Tutorial 22_Image_Captioning.ipynb is not working in Colab because of disk space limit
neuralcomputer opened this issue · comments
Tutorial 22_Image_Captioning.ipynb is not working in Google Colab because of disk space limit. It fails on dataset unzipping line. Colab provide only 68Gb with GPU backend (25-30 already wasted).
Is it possible to provide a trained generator and\or saved transfer-values files to avoid downloading images themselves?
I think it was maybe 4 years ago that I made that tutorial, so it's a lot to ask for me to do follow-up work on it now. If downloading large files is the problem, then you could try and go to a public library with a good internet connection. You are very welcome to modify the tutorial and make it available to others on your own github account, with the modifications you would prefer.
Thanks for your answer.
Cache files of train and validation transfer-values available at:
- transfer_values_train.pkl (900Mb): https://drive.google.com/file/d/1z-IIMhEccKO6IbEOeWa2QUP37NamYEVb/view?usp=sharing
- transfer_values_val.pkl (40Mb): https://drive.google.com/file/d/1H0XgMkpm1H3Qy5zYAJFByUwoOVc8IY6j/view?usp=sharing