Closed pilchat closed 3 years ago
Thanks for pinging me about this again! I'm not sure how much help I can be with this issue, since it seems to be on the MAST server side. I've had trouble in the past downloading a lot of light curves at once, but usually if I try again a few hours or another day, it works just fine.
Alternatively, if you're not hoping to retrain all of the CNN models from scratch, I just wrote a new function to download the models, which are now hosted on MAST. This downloads as a .tar.gz
file, so is much faster! You can get the models through stella
by calling: ds = stella.DownloadSets.download_models()
This will set the attributes
ds.model_dir
= where the files were downloadedds.models
= the path + filenames of the modelsBy default, ds.models
will be the 10 models used in the analysis of our paper. If you want access to all 100 models, you can simply set ds = stella.DownloadSets.download_models(all_models=True)
. I will work on incorporating this into the docs over the next few days, but let me know if you have any questions until then!
This new function is available on Github, but not PyPI yet.
hi,
I think I will wait for the update of the docs: can you notify me when you will be done?
Moreover, what should I do to get the new functionality? I tried pip upgrade but no newer version of stella has been installed. Maybe I have to clone the git repo?
thanks a lot
On Tue, Oct 27, 2020 at 6:06 PM Adina Feinstein notifications@github.com wrote:
Thanks for pinging me about this again! I'm not sure how much help I can be with this issue, since it seems to be on the MAST server side. I've had trouble in the past downloading a lot of light curves at once, but usually if I try again a few hours or another day, it works just fine.
Alternatively, if you're not hoping to retrain all of the CNN models from scratch, I just wrote a new function to download the models, which are now hosted on MAST. This downloads as a .tar.gz file, so is much faster! You can get the models through stella by calling: ds = stella.DownloadSets.download_models()
This will set the attributes
- ds.model_dir = where the files were downloaded
- ds.models = the path + filenames of the models
By default, ds.models will be the 10 models used in the analysis of our paper. If you want access to all 100 models, you can simply set ds = stella.DownloadSets.download_models(all_models=True). I will work on incorporating this into the docs over the next few days, but let me know if you have any questions until then!
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/afeinstein20/stella/issues/11#issuecomment-717389319, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQ5BYZJRY4WYTNRTIOPASDSM3423ANCNFSM4S3EBDGA .
I can definitely notify you when that's up. And yes, you have to clone the git repo to get that function right now!
Hi Adina,
any updates?
thanks
I believe the docs have been properly updated now! Sorry for such a delay. Let me know if you are still having issues.
Hi,
I am using stella 0.1.0 installed via pip. I am trying to download the lightcurves with:
but it is extremely slow: it created ~70 .npy files after 3 hours though my internet connection is superfast. After that, it gave me this error:
is it ok that the process is this slow and eventually gets killed by the server?
thanks