![slide_bn](https://cdn.statically.io/img/hancomacademy.com/_img/comn/1_slide_bn.png)
![slide_bn02](https://cdn.statically.io/img/hancomacademy.com/_img/comn/slide_bn02.png)
NVIDIA DEEP LEARNING INSTITUTE
NVIDIA Deep Learning Institute
NVIDIA DEEP LEARNING INSTITUTE
���ľ�ī���̴� ���� ������ NVIDIA DLI(Deep Learning Institute) ���� �������ͷ�, Multi GPU, CUDA �� NVIDIA�� ����� ������ �ǽ� ���� ������ �����մϴ�.
NVIDIA DLI�� ������, ������ ������ �� ���������� �ΰ����� �� ����ȭ�� ��ǻ���� ����Ͽ� �������� �ڵ���, �コ�ɾ�, �繫 �� �پ��� ������ ���� ������ �ذ��ϴ� ����� �����մϴ�.
������ ���������� �ֽ� ����, ������ ��ũ �� ����� ����Ͽ� �Ű� ��Ʈ��ũ�� �н��ϰ�, ����ȭ �� �����ϴ� ����� �н��� �� �ֽ��ϴ�. ���� ��ǻ�� ���¿����� �������� ���� ���α� �������� GPU ���� ��ǻ�� ���� ���α��� ��, ����ȭ, ����ȭ �� �����ϴ� ����� ���� �˴ϴ�.
![logo](https://cdn.statically.io/img/hancomacademy.com/../_img/cont/nvidia_logo02.png)
NVIDIA DLI ���� ����
���ľ�ī���̴� ��ٷο� �ɻ縦 ����� DLI ���� ���縦 �����ϰ� �ֽ��ϴ�.
![learn](https://cdn.statically.io/img/hancomacademy.com/../_img/cont/nvidia_learn2.jpg)
NVIDIA DLI ���� ����
Fundamentals of Accelerated Computing With CUDA C/C++(6h)
- Accelerating Applications with CUDA C/C++ (120��)
- Managing Accelerated Application Memory with CUDA C/C++ (120��)
- Asynchronous Streaming and Visual Profiling for Accelerated Applications with CUDA C/C++ (120��)
Fundamentals of Accelerated Computing With CUDA Python(6h)
- Introduction to CUDA Python with Numba (120')
- Custom CUDA Kernels in Python with Numba(120')
- RNG, Multidimensional Grids, and Shared Memory for CUDA Python with Numba (120')
Accelerating CUDA�� C++ Applications With Multiple GPUs(6h)
- Using JupyterLab (15')
- Application Overview (15')
- Introduction to CUDA Streams (90')
- Copy/Compute Overlap with CUDA Streams (90')
- Multiple GPUs with CUDA C++ (60')
- Copy/Compute Overlap with Multiple GPUs (60')
- Course Assessment (30')
Fundamentals of DeepLearning(6h)
- The Mechanics of Deep Learning (120')
- Pre-trained Models and Recurrent Networks(120')
- Final Project: Object Classification (120')
Building AI-Based Cybersecurity Pipelines(6h)
- An Overview of the NVIDIA Morpheus AI Framework (30')
- Morpheus Pipeline Construction (45')
- Inference in Morpheus Pipelines (45')
- Case Study: AI-Based Machine Log Parsing at Splunk (30')
- Digital Fingerprinting Pipeline (45')
- Time Series Analysis (45')
- Case Study: Cybersecurity Flyaway Kit at Booz Allen Hamilton (30')
- Assessment 1: Test Your Understanding(45')
- Assessment 2: Practical Demonstration (45')
Model Parallelism: Building and Deploying Large Neural Networks(6h)
- Introduction to Training of Large Models (120��)
- Model Parallelism: Advanced Topics (120��)
- Inference of Large Models (120��)
Data Parallelism: How to Train Deep Learning Models on Multiple GPUs(6h)
- Stochastic Gradient Descent and the Effects of Batch Size (120')
- Training on Multiple GPUs with PyTorch Distributed Data Parallel (DDP) (120')
- Maintaining Model Accuracy when Scaling to Multiple GPUs (90')
- Workshop Assessment(30')
���� ��û�ϱ� �����Ұ��� �ٿ�ε�