Train the model on chinese OntoNotes 5.0 but eat up all my 64GB memory

[ITonly]

I am training it for Chinese Model to use for event extraction for my company, but I only use part of English data for testing the model, as show in first image(train data ). But it eats up 50% memory(64GB) when build coreference clusters，when I use all data from conll-2012, my memery is 100% and can't run any more. Can you tell me how much memory should I use to train the model, or any other requirements? The attach file is my training machine performance.

[svlandeg]

Hi @ITonly, can you give a bit more detail on which code you were executing exactly, and where you run into memory issues?

Unfortunately I personally don't have experience with retraining this for Chinese.
Have you been able to solve it in the meantime with more RAM?

[svlandeg]

Also - this is probably a stab in the dark - but just to make sure: are you running the code from a console directly or from within an IDE? (the latter may be restricting your memory usage)

[svlandeg]

One more suggestion: could you try setting use_gold_mentions=True (see related Issue #212)

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[Abhijit-2592]

@ITonly Check my pull request #231. This problem is now resolved. Feel free to get back if you still find memory issues