Add a simple testing method
RainBoltz opened this issue · comments
since we can train
then get the embeddings efficiently, why not also do testing
in the pipeline?
for example:
import pysmore.utils.eval as Evaluator
import pysmore.models.mf as MF
embeddings, index_mapper = MF.create_graph('/training/data')
MF.train()
Evaluator.create_graph('/testing/data')
result_dict = Evaluator.top_k_result(embeddings, index_mapper, k=10)
with result_dict
as:
{
'eval': 'top_k',
'settings':{
'k': int,
'task': 'i-i'
},
'results': {
'map': float,
'hit': float,
'rec': float
}
}
I can't find the testing method in the pipeline. Can you help me with it
sorry @neenerrh it's still in progress.
COVID19 messed everything up so this feature will not bring up recently 😞