this i'm less certain about, but would it be relevant to cross-reference the following?
when mentioning vandam-data -- the instructions to create the vandam dataset, and also instructions to adapt datasets to this format
in sampling, the issue where we laid out alternative sampling schemes
between sampling and preparing chunks for zooniverse, there is a conceptual distinction bet/ clips & chunks that is explained. In addition, should we point to the paper with Chiara, or perhaps steal the figure where this is explained and put it here? or put it in the "book" and point there? In the book, we do talk about these two alternative approaches (vet the clips before sharing, or use chunking)
matching classifications back & importing: perhaps to illustrate how things can be done differently, we can point to Cychosz et al's paper, where chunks are analyzed, versus Semenzin et al, where chunks are related back to segments, and then segments are analyzed?
The "going big" section is great!
Shall we also incorporate a link to zooniverse quota?
how about a link to a sample gitignore ? it feels almost obvious that one should look at the present repo, but perhaps say it, and link it?
cluster using several CPU cores: we have a chapter on this in the "book" (which isn't a great chapter, but at least it's a bit more info)
this i'm less certain about, but would it be relevant to cross-reference the following?
- when mentioning vandam-data -- the instructions to create the vandam dataset, and also instructions to adapt datasets to this format
- in sampling, the issue where we laid out alternative sampling schemes
- between sampling and preparing chunks for zooniverse, there is a conceptual distinction bet/ clips & chunks that is explained. In addition, should we point to the paper with Chiara, or perhaps steal the figure where this is explained and put it here? or put it in the "book" and point there? In the book, we do talk about these two alternative approaches (vet the clips before sharing, or use chunking)
- matching classifications back & importing: perhaps to illustrate how things can be done differently, we can point to Cychosz et al's paper, where chunks are analyzed, versus Semenzin et al, where chunks are related back to segments, and then segments are analyzed?
The "going big" section is great!
- Shall we also incorporate a link to zooniverse quota?
- how about a link to a sample gitignore ? it feels almost obvious that one should look at the present repo, but perhaps say it, and link it?
- cluster using several CPU cores: we have a chapter on this in the "book" (which isn't a great chapter, but at least it's a bit more info)
I think it should be good now -- except for the link towards the VanDam tutorial which people can already find from ChildProject's docs and I am going to add a link to the tutorial in our GIN version of VanDam daylong.
I think it should be good now -- except for the link towards the VanDam tutorial which people can already find from ChildProject's docs and I am going to add a link to the tutorial in our GIN version of VanDam daylong.
this i'm less certain about, but would it be relevant to cross-reference the following?
The "going big" section is great!
I think it should be good now -- except for the link towards the VanDam tutorial which people can already find from ChildProject's docs and I am going to add a link to the tutorial in our GIN version of VanDam daylong.