Not converge when training? about pyfm

[coreylynch]

This should be fixed now. Thanks.

[geekan]

@coreylynch Thanks for your work! But I test it for 300 iters and it cannot converge too, shouldn't I set a big iteration limit?

[coreylynch]

Hey @geekan I believe the learning rate is just too high in the example (I didn't do any tuning or grid search to find the most appropriate one). If you take a look at this: https://gist.github.com/coreylynch/7a33ba902b2e14315388, decreasing the learning rate to 0.001 allows the model to train for 600 iterations without diverging. Thanks!