Stanford CME295 – LLM Training Lecture (Transformers & Large Language Models)

عدد الدروس : 1 عدد ساعات الدورة : 01:47:27 شهادة معتمدة : نعم التسجيل في الدورة للحصول على شهادة

للحصول على شهادة

  • 1- التسجيل
  • 2- مشاهدة الكورس كاملا
  • 3- متابعة نسبة اكتمال الكورس تدريجيا
  • 4- بعد الانتهاء تظهر الشهادة في الملف الشخصي الخاص بك
An advanced Stanford lecture covering how large language models are trained, including optimization, scaling, and transformer-based training pipelines.
عن الدورة

This Stanford CME295 lecture focuses on the training process of large language models (LLMs) built using transformer architectures. It provides a deep technical understanding of how modern AI systems are trained at scale and optimized for performance.

You will begin by revisiting transformer fundamentals and how they are used as the backbone of large language models. The lecture explains how data flows through transformer layers and how models learn contextual relationships in language.

Next, the course explores the LLM training pipeline in detail. This includes data preparation, tokenization, batching strategies, and optimization techniques used during large-scale training. You will also learn how models are trained on massive datasets to capture language patterns and knowledge.

The lecture covers key training concepts such as loss functions, gradient descent, learning rates, and scaling laws that influence model performance. These ideas are essential for understanding how improvements in data and compute affect model quality.

In addition, the course discusses practical challenges in training LLMs, including computational cost, stability, and efficiency at scale.

By the end of this lecture, you will have a strong understanding of how large language models are trained from raw data into powerful AI systems capable of generating human-like text.

This course is ideal for students and researchers interested in deep learning, NLP, and large-scale AI training systems.