rambler-digital-solutions / criteo-1tb-benchmark

Benchmark of different ML algorithms on Criteo 1TB dataset
145 stars 20 forks source link

vowpal wabbit only run linearly #3

Open seanv507 opened 5 years ago

seanv507 commented 5 years ago

hi, its always hard to compare different algorithms. BUT IMO you should have laid out the configurations in the readme. in particular you are comparing xgboost max depth 7 with logistic regression with no interactions (ie just linear ? which corresponds to max depth 1? xgboost?? (also its a bit surbrising that you use 2^29 hash in vw and 10^5 in spark?)

Sandy4321 commented 4 years ago

do you have any news about this?