hummingbird
swift-evolution
hummingbird | swift-evolution | |
---|---|---|
9 | 125 | |
3,304 | 15,030 | |
0.5% | 0.5% | |
7.1 | 9.7 | |
17 days ago | 3 days ago | |
Python | Markdown | |
MIT License | Apache License 2.0 |
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.
hummingbird
- Treebomination: Convert a scikit-learn decision tree into a Keras model
-
[D] GPU-enabled scikit-learn
If are interested in just predictions you can try Hummingbird. It is part of the PyTorch ecosystem. We get already trained scikit-learn models and translate them into PyTorch models. From them you can run your model on any hardware support by PyTorch, export it into TVM, ONNX, etc. Performance on hardware acceleration is quite good (orders of magnitude better than scikit-learn is some cases)
-
Machine Learning with PyTorch and Scikit-Learn – The *New* Python ML Book
I think Rapids AI's cuML tried to go into this direction (essentially scikit-learn on the GPU): https://docs.rapids.ai/api/cuml/stable/api.html#logistic-reg.... For some reason it never took really off though.
Btw., going on a tangent, you might like Hummingbird (https://github.com/microsoft/hummingbird). It allows you trained scikit-learn tree-based models to PyTorch. I watched the SciPy talk last year, and it's a super smart & elegant idea.
-
Export and run models with ONNX
ONNX opens an avenue for direct inference using a number of languages and platforms. For example, a model could be run directly on Android to limit data sent to a third party service. ONNX is an exciting development with a lot of promise. Microsoft has also released Hummingbird which enables exporting traditional models (sklearn, decision trees, logistical regression..) to ONNX.
-
Supreme Court, in a 6–2 ruling in Google v. Oracle, concludes that Google’s use of Java API was a fair use of that material
And Python.
-
[D] Here are 3 ways to Speed Up Scikit-Learn - Any suggestions?
For inference, you can convert your models to other formats that support GPU acceleration. See Hummingbird https://github.com/microsoft/hummingbird
-
[D] Microsoft library, Hummingbird, compiles trained ML models into tensor computation for faster inference.
The surprising thing is that Hummingbird can be faster than the GPU implementation of LightGBM (and XGBoost) if you use tensor compilers such as TVM. [The paper](https://www.usenix.org/conference/osdi20/presentation/nakandala) describes our findings. We have also open sourced the [benchmark code](https://github.com/microsoft/hummingbird/tree/main/benchmarks) so you try yourself!
-
I learned about Microsoft's Hummingbird library today. 1000x performance??
I took their sample code from Github and tweaked it to spit out times for each model's prediction, as well as increase the number of rows to 5 million. I used Google's Colab and selected GPU for my hardware accelerator. This gives an option to run code on GPU, not that all computations will happen on the GPU.
swift-evolution
-
Swift's native Clocks are inefficient
According to their changelog[0], Clock was added to the standard library with Swift 5.7, which shipped in 2022, at the same time as iOS 16. It looks like static linking by default was approved[1] but development stalled[2].
I expect that it's as simple as that: It's supported on iOS 16+ because it's dynamically linked by default, against a system-wide version of the standard library. You can probably try to statically link newer versions on old OS versions, or maybe ship a newer version of the standard library and dynamically link against that, but I have no idea how well those paths are supported.
0. https://github.com/apple/swift/blob/main/CHANGELOG.md
1. https://github.com/apple/swift-evolution/blob/main/proposals...
2. https://github.com/apple/swift-package-manager/pull/3905
-
Byte-Sized Swift: Building Tiny Games for the Playdate
[A Vision for Embedded Swift](https://github.com/apple/swift-evolution/blob/main/visions/e...) has the details on this new build mode and is quite interesting.
> Effectively, there will be two bottom layers of Swift, and the lower one, “non-allocating” Embedded Swift, will necessarily be a more restricted compilation mode (e.g. classes will be disallowed as they fundamentally require heap allocations) and likely to be used only in very specialized use cases. “Allocating” Embedded Swift should allow classes and other language facilities that rely on the heap (e.g. indirect enums).
Also, this seems to maybe hint at the Swift runtime eventually being reimplemented in non-allocating Embedded Swift rather than the C++ (?) that it uses now:
> The Swift runtime APIs will be provided as an implementation that’s optimized for small codesize and will be available as a static library in the toolchain for common CPU architectures. Interestingly, it’s possible to write that implementation in “non-allocating” Baremetal Swift.
-
Borrow Checking Without Lifetimes
I may be out of my depth here as I've only casually used Rust, but this seems similar to Swift's proposed lifetime dependencies[1]. They're not in the type system formally so maybe they're closer to poloneius work
[1]: https://github.com/apple/swift-evolution/blob/3055becc53a3c3...
-
Functional Ownership Through Fractional Uniqueness
Swift recently adopted a region-based approach for safe concurrency that builds on Milano et al’s ideas: https://github.com/apple/swift-evolution/blob/main/proposals...
- Swift-evolution/proposals/0373-vars-without-limits-in-result-builders.md
- The Swift proposal that removed the ++ and –- operators (2017)
-
Crafting Self-Evident Code with D
No, it's not. Refcounting CAN be a garbage collection algorithm, but in Swift it's deterministic and done at compile time. Not to mention recently added support for non-copyable types that enforces unique ownership: https://github.com/apple/swift-evolution/blob/main/proposals...
- Statically link Swift runtime libraries by default on supported platforms
- (5.9) What is the point of a SerialExecutor that can silently re-order jobs?
-
Mac shipments grow 10%, as all major PC brands see downturns.
You can stackallocate buffers with unsafe Swift but it's not exactly fun to use. https://github.com/apple/swift-evolution/blob/main/proposals/0322-temporary-buffers.md
What are some alternatives?
onnx - Open standard for machine learning interoperability
compose-multiplatform - Compose Multiplatform, a modern UI framework for Kotlin that makes building performant and beautiful user interfaces easy and enjoyable.
swift - The Swift Programming Language
foundationdb - FoundationDB - the open source, distributed, transactional key-value store
sentence-transformers - Multilingual Sentence & Image Embeddings with BERT
kotlinx-datetime - KotlinX multiplatform date/time library
cuml - cuML - RAPIDS Machine Learning Library
okio - A modern I/O library for Android, Java, and Kotlin Multiplatform.
docker - Docker - the open-source application container engine
PeopleInSpace - Kotlin Multiplatform project with SwiftUI, Jetpack Compose, Compose for Wear, Compose for Desktop, Compose for Web and Kotlin/JS + React clients along with Ktor backend.
chemprop - Message Passing Neural Networks for Molecule Property Prediction
swift-algorithms - Commonly used sequence and collection algorithms for Swift