Ayush Chaurasia
Ayush Chaurasia
@cyndyNKCM val.py already does both the things. It prints out the scores and also returns them. Here's val.py executing on a colab And here you can see that it returns...
@cyndyNKCM yes that line returns maps already. I think you're trying to import the functionality in a custom module. Maybe you should try something like this. ``` import val metrics...
Can you share the colab?
@cyndyNKCM you're not using the snippet that I shared..Please use that. I've used it like this before: ``` import val metrics = val.run(...) return metrics ``` Maybe what you're doing...
@thepycoder can you take a look please? @abcsunshine is this error related to clearml logger? Does it work fine without clearml?
@PraveenMNaik I like the idea of model visualizer too but I think there's a problem in creating such tool. The model definitions and structure need to be parsed and represented...
@RXZ-BUPT Have you tried training on non-transformers variants. In your console log, all train losses are nans so there might be a problem with th e training data. On training...
@ElefHead We currently have other higher priority tasks. We'll probably come back to this after 2 releases
@thepycoder Hey can you take a look whenever you're back? Thanks!
@glenn-jocher sure I think we can come up with a nice sweeps tutorial. Let's discuss the details in our meeting tonight.