Jaxonmodels
🚨 This library is still under HEAVY development and won't reach version 1.0.0 in a long time!
This library consists of deep learning model implementations in JAX using Equinox as the neural network library.
The goal of this library is to provide simple, yet performant and easy to understand implementations with the aim to give exactly the same output as their Pytorch counterparts. As such, great emphasis is placed on making sure that the layers and the models behave accordingly.
Using statedict2pytree we can also load the Pytorch model weights into the JAX models.
Some models will have inadvertently repeated code, but this is fine so long as the model remains self contained for the most part.
Implemented Models
These models have been implemented:
- AlexNet
- CLIP
- EfficientNet
- ResNet
- ViT
- Mamba
- ConvNext
- Swin Transformer
- Siglip (in progress)
- VQ-VAE
- ESMC
- ESM3
Contributing
If you have a model that you would like to include, then just open up a PR. It should contain your model and ideally a few tests showcasing that the model (and its components) behave like their Pytorch versions.
1 maintainer
Metadata
1 maintainer
Release files for jaxonmodels 0.5.5
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| jaxonmodels-0.5.5.tar.gz | 61.5 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| jaxonmodels-0.5.5-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 117.6 kB
Release files / jaxonmodels-0.5.5.tar.gz
| Download URL | jaxonmodels-0.5.5.tar.gz |
|---|---|
| Size | 61.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
025088e025ffc843b4bf558bba273b565714ac5e15686839694623e7ca5b55f6
|
|
BLAKE2b-256 checksum How to use checksums |
8fc11bda65e1150c53016bee98bd217b19d811b209e08010b768dbca8160103f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.9.7
|
Release files / jaxonmodels-0.5.5-py3-none-any.whl
| Download URL | jaxonmodels-0.5.5-py3-none-any.whl |
|---|---|
| Size | 56.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
16fd2ccccd210ea1a5be03bc747610f287db11ea96d3b1d50373e65e2c571a35
|
|
BLAKE2b-256 checksum How to use checksums |
9b602eae1f02b7e0a581c7b5d78c73cb555bf2524950242a7f3ceea65f9b53e2
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
uv/0.9.7
|
