search

psorus / eereid

Python module providing tools for training and evaluating re-identification models.
Apache License 2.0
2 stars 0 forks source link

Default parameters not set #21

Closed fpolachowski closed 5 days ago

fpolachowski commented 5 days ago

Add default parameters

class ghost():
    def __init__(self,*tags,dataset=None, distance=None, loss=None, model=None, novelty=None, experiments=None,modifier=None,preproc=None,prepros=None,preprocessing=None, **kwargs):
        #add kwargs 
        self.dataset=ee.datasets.mnist()
        self.distance=ee.distances.euclidean()
        self.experiments={}
        self.loss=ee.losses.triplet()
        self.model=ee.models.conv()
        self.modifier=mods()
        self.novelty=None
        self.prepro={}
psorus commented 5 days ago

I dont get the problem tbh. the standarts are set (but not in the standard way). Like the loss is usually triplet. And if you provide a new loss it gets overwritten. Also I dont find the pull request anymore?

psorus commented 5 days ago

ahh the email has more info basically when you set another loss, the old one gets overwritten always. Same for the other unique variables

fpolachowski commented 5 days ago

I saw the overwrite functions at the bottom right after I created this issue. Hence closed it as already resolved.