Samvel K.

Results 10 comments of Samvel K.

> > It is currently possible to predict on new data in the UI by > * Enable "Use previous experiment weights" > * Select the dataset you want to...

> Here's a video (for next token prediction, classification problem type works similar) with the required steps. I've also expanded the steps above. We will probably provide a smoother solution...

Doesn't work on Kaggle :( Any suggestions how to fix it?

Up for specs in audio labeling

> @AlistairKeiller @samvelkoch @tolgayan, refer to @CO0l-github comment above, maybe this fixes? Works for me at Kaggle core. Thank you!

new problems appears at Kaggle for MultiModal after AG 1.1.1 release ``` %%capture !pip install autogluon %%capture !pip install -U 'aiobotocore[awscli,boto3]' ``` `from autogluon.multimodal import MultiModalPredictor` ``` `/opt/conda/lib/python3.10/site-packages/torchvision/io/image.py:13: UserWarning: Failed...

Works for me at Kaggle: `!pip install autogluon==1.1.0` `!pip install -U 'aiobotocore[awscli,boto3]'`

https://github.com/autogluon/autogluon/issues/3918 addon to the bug