11_raafi.py

import os
import wget
import zipfile

model_name = "model_raafi"
if not os.path.exists(f"{model_name}.zip"):
    url = f"https://harmonyapistorage.z33.web.core.windows.net/{model_name}.zip"
    print (f"Downloading model from {url}")
    path_to_zip_file = wget.download(url)
else:
    path_to_zip_file = f"{model_name}.zip"

if not os.path.isdir(model_name):
    with zipfile.ZipFile(path_to_zip_file, 'r') as zip_ref:
        zip_ref.extractall(model_name)


os.environ["HARMONY_SENTENCE_TRANSFORMER_PATH"] = model_name


import re

import harmony
import numpy as np
from harmony.schemas.requests.text import Instrument, Question

import evaluation_helper



re_tokeniser = re.compile(r'([a-z]+)')

for input_file, data in evaluation_helper.get_datasets():
    all_questions = list(sorted(set(data.text_1).union(set(data.text_2))))
    question_text_to_idx = dict([b, a] for a, b in enumerate(all_questions))
    questions = []
    for idx, question_text in enumerate(all_questions):
        questions.append(Question(question_text=question_text, question_no=f"{idx}"))
    instrument = Instrument(questions=questions)
    match_response = harmony.match_instruments([instrument])
    similarity = match_response.similarity_with_polarity
    preds = [0] * len(data)
    for idx in range(len(data)):
        text_1 = data.text_1.iloc[idx]
        text_2 = data.text_2.iloc[idx]
        idx_1 = question_text_to_idx[text_1]
        idx_2 = question_text_to_idx[text_2]
        preds[idx] = np.abs(similarity[idx_1, idx_2])

    data["y_pred"] = preds

    evaluation_helper.save_results(input_file, data)