Age | Commit message (Collapse) | Author |
|
|
|
require an additional module dependency
|
|
|
|
|
|
|
|
library dependency
|
|
|
|
* Add NER dataset for Polish
Co-authored-by: ryszardtuora <ryszardtuora@gmail.com>
Co-authored-by: Karol Saputa <ksaputa@gputrain.dariah.ipipan.waw.pl>
This PR adds Polish NER dataset
#1070
|
|
write sentences so that we can later manipulate the sentences as needed in write_sentences
|
|
|
|
|
|
|
|
conll18 version
|
|
|
|
|
|
fork of HTB. Addresses #1109
|
|
list of NER models in the package argument when creating a pipeline. From @mpenalver in #928
|
|
|
|
|
|
end of epoch save)
|
|
space and time
|
|
|
|
|
|
checkpoint knows when to save a better model
|
|
|
|
|
|
epoch will not be incremented properly when reloading)
|
|
pass checkpoint_file to train_model in the unittest, but TODO: need to add tests for checkpointing
|
|
loads the pretrain, charlm, etc
|
|
addition to the model, this saves and loads the optimizer and the number of epochs trained. Purpose: to make it so that it is easy to checkpoint model training the same way the charlm is checkpointed
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Includes a matrix trained to connect the 3 layers of elmo instead of using the default averaging
Also, a projection from elmo dim to a lower dimension (although this was less useful)
Add a comment on how the sentiment processor doesn't load Elmo.
Actually, in general this integration is unlikely to be used for much,
but there's also no specific reason to throw this code away.
|
|
|
|
sentiment yet
|
|
|
|
|
|
|
|
|
|
latest version
|
|
|
|
leaving this as an option.
|
|
using the dataset. Works better for smaller items... need to fix larger dataset as well
|
|
|
|
|