train

#!/usr/bin/env python3
import argparse
import json
import subprocess
from splittools import get_topics_by_ids, read_topics, read_ids
import tune

parser = argparse.ArgumentParser()
parser.add_argument("--json", type=json.loads, required=True, help="the json input")

args, unknown = parser.parse_known_args()

TEST_SPLIT_PATH = '/data/splits/test_split.txt'
VALIDATION_SPLIT_PATH = '/data/splits/validation_split.txt'

QRELS_PATH = args.json["qrels"]["path"]
TOPICS_PATH = args.json["topic"]["path"]
TOPIC_FORMAT = args.json["topic"]["format"]
COLLECTION = args.json["collection"]["name"]
model_folder = args.json["model_folder"]["path"]


if __name__ == "__main__":
    dev_ids = read_ids(VALIDATION_SPLIT_PATH)
    test_ids = read_ids(TEST_SPLIT_PATH)
    topics = read_topics(TOPICS_PATH)

    dev_topics_path = "/tmp/dev_topics"
    test_topics_path = "/tmp/test_topics"

    get_topics_by_ids(topics, dev_ids, dev_topics_path)
    get_topics_by_ids(topics, test_ids, test_topics_path)

    tune.main(COLLECTION, dev_topics_path, "/tmp", QRELS_PATH)