Open james-tanner opened 5 years ago
@mmcauliffe any thoughts on this? I know you probably won't have time to fix before leaving, but any guidance appreciated. like, do you suspect the issue will have been resolved with your recent memory optimizations -- or does the issue seem like an actual bug?
I've been trying to use the
encode_baseline
measure for words inside of a SPADE script, currently:This works fine on smaller corpora (like ICE-Can or Modern RP), but exceeds the memory limit (even on Roquefort) for corpora of SOTC-size and larger.