a tiny error in character_rnn example kurfile

[EmbraceLife]

Hi @noajshu , Thanks for your character example, it is great!

@ajsyp has pointed to me something is not exactly right with the weights, and now I realized the tiny issue there. Inside kurfile.yaml, is the following weights what you wanted, instead of here:

train:
  data:
    - jsonl: ../data/train.jsonl
  epochs: 5
  weights:
    initial: best.w.kur
    best: best.w.kur
    last: last.w.kur
  log: log

validate:
  data:
    - jsonl: ../data/validate.jsonl
  weights: best.w.kur

test:
  data:
    - jsonl: ../data/test.jsonl
  weights: best.w.kur

[ajsyp]

My recommendation is to do the following:

• Use initial and last keys under the training weights (but not best). The reason for this is that if best is specified under training weights, then Kur will save the best weights with respect to the training loss to the file at best. Best under validation tells Kur to save the best weights with respect to the validation loss to the file at best. Thus, using both puts that file in competition. So either drop best out of training weights entirely, or just change the name, something like best: best.train.w.kur.
• Use the best weights with respect to validation loss in the test and evaluate sections. This is what is shown in your examples, @EmbraceLife.

If you feel up to it, go ahead and submit a pull request for this!