Closed dacian7 closed 1 year ago
of course, the repo should be updated later today!
fixed! Merge is here: https://github.com/lisadunlap/LADS/pull/14
Keep in mind our experiments are on a specific subset of DomainNet, you can run those by setting DATA.DATASET=DomainNetMini
. I also uploaded the embeddings for CUB, DomainNetMini, and Waterbirds if you don't want to regenerate them
fixed! Merge is here: #14
Keep in mind our experiments are on a specific subset of DomainNet, you can run those by setting
DATA.DATASET=DomainNetMini
. I also uploaded the embeddings for CUB, DomainNetMini, and Waterbirds if you don't want to regenerate them
Thanks for your help. But I still have some questions.
As you mentioned, experiments are on a specific subset of DomainNet. If I want to run, for instance, "real -> sketch" in DomainNet, what I should do?
If I run 'python main.py --config configs/DomainNet/lads.yaml', why are there multiple TEXT_PROMPTS and NEUTRAL_TEXT_PROMPTS from multiple domains? From my understanding, NEUTRAL_TEXT_PROMPTS (source) should be 'real' and TEXT_PROMPTS (target) should be 'sketch'?
And in data_helpers.py:
` elif dataset_name == "DomainNet": ...
elif dataset_name == "DomainNetMini": ... `
May I ask the differences between "DomainNet" and "DomainNetMini"? Do we get a chance to use "DomainNet"? According to the paper, we seem to have only one source domain all the time, does it mean we actually will never set dataset_name == "DomainNet" in 'configs/DomainNet/lads.yaml'?
Many thanks!
When I tried to runexperiments with DomainNet dataset, I got the error:
DomainNet Dataset config is not found in data_configs/.
Can you upload the data config for domain net? Thanks