So should I always aim to set eta as low as possible for time/processor constraints?
James Chuter
1

The lower it gets the smaller the steps done at each boosting iteration.

It can get absurdly slow and you might get diminishing returns quickly when using a small eta. Ideally the eta should be small enough but not too small, in order to allow an acceptable convergence speed while maintaining a good enough performance.

Note that sometimes a too small learning rate prevents learning properly.

Cross-validation can help finding an estimation of the appropriate number of rounds automatically.