clab / clab/dynet

Example missing info

Open
#1,157 1 comment 0 reactions 0 assignees View on GitHub
minor bug
Dominant language
C++
Stars
3.4k
Forks
701
PR merge metrics
No merged PRs in 30d

Description

Regarding this example:
https://github.com/clab/dynet/blob/master/examples/tagger/bilstmtagger.py

It is missing:
```python
import numpy as np
```

and it imports "util" which is not a module. The module can be found here: https://github.com/clab/dynet/blob/master/examples/python-utils/util.py

The train and test files are local for @yoavg 's machine:
```python
train_file="/Users/yogo/Vork/Research/corpora/pos/WSJ.TRAIN"
test_file="/Users/yogo/Vork/Research/corpora/pos/WSJ.TEST"
```

And I tried to find them, and couldn't.
I did find some good train and test files in a homework assignment he gave.

Also, I am literally a 30 minutes DyNet user, so excuse me if I'm wrong, but [this line](https://github.com/clab/dynet/blob/master/examples/tagger/bilstmtagger.py#L41) seems like a horrible practice. to add the test words in the words index, so you should in theory never hit the _UNK_ word.. But I might misunderstood what is it doing there.

Finally, the test file is not really a test file, it is a tagged dev file.

Other than those, this example works great for POS and NER data (loss is dropping, accuracy is increasing, and homework are getting done)

Contributor guide

No contributing guide indexed for this repository

Research direction

Start with examples/tagger/bilstmtagger.py and compare its imports with examples/python-utils/util.py. Check the hard-coded train and test paths, the vocabulary handling around line 41, and whether the referenced test data is actually development data. Done means the example's dependencies and data references are usable and its POS/NER behavior is clearly documented or corrected.

Written by the indexing model from the issue text.

Assessment

Tech stack
numpy, python
Domain
machine-learning
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.