Open brando90 opened 2 years ago
Hi @brando90,
Sorry for my latency. Not very familiar this this library, so not sure what you exactly want to achieve, but if you want to wrap Meta-Dataset in an l2l loader, I guess you could have a look at the way I construct the ZipDataset: https://github.com/mboudiaf/pytorch-meta-dataset/blob/61e8555dda0ef75e80cf3d03cb0d7a6822b0e082/src/datasets/pipeline.py#L240 and see if that be interfaced with l2l library. Please provide more details and we can probably work that out together :)
Malik
HI Malik!
Appreciate the response!
It would be awesome to work together on this. However, something change in my current things that are essential and meta-data set is not it unfortunately :( I do think it will be in the future and I post again here and we can tackle this problem then!
Thanks for the message and hope to post back soon in a couple of months :)
On Apr 16, 2022, at 1:16 PM, Malik Boudiaf @.***> wrote:
Hi @brando90 https://github.com/brando90,
Sorry for my latency. Not very familiar this this library, so not sure what you exactly want to achieve, but if you want to wrap Meta-Dataset in an l2l loader, I guess you could have a look at the way I construct the ZipDataset: https://github.com/mboudiaf/pytorch-meta-dataset/blob/61e8555dda0ef75e80cf3d03cb0d7a6822b0e082/src/datasets/pipeline.py#L240 https://github.com/mboudiaf/pytorch-meta-dataset/blob/61e8555dda0ef75e80cf3d03cb0d7a6822b0e082/src/datasets/pipeline.py#L240 and see if that be interfaced with l2l library. Please provide more details and we can probably work that out together :)
Malik
— Reply to this email directly, view it on GitHub https://github.com/mboudiaf/pytorch-meta-dataset/issues/15#issuecomment-1100724701, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAOE6LREY3BBN63DE6W7PG3VFL7XXANCNFSM5S464DPA. You are receiving this because you were mentioned.
No worries, let's keep that issue open for the future ! :)
Is there learn2learn support for this https://github.com/learnables/learn2learn/issues/286
I am happy to help add it. What would be a good starting point?