pytorch-image-models

pytorch-image-models copied to clipboard

**Is your feature request related to a problem? Please describe.** [ViTAE/ViTAEv2](https://github.com/ViTAE-Transformer/ViTAE-Transformer) has been open sourced, which implemented using timm. 

seefun

Before, the one_hot could only run in device='cuda'. Now it will run on input device automatically.

Leoooo333

Previously, this would only log a warning. ## Motivation Calling e.g. `timm.create_model('mobilenetv3_small_075', pretrained=True)` to retrieve a pre-trained model will sometimes fail silently (or: only with a warning), potentially leading to...

dnouri

Tkx for your great work, I found this repository not contain RepVGG-A0 and RepVGG-A1, could you add these models in this repository reference link [repvgg](https://github.com/DingXiaoH/RepVGG/blob/9f4de3c7a065f8d0e56ad73ee4c1f53df9e66014/repvgg.py#L196)

flywheel1412

**Is your feature request related to a problem? Please describe.** Add support for S^2-MLP: Spatial-Shift MLP Architecture for Vision. https://arxiv.org/abs/2106.07477 **Describe the solution you'd like** A clear and concise description...

Lloyd-Pottiger

[FEATURE] Creating Scheduler V2 (same as create_optimizer_v2 concept)

2

When building optimizers now, we can use the function `create_optimizer_v2` to take in keyword arguments instead of the parser arguments. This makes it easier to set up configuration files (can...

timothylimyl

[FEATURE] Add more Hybrid Vision Transformers

1

The timm library currently only has Vit and resnet as Hybrid Models. I would really love to See Swin transformers And EffNets As Hybrid ones Being added into the Library

king398

[FEATURE] Adding a column to CSVs to sort by architecture type

3

For certain contexts, it is interesting to look at only convolution-based models (and/or compare to attention-based architectures). To facilitate this, there could be an optional column in the [CSV](https://github.com/rwightman/pytorch-image-models/blob/master/results/results-imagenet.csv) files....

rasbt

[FEATURE] Validation labels txt or csv file

1

I have trained model using timm distributed_train.sh, but I have 0.1 validation accuracy on imagenet. Images in validation folder are all placed in root directory going like `ILSVRC2012_val_00000001.JPEG ... ILSVRC2012_val_00050000.JPEG`....

IgorKasianenko

[FEATURE] Vision Transformer Feature Extraction

7

It seems that ```VisionTransformer``` doesn't support feature extraction of all outputs in the ```forward_features``` method. Only returning of the cls token or [cls_token, distillation_token] is available [timm/models/vision_transformer.py#L291-L304](https://github.com/rwightman/pytorch-image-models/blob/23c18a33e4168dc7cb11439c1f9acd38dc8e9824/timm/models/vision_transformer.py#L291-L304). This functionality seems...

isaaccorley