autograd
qha
autograd | qha | |
---|---|---|
6 | 1 | |
6,797 | 28 | |
0.7% | - | |
6.0 | 8.3 | |
7 days ago | 3 months ago | |
Python | Python | |
MIT License | GNU General Public License v3.0 only |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
autograd
-
JAX – NumPy on the CPU, GPU, and TPU, with great automatic differentiation
Actually, that's never been a constraint for JAX autodiff. JAX grew out of the original Autograd (https://github.com/hips/autograd), so differentiating through Python control flow always worked. It's jax.jit and jax.vmap which place constraints on control flow, requiring structured control flow combinators like those.
-
Autodidax: Jax Core from Scratch (In Python)
I'm sure there's a lot of good material around, but here are some links that are conceptually very close to the linked Autodidax.
There's [Autodidact](https://github.com/mattjj/autodidact), a predecessor to Autodidax, which was a simplified implementation of [the original Autograd](https://github.com/hips/autograd). It focuses on reverse-mode autodiff, not building an open-ended transformation system like Autodidax. It's also pretty close to the content in [these lecture slides](https://www.cs.toronto.edu/~rgrosse/courses/csc321_2018/slid...) and [this talk](http://videolectures.net/deeplearning2017_johnson_automatic_...). But the autodiff in Autodidax is more sophisticated and reflects clearer thinking. In particular, Autodidax shows how to implement forward- and reverse-modes using only one set of linearization rules (like in [this paper](https://arxiv.org/abs/2204.10923)).
Here's [an even smaller and more recent variant](https://gist.github.com/mattjj/52914908ac22d9ad57b76b685d19a...), a single ~100 line file for reverse-mode AD on top of NumPy, which was live-coded during a lecture. There's no explanatory material to go with it though.
-
Numba: A High Performance Python Compiler
XLA is "higher level" than what Numba produces.
You may be able to get the equivalent of jax via numba+numpy+autograd[1], but I haven't tried it before.
IMHO, jax is best thought of as a numerical computation library that happens to include autograd, vmapping, pmapping and provides a high level interface for XLA.
I have built a numerical optimisation library with it, and although a few things became verbose, it was a rather pleasant experience as the natural vmapping made everything a breeze, I didn't have to write the gradients for my testing functions, except for special cases that involved exponents and logs that needed a bit of delicate care.
[1] https://github.com/HIPS/autograd
-
Run Your Own DALL·E Mini (Craiyon) Server on EC2
Next, we want the code in the https://github.com/hrichardlee/dalle-playground repo, and we want to construct a pip environment from the backend/requirements.txt file in that repo. We were almost able to use the saharmor/dalle-playground repo as-is, but we had to make one change to add the jax[cuda] package to the requirements.txt file. In case you haven’t seen jax before, jax is a machine-learning library from Google, roughly equivalent to Tensorflow or PyTorch. It combines Autograd for automatic differentiation and XLA (accelerated linear algebra) for JIT-compiling numpy-like code for Google’s TPUs or Nvidia’s CUDA API for GPUs. The CUDA support requires explicitly selecting the [cuda] option when we install the package.
-
Trade-Offs in Automatic Differentiation: TensorFlow, PyTorch, Jax, and Julia
> fun fact, the Jax folks at Google Brain did have a Python source code transform AD at one point but it was scrapped essentially because of these difficulties
I assume you mean autograd?
https://github.com/HIPS/autograd
-
JAX - COMPARING WITH THE BIG ONES
These four points lead to an enormous differentiation in the ecosystem: Keras, for example, was originally thought to be almost completely focused on point (4), leaving the other tasks to a backend engine. In 2015, on the other hand, Autograd focused on the first two points, allowing users to write code using only "classic" Python and NumPy constructs, providing subsequently many options for point (2). Autograd's simplicity greatly influenced the development of the libraries to follow, but it was penalized by the clear lack of the points (3) and (4), i.e. adequate techniques to speed up the code and sufficiently abstract modules for neural network development.
qha
-
Numba: A High Performance Python Compiler
Software from our group (cij[1], qha[2]) were developed when numba seems to be the best option for JIT. It generates more pain in the hindsight. It generates a lot of depreciated warning due to unstable API, locked numpy to a certain version (i remember 1.21) due to compatibility issues, and when M1 Mac comes out, there were for a long time lack of llvmlite porting to the new platform, so cannot run on these new Macs.
If I had to do it again I would just use plain numpy or use the JAX from Google if JIT is really necessary.
[1]: https://github.com/MineralsCloud/cij
[2]: https://github.com/MineralsCloud/qha
What are some alternatives?
Enzyme - High-performance automatic differentiation of LLVM and MLIR.
pure_numba_alias_sampling - Pure numba version of Alias sampling algorithm from L. Devroye's, "Non-Uniform Random Random Variate Generation"
SwinIR - SwinIR: Image Restoration Using Swin Transformer (official repository)
hn-search - Hacker News Search
jaxonnxruntime - A user-friendly tool chain that enables the seamless execution of ONNX models using JAX as the backend.
fbpic - Spectral, quasi-3D Particle-In-Cell code, for CPU and GPU
autodidact - A pedagogical implementation of Autograd
dalle-flow - 🌊 A Human-in-the-Loop workflow for creating HD images from text
AD-Rosetta-Stone - Examples of Automatic Differentiation (AD) in many different languages and systems
Keras - Deep Learning for humans