fatih akyon

it may be related to this issue: https://github.com/nodesource/distributions/issues/1266

This is a very vital issue in our case too, waiting for the response.

What is your stance on this @Narsil ?

@glenn-jocher how can i visualize iteration based train loss in wandb?

@glenn-jocher by iteration based, i mean logging the training loss **at every 100 iteration**, for instance. Other frameworks such as **mmdetection** send the **training loss logs after each iteration windows**...

@glenn-jocher i see, thanks for the fast response!

@patil-suraj what is needed to migrate this project into huggingface datasets library since the nlp library is deprecated? I can work on a PR if you guide me on this.

@sherlocked777 you can use [labelme2coco](https://pypi.org/project/labelme2coco/) package to convert your annotations to coco format. It is really easy. Installation: ```bash pip install labelme2coco ``` Usage: ```python # import package import labelme2coco...

@RiyazAina-DeepML you should use [labelme2coco](https://pypi.org/project/labelme2coco/) package to convert your annotations to coco format. It is really easy. Installation: ```bash pip install labelme2coco ``` Usage: ```python # import package import labelme2coco...

@wkentaro no, you have to check if the given output is a directory with `os.path.isdir(output)` instead of `output.endswith(osp.sep)`. Because `output.endswith(osp.sep)` fails on windows while `os.path.isdir(output)` works on linux/mac/windows.