitigate them, alongside GPU inference optimizations including KV caching, quantization, flash attention, and speculative decoding.
✓Interactive visualizations are used throughout to build intuition for transformer mechanics including token-by-token generation, sampling, and attention—topics typically difficult to grasp through theory alone.

More from DeepLearning.AI

📰Today⚡Feed📡Signals💰Capital
The Complete Guide to Transformers Just Dropped! 👉 New Course — DeepLearning.AI | subtl