Thank you for sharing the code. I successfully ran the code on a regression datasets (tsra). Additionally, I also explored two ways to run my custom dataset
a) I created my own .ts file, although, as you mentioned in other discussions, this might not be necessary and I can create a custom class to load the dataset and simply update the 'readdata' function.
b) I developed a custom class that traverses the directory and loads all the pickle datasets (multiple files). However, I've noticed that it loads all the data regardless of the batch size. Since I have multiple files, each ranging from a few megabytes to gigabytes in size, it can be a bottleneck to load all the data at once into memory. Is there any workaround for this? I was wondering dataloader but if it was already implemented or if you have explored then will appreciate your insights on this.
Hi gzerveas,
Thank you for sharing the code. I successfully ran the code on a regression datasets (tsra). Additionally, I also explored two ways to run my custom dataset
a) I created my own .ts file, although, as you mentioned in other discussions, this might not be necessary and I can create a custom class to load the dataset and simply update the 'readdata' function.
b) I developed a custom class that traverses the directory and loads all the pickle datasets (multiple files). However, I've noticed that it loads all the data regardless of the batch size. Since I have multiple files, each ranging from a few megabytes to gigabytes in size, it can be a bottleneck to load all the data at once into memory. Is there any workaround for this? I was wondering dataloader but if it was already implemented or if you have explored then will appreciate your insights on this.
Thanks,