Pyatkin et al. (2021)#

Publication#

Asking It All: Generating Contextualized Questions for any Semantic Role

Repositories#

https://github.com/ValentinaPy/RoleQGeneration

This implementation uses our fork with some additional code modifications: https://github.com/danieldeutsch/RoleQGeneration

Available Models#

RoleQuestionGenerator

Description: A model which generates role questions. This model is the one released by the authors.
Name: pyatkin2021-role-question-generator

Usage:

from repro.models.pyatkin2021 import RoleQuestionGenerator
model = RoleQuestionGenerator()
inputs = [
    {
        "sentence": "Tom brings the dog to the park.",
        "token_index": 1,
        "lemma": "bring",
        "pos": "v",
        "sense": 1
    }
]
outputs = model.predict_batch(inputs)

The input keys are the sentence, the token index of the predicate in the sentence, the lemma of the predicate in OntoNotes, the part-of-speech of the predicate, and the OntoNotes sense index of the predicate.

Implementation Notes#

Docker Information#

Image name: danieldeutsch/pyatkin2021:1.0
Build command:
```
repro setup pyatkin2021
```
Requires network: Yes, the code does a network call related to ensuring NLTK libraries are installed.

Testing#

repro setup pyatkin2021
pytest models/pyatkin2021/tests

Status#

[x] Regression unit tests pass
[ ] Correctness unit tests pass
[ ] Model runs on full test dataset
[ ] Predictions approximately replicate results reported in the paper
[ ] Predictions exactly replicate results reported in the paper