coqa.py

# REMINDER: this code needs to be rewritten for the new framework. Remove this comment when the code is fully converted.

import json
import random
from lm_eval.base import Dataset
from ..utils import sh
import itertools

class CoQA(Dataset):
    def __init__(self):
        self.download()
    def download(self):
        sh ("""
            mkdir -p data/coqa 
            wget --no-clobber http://downloads.cs.stanford.edu/nlp/data/coqa/coqa-train-v1.0.json -O data/coqa/coqa-train-v1.0.json
            wget --no-clobber http://downloads.cs.stanford.edu/nlp/data/coqa/coqa-dev-v1.0.json -O data/coqa/coqa-dev-v1.0.json
            """)

    @classmethod
    def get_answers(cls, doc, turn_id):
        answers = zip(doc["answers"], zip(doc["additional_answers"]))
        return answers[turn_id - 1]

    def has_training_docs(self):
        return True

    def has_validation_docs(self):
        return True

    def has_test_docs(self):
        return False

    def training_docs(self):
        return json.load(open('data/coqa/coqa-train-v1.0.json'))['data']

    def validation_docs(self):
        return  json.load(open('data/coqa/coqa-dev-v1.0.json'))['data']  

    def test_docs(self):
        pass
    
    def fewshot_description(self):
        return "Given a passage and a conversation so far, answer the next question in the conversation."
    
    def doc_to_text(self, doc):
        qa_pairs = [(q, a) in zip_longest(doc["questions"], doc["answers"][:-1])]  # truncate target answer
        return "{}\n\n{}".format(doc["story"], f"Q: {q}"+ '\n\n' + f"A: {a}")
    
    def doc_to_target(self, doc):
        # TODO: all distinct answers taking into account whitespace?
        return get_answers(doc, len(doc["questions"]))

    def construct_requests(self, doc, ctx):
        """ Uses RequestFactory to construct Requests and returns an iterable of 
        Requests which will be sent to the LM.

        :param doc:
            The document as returned from training_docs, validation_docs, or test_docs.
        :param ctx: str
            The context string, generated by fewshot_context. This includes the natural 
            language description, as well as the few shot examples, and the question
            part of the document for `doc`. 
        """
        ll_alternative_answers = [
            rf.loglikelihood(ctx, " " + answer) for answer in get_answers(doc, len(doc["questions"]))
        ]

        return ll_alternative_answers
    
    def process_results(self, doc, results):
        """Take a single document and the LM results and evaluates, returning a 
        dict where keys are the names of submetrics and values are the values of 
        the metric for that one document

        :param doc:
            The document as returned from training_docs, validation_docs, or test_docs.
        :param results:
            The results of the requests created in construct_requests.
        """
        golds = get_answers(doc, len(doc["questions"]))
        pred = np.argmax(results)
        return {
            "acc": pred in golds,
            # "f1": (golds, pred),    # TODO:  Fix
        }

    def aggregation(self):
        """
        :returns: {str: [float] -> float}
            A dictionary where keys are the names of submetrics and values are 
            functions that aggregate a list of metrics
        """
        # TODO: implement evaluation.
        raise NotImplementedError('Evaluation not implemented')

    def higher_is_better(self):
        """
        :returns: {str: bool}
            A dictionary where keys are the names of submetrics and values are 
            whether a higher value of the submetric is better
        """
        # TODO: implement evaluation.
        raise NotImplementedError('Evaluation not implemented')