rwightman/pytorch-image-models: MaxxVit (CoAtNet, MaxVit, and related experimental weights)
Creators
- Ross Wightman
- Alexander Soare
- Aman Arora1
- Chris Ha2
- Christoph Reich3
- Nathan Raw4
- Jakub Kaczmarzyk5
- mrT23
- Mike
- SeeFun
- contrastive6
- Mohammed Rizin7
- Hyeongchan Kim8
- Csaba Kertész9
- Dushyant Mehta
- Guillem Cucurull
- Kushajveer Singh
- han10
- Yuki Tatsunami
- Andrew Lavin
- Juntang Zhuang
- Matthijs Hollemans11
- Sepehr Sameni12
- Vyacheslav Shults
- Wang, Xiao13
- Yonghye Kwon14
- Yusuke Uchida15
- Zhun Zhong16
- Comar
- Kim, Taehoon17
- 1. Weights & Biases
- 2. independent
- 3. Technical University of Darmstadt
- 4. @huggingface
- 5. Stony Brook Medicine
- 6. MIT
- 7. Kaggle Competition Master
- 8. @toss
- 9. Neuro Event Labs Oy
- 10. Ajou University
- 11. Indie developer
- 12. University of Bern
- 13. @NVIDIA
- 14. MarkAny
- 15. Mobility Technologies Co., Ltd.
- 16. Xiamen University
- 17. LG AI Research
Description
CoAtNet (https://arxiv.org/abs/2106.04803) and MaxVit (https://arxiv.org/abs/2204.01697) timm trained weights
Weights were created reproducing the paper architectures and exploring timm sepcific additions such as ConvNeXt blocks, parallel partitioning, and other experiments.
Weights were trained on a mix of TPU and GPU systems. Bulk of weights were trained on TPU via the TRC program (https://sites.research.google/trc/about/).
CoAtNet variants run particularly well on TPU, it's a great combination. MaxVit is better suited to GPU due to the window partitioning, although there are some optimizations that can be made to improve TPU padding/utilization incl using 256x256 image size (8, 8) windo/grid size, and keeping format in NCHW for partition attention when using PyTorch XLA.
Files
rwightman/pytorch-image-models-v0.1-weights-maxx.zip
Files
(1.4 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:0c5d6d60b7ba58ae3cae0218b2a4b02d
|
1.4 MB | Preview Download |
Additional details
Related works
- Is supplement to
- https://github.com/rwightman/pytorch-image-models/tree/v0.1-weights-maxx (URL)