results is not in line with yours

[sue2xlh]

hi, i run the program according to the procedure, train, evaluate, tag... but the result is not in line with yours

[sue2xlh]

and i run the same sent "President Obama is speaking at the White House.", the results are different at each time, i couldn't know this is normal condition

[sue2xlh]

and why the eval results are different at each time

[sue2xlh]

and i also want to know if i use B I E O tag , the program is normal

[Hironsan]

Hi, @Sue52nlp

You can use BIEO tags. But you need to rewrite evaluation code, especially get_entities function because It is assumed that the function is used for BIO tag.

[sue2xlh]

hi,Hironsan Thank you for your reply, can you tell me why the eval results are different at each time

[lengocthuan]

How to run an example as mentioned? I do not know anything about it.

[lengocthuan]

Please help me,

[mvallet91]

I have a similar issue. I simply downloaded the files and did the proper setup to run on GPU. Didn't change any variables.

While running the training, -f1 starts at 86, and goes up to 91.88 after a few rounds. However, when evaluating I get the following results:

evaluator = anago.Evaluator(model_config, weights, save_path=SAVE_ROOT, preprocessor=p) evaluator.eval(x_test, y_test)

• f1: 3.09

tagger = anago.Tagger(model_config, weights, save_path=SAVE_ROOT, preprocessor=p) sent = 'President Obama is speaking at the White House.' tagger.tag(sent) [('President', 'O'), ('Obama', 'O'), ('is', 'O'), ('speaking', 'O'), ('at', 'O'), ('the', 'O'), ('White', 'O'), ('House.', 'O')]

Any ideas? Maybe I have an incomplete corpus or other errors, but I don't understand the difference in F score while training and then when evaluating.