Training table doubts

[alvaro.marlo (Álvaro Martínez López)]

Hi,

What exactly means each column of the training table?
E | # | LOSS TOK2VEC | LOSS NER | ENTS_F | ENTS_P | ENTS_R | SCORE
Is somewhere at the documentation? I don't see anything about it.

Also, I have trained a model that displays a 0 at every row of the training table at E column. Why is this happening? I have something wrong?

Thanks,
Alvaro

[ryanwesslen]

hi @alvaro.marlo!

Check out this previous post:

understanding the different terminology in the command line output of a training pipeline

Hi @nanyasrivastav! Thank you for your questions! These concepts are very important. I'm glad you've asked because it's hard to effectively use Prodigy without understanding them . Initial learn rate: The initial learning rate used. The learning rate is a tuning parameter in an optimization algorithm that determines the step size at each iteration while moving toward a minimum of a loss function. This is a default value but can be optimized for training. Optimizing the learning …

@alvaro.marlo

Also, I have trained a model that displays a 0 at every row of the training table at E column. Why is this happening? I have something wrong?

The "E" is epochs and it seems like your dataset is so large (240,000+ training), it can't make a full pass (epoch) on your entire dataset.

I found this spaCy discussion post on advice on training large NER datasets. They recommended a few suggestions like creating a labels file (more to speed up the process), modifying the learning rate (will need some experimentation), and reducing the evaluation set. If you have further questions related to training, I would recommend posting on the spaCy discussion as the spaCy core team answers those posts (and prodigy train is really just a wrapper for spacy train, so your questions are really spaCy questions).