Choosing the right Deep Learning Framework: A Deep Learning Approach
summary
Collected ~9,000 Jupyter notebooks from GitHub labeled by whether they reference TensorFlow or PyTorch, creating a dataset split ~57% TensorFlow / ~43% PyTorch; models compare 57 million parameters bag‑of‑words networks vs 5 million parameters text‑embedding networks, where embedding cuts epoch time from 420s to <60s and raises accuracy; Fleiss kappa ≈ 0.87; a Flask front end and published GitHub repo let users upload notebooks to get framework predictions.
AI-written summary · about
chapters (26)
- 0:00Introduction
- 0:35IBM Data Science Community
- 1:05Which Deep Learning Framework Should I be Using
- 2:07Deep Learning Framework Content
- 3:12subjectivity and anecdotal evidence
- 4:18agenda
- 5:03what is a deep learning framework
- 5:26why use a deep learning framework
- 6:43tensorflow
- 8:15case study
- 9:02PITorch
- 9:38Pokemon Gang
- 10:03Tensorflow vs Pie Charts
- 10:34Deep Learning Scale Example
- 10:47Deep Learning Framework
- 11:22PI Github
- 11:56Notebooks
- 13:02Dataset Stats
- 14:30Data Preprocessing
- 15:41Text Embedding
- 17:07Models
- 18:21Results
- 21:15Resources
- 23:20Github
- 24:18Data Science Community
- 24:45Call for Code
related talks
38mDeep Learning in Python: How to Choose a Framework?
42mDeep learning in Python: how to choose a framework?
01h 59mDeep Learning and Modern NLP
01h 40mA Deep Dive into NLP with PyTorch
17mF8 Refresh | Deep Learning with PyTorch
01h 30mIntroduction to Deep Learning & Natural Language Processing
32mDeep Learning Battle: TensorFlow 2 vs. PyTorch
52mMachine Learning Platforms