pytorch-image-models

pytorch-image-models copied to clipboard

Hi, just a very small typo fix :)

ChristophReich1996

[FEATURE] Detection models with loss functions and trainers

8

It would be great if this repo started benchmarking detection models such as yolo family and Detr on COCO dataset. Big ask I know but it would be fantastic.

pfeatherstone

Can you add "An Improved One millisecond Mobile Backbone" (MobileOne)

5

As above

ghoshaw

hack ln implementation in convnext

28

See https://github.com/pytorch/pytorch/issues/71465 Slightly changes LayerNorm2d implementation, 1) currently when ln2d is called on a contiguous tensor, it accidentally turns most of the network into channels last mode, line 114 undoes...

ngimel

Should Fix errors like RuntimeError: Error(s) in loading state_dict for EfficientDet: Missing key(s) in state_dict: "backbone.conv_stem.weight", "backbone.bn1.weight", ... Unexpected key(s) in state_dict: "model.model.backbone.conv_stem.weight", "model.model.backbone.bn1.weight", "model.model.backbone.bn1.bias", "model.model.backbone.bn1.running_mean", "model.model.backbone.bn1.running_var", "model.model.backbone.bn1.num_batches_tracked", "model.model.backbone.blocks.0.0.conv_dw.weight",

AndreyGurevich

the code is released in [https://github.com/hustvl/TopFormer](url)

wmkai

Straight forward conversion of models trained with timm {all checkpoints; model_best; last; specific_model_file}

swenkel

[FEATURE] Adding Squeezenet in timm

2

Is there a plan for adding [Squeezenet](https://pytorch.org/hub/pytorch_vision_squeezenet/) as part of `timm` library?

nahidalam

1 Channel images not working

7

I am using `timm` version `0.5.0`, and I plan to train a greyscale image, but the data loader is outputting RGB images. The model I intend to use works fine...

etetteh

[FEATURE] Add ViT GSAM

2

Is your feature request related to a problem? Please describe. ViT GSAM checkpoint have been released in https://console.cloud.google.com/storage/browser/vit_models/gsam

banda-larga