Even if I set all the parameters the same, you get different results. How can I get the same result in both versions?
How do I get the same results in versions 0.9 and 1.3.3?
1 Like
You can’t. Different versions tend to use different implementations of algorithms.
Hello. I faced the same problem. I get dramatically different results with same parameters - 50 iterations instead of about 1000. With the help of the debugger, I found a line of code where the differences appeared, but I do not understand what affects it. I keep getting the same pred value of 0.5 all the time. What is the possible reason for this problem?? Maybe we missed something important that broke backward compatibility?
xgboost/core.py
