dart_pytorch 0.1.2 
dart_pytorch: ^0.1.2 copied to clipboard
A minimal PyTorch-style tensor library for Dart with reverse-mode autograd, hand-written CUDA kernels via dart:ffi, and an nn.Module scaffolding for Transformers, GPT, MoE, ViT, and LC0 chess-network [...]
We analyzed this package 11 hours ago, and awarded it 130 pub points (of a possible 160):
0/10 points: Provide a valid pubspec.yaml
The package description is too long.
Search engines display only the first part of the description. Try to keep the value of the description field in your package's pubspec.yaml file between 60 and 180 characters.
5/5 points: Provide a valid README.md
5/5 points: Provide a valid CHANGELOG.md
10/10 points: Use an OSI-approved license
Detected license: MIT.
10/10 points: 20% or more of the public API has dartdoc comments
845 out of 1972 API elements (42.8 %) have documentation comments.
Some symbols that are missing documentation: adam.Adam, adam.Adam.Adam.new, adam.Adam.beta1, adam.Adam.beta2, adam.Adam.eps.
0/10 points: Package has an example
No example found.
See package layout guidelines on how to add an example.
20/20 points: Supports 5 of 6 possible platforms (iOS, Android, Web, Windows, macOS, Linux)
-
✓ Android
-
✓ iOS
-
✓ Windows
-
✓ Linux
-
✓ macOS
These platforms are not supported:
Package not compatible with platform Web
Because:
package:dart_pytorch/dart_pytorch.dartthat imports:package:dart_pytorch/core/vector_store/faiss_io.dartthat imports:package:dart_pytorch/core/vector_store/vector_transform.dartthat imports:package:dart_pytorch/core/vector_store/index_io.dartthat imports:dart:io
40/50 points: code has no errors, warnings, lints, or formatting issues
Found 62 issues. Showing the first 2:
INFO: Use 'isEmpty' instead of 'length' to test whether the collection is empty.
bin/_gpt2_debug.dart:25:10
╷
25 │ return xs.length == 0 ? 0 : (s / xs.length);
│ ^^^^^^^^^^^^^^
╵
To reproduce make sure you are using the lints_core and run dart analyze bin/_gpt2_debug.dart
INFO: The local variable '_mx' starts with an underscore.
bin/_gpt2_debug.dart:109:16
╷
109 │ double _mx(List<double> xs) {
│ ^^^
╵
To reproduce make sure you are using the lints_core and run dart analyze bin/_gpt2_debug.dart
10/10 points: All of the package dependencies are supported in the latest version
| Package | Constraint | Compatible | Latest | Notes |
|---|---|---|---|---|
ffi |
^2.2.0 |
2.2.0 | 2.2.0 | |
image |
^4.8.0 |
4.9.2 | 4.9.2 | |
path |
^1.9.0 |
1.9.1 | 1.9.1 |
Transitive dependencies
| Package | Constraint | Compatible | Latest | Notes |
|---|---|---|---|---|
archive |
- | 4.3.0 | 4.3.0 | |
meta |
- | 1.19.0 | 1.19.0 | |
posix |
- | 6.5.2 | 6.5.2 |
To reproduce run dart pub outdated --no-dev-dependencies --up-to-date --no-dependency-overrides.
10/10 points: Package supports latest stable Dart and Flutter SDKs
20/20 points: Compatible with dependency constraint lower bounds
pub downgrade does not expose any static analysis error.
Analyzed with Pana 0.23.19, Dart 3.13.1.
Check the analysis log for details.
Weekly downloads
Metadata
Documentation
Publisher
unverified uploader
Weekly Downloads
Metadata
A minimal PyTorch-style tensor library for Dart with reverse-mode autograd, hand-written CUDA kernels via dart:ffi, and an nn.Module scaffolding for Transformers, GPT, MoE, ViT, and LC0 chess-network inference.
Repository (GitHub)
View/report issues
Topics
#machine-learning #deep-learning #tensor #cuda #autograd
License
MIT (license)