Named Entity Recognition #167
Conversation
src/sequence/ner_datadeps.jl
Outdated
| """ | ||
| The weights for NER Sequence Labelling Model. | ||
| """, | ||
| "https://raw.githubusercontent.com/Ayushk4/Random_set_of_codes/weights/weights.tar.xz", |
There was a problem hiding this comment.
I don't get what you mean by work on windows?
There was a problem hiding this comment.
If you mean how to extract/unzip the file, I think WinRar should let you extract it.
Or if you are referring to the file being unpacked, then DataDeps package function unpack handles it.
There was a problem hiding this comment.
I meant will DataDeps be able to unzip the file without the user having to install additional things?
There was a problem hiding this comment.
It should work - DataDeps.jl has a test for checking the unpack for tarballs. In this line - https://github.com/oxinabox/DataDeps.jl/blob/12bf837dc4ba09edca19fe617d1408ec7dae1796/test/examples.jl#L137
There was a problem hiding this comment.
I added Windows to the os in travis CI and sent a PR to my own fork of DataDeps.jl. The tests were ran on linux, mac osz and windows and passed on all the os - Ayushk4/DataDeps.jl#1
|
the weights are now in the repo as release assets: |
|
@aviks can you review this Patch? I am proceeding with POS API, most of the functions for NER API will be reused on that. |
|
@aviks I have written the POS API here. Can you review this patch, since the POS API uses the same model and functions for sequence labelling? Also, that seems to be working much better than the perceptronTagger |
2 participants
Porting the NER API from https://github.com/Ayushk4/NER.jl/tree/master/src