facebookresearch / StarSpace

Learning embeddings for classification, retrieval and ranking.
MIT License
3.95k stars 531 forks source link

Calculating time taken to predict #247

Closed tharangni closed 5 years ago

tharangni commented 5 years ago

I am doing a document classification task and my test set is quiet large: around 250k samples to be compared against 55k labels. Is there a way to calculate the time it takes to predict for these samples? I know it works fine since I can predict for 10k samples (even though it takes like 5-10 mins). Issue #68 was also referred and i changed the number of threads from 10 to 50, but it still takes time.

ledw commented 5 years ago

@tharangni Hi, the time it takes in test is proportional to the size of the examples, so it takes 25x more time when you increase examples from 10k to 250k.

tharangni commented 5 years ago

Awesome! Thanks for the info! Closing it .