Development of prototype based on NVIDIA DIGITS for PyTorch
Deep learning frameworks such as Caffe and Tensorflow has significantly eased the process of creating deep neural networks. However, these frameworks do not come with a built in visual interface which can be difficult to use. NVIDIA Deep Learning GPU Training System (DIGITS) bridges the gap by provi...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/75377 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Deep learning frameworks such as Caffe and Tensorflow has significantly eased the process of creating deep neural networks. However, these frameworks do not come with a built in visual interface which can be difficult to use. NVIDIA Deep Learning GPU Training System (DIGITS) bridges the gap by providing a Graphic User Interface (GUI) for using deep learning frameworks. This report presents the process behind building a prototype based on NVIDIA DIGITS for a new framework, PyTorch. This prototype aims to have basic functionalities to create a dataset and to train a neural network model. The report covers three main sections. First, the NVIDIA DIGITS section will cover the basic features and information on using the interface. Secondly, the Materials and Methods section will cover the hardware and software used in the project and the brief outline of the procedures taken to develop the prototype. Thirdly, the Results and Discussion section covers the results from the prototype and discussions on how it can be further improved towards full integration of the PyTorch framework. |
|---|