UNet Tutorial in JAX
Вставка
- Опубліковано 14 чер 2024
- UNets are a famous architecture for image segmentation. With their hierarchical structure they have a wide receptive field. Similar to multigrid methods, we will use them in this video to solve the Poisson equation in the Equinox deep learning framework. Here is the code: github.com/Ceyron/machine-lea...
-------
👉 This educational series is supported by the world-leaders in integrating machine learning and artificial intelligence with simulation and scientific computing, Pasteur Labs and Institute for Simulation Intelligence. Check out simulation.science/ for more on their pursuit of 'Nobel-Turing' technologies (arxiv.org/abs/2112.03235 ), and for partnership or career opportunities.
-------
📝 : Check out the GitHub Repository of the channel, where I upload all the handwritten notes and source-code files (contributions are very welcome): github.com/Ceyron/machine-lea...
📢 : Follow me on LinkedIn or Twitter for updates on the channel and other cool Machine Learning & Simulation stuff: / felix-koehler and / felix_m_koehler
💸 : If you want to support my work on the channel, you can become a Patreon here: / mlsim
🪙: Or you can make a one-time donation via PayPal: www.paypal.com/paypalme/Felix...
-------
⚙️ My Gear:
(Below are affiliate links to Amazon. If you decide to purchase the product or something else on Amazon through this link, I earn a small commission.)
- 🎙️ Microphone: Blue Yeti: amzn.to/3NU7OAs
- ⌨️ Logitech TKL Mechanical Keyboard: amzn.to/3JhEtwp
- 🎨 Gaomon Drawing Tablet (similar to a WACOM Tablet, but cheaper, works flawlessly under Linux): amzn.to/37katmf
- 🔌 Laptop Charger: amzn.to/3ja0imP
- 💻 My Laptop (generally I like the Dell XPS series): amzn.to/38xrABL
- 📱 My Phone: Fairphone 4 (I love the sustainability and repairability aspect of it): amzn.to/3Jr4ZmV
If I had to purchase these items again, I would probably change the following:
- 🎙️ Rode NT: amzn.to/3NUIGtw
- 💻 Framework Laptop (I do not get a commission here, but I love the vision of Framework. It will definitely be my next Ultrabook): frame.work
As an Amazon Associate I earn from qualifying purchases.
-------
Timestamps:
00:00 Intro
01:13 About the 1d Poisson problem
02:23 Intro to the UNet architecture
07:04 Concrete scenario for this video
09:30 Imports
11:23 Data generation & visualization
19:28 Implement UNet architecture
47:07 Check implementation
50:16 Training Loop
57:22 Inspect Loss history
59:00 Qualitative inspection
01:01:33 Test error metric
01:04:09 Outro
What a great video so far !!!!! 😮
Thank you!! 😁
Thank you for the content!!! 😊😊
My pleasure!!
Are you german? Sound very german!
Great videos BTW!
Yes 😁
Thanks for the kind comment 😊
would you learn jax or pytorch as someone wnating to implement everything from scratch as trainig.
Great question! 😊 Both are valid frameworks to do so. I prefer JAX because it is lower level, not monolithic, and better suited for scientific ML. It also has features that PyTorch does not have (or at least not yet as good): automatic vectorization, forward mode AD, higher order AD, IMHO a better (and mathematically more grounded) approach to autodiff, exact implementation of the numpy API (not one that differs at some point), and it has the amazing concept of pytree. Some Microbenchmarks (don't have the links unfortunately) showed that JAX produced slightly faster execution times over PyTorch. For most real-world cases, that likely does not matter though.
However, JAX has a steeper learning curve. You need to be more familiar with concepts of Python. Likely, there are also way more tutorials and papers using PyTorch.
Both choices are valid, if you have more time to invest, JAX is a great choice. I don't think I will make PyTorch videos on the channel
@@MachineLearningSimulation nice, i d bias towards jax anyway, thx for vids, i want to learn to be like you in ml n simulator
@@MachineLearningSimulation How good is the future of Jax as Compare to Pytorch ?