Twitter threads
I sometimes get asked for a list of the more memorable Twitter threads and tweets that I wrote and hence I decided to put it together. Here it is 🙂
learning resources
- Can you start learning cutting-edge deep learning without specialized hardware?
- Summary of BatchNorm, LayerNorm, InstanceNorm and GroupNorm
- How to debug in Jupyter notebook
- What differentiates fast.ai students who do really well from the rest?
- Ever wondered what a class in Python is and how come everything is an object?
- Why bottlenecks in NN design are important?
- Everything you never knew you wanted to know about Python decorators (and more)!
- This is what overemphasizing math when learning ML can lead to
- How to win at Kaggle.
- Thoughts on the fastai v4 deep learning course
- [notes] outstanding lecture on ethics by Rachel Thomas
- [notes] how to make your workplace more inclusive
- [notes] Journey to Kaggle #1 — a talk by Philipp Singer
- How I started blogging
- How to process pandas DataFrames in parallel
- Can one grab a matplotlib plot as a numpy array?
- How to work with sound in Python? 🎶🎙️🥁
- Using the validation set for training
- Adding functionality to fastai on the fly
- Video by Yannic Kilcher on the "Attention is all you need" paper
- Links to a great set of resources on introduction to probability
- What is hard negative mining?
- What does it mean for UMAP to preserve local and global structure?
- What are packed sequences in PyTorch?
- Excel-style conditional formatting using pandas
projects done with fast.ai
- Implementation of a paper on sperm whale bioacoustics using fastai (Random Forest)
- A thread of people sharing their projects
- Classification / metric learning using fastai
- Learning an embedding space for locations by @sentiance
- Tattoo removal from images by Vijish Madhavan
- Class activation mapping
assorted thoughts
- Why did I write Meta Learning?
- [notes] A vision for the future of ML frameworks by Soumith Chintala
- [notes] Modern Artificial Intelligence 1980s - 2021 by Jürgen Schmidhuber
- [notes] Art in Light of AI — a talk on AI art by Helena Sarin
- Being new to something is a superpower 🦸
- What I thought becoming employable in deep learning would be vs what it ended up being
- "We do not learn from experience… we learn from reflecting on experience." - John Dewey
- How I got my first two DL assignments
- I'm falling behind in fast.ai lectures, what do I do?
- Everything worth learning feels too hard to learn before we learn it.
- The Origin of Wealth — a great book to read with an extensive section on algorithms
- You can't learn a profession by studying a textbook.
- school vs life
- How to build your personal website and serve it using GithubPages for free