evaluate.py

import copy
import numpy as np
from hparams import get_sampler_hparams
from utils import evaluate, get_sampler, load_model, EMA, log

if __name__ == '__main__':
    H = get_sampler_hparams('eval')

    if (H.gap_start < 0 or H.gap_end < 0) and H.mode == 'infilling':
        H.gap_start = H.NOTES // 4
        H.gap_end = (H.NOTES * 3) // 4
        log(f'Gap not specified - masking out {H.gap_start} to {H.gap_end}')

    if H.mode != 'self':
        sampler = get_sampler(H).cuda()
        sampler = load_model(sampler, H.sampler, H.load_step, H.load_dir).cuda()
        ema = EMA(H.ema_beta)
        ema_sampler = copy.deepcopy(sampler)

        try:
            ema_sampler = load_model(
                ema_sampler, f'{H.sampler}_ema', H.load_step, H.load_dir)
        except Exception as e:
            ema_sampler = copy.deepcopy(sampler)

    else:
        sampler, ema_sampler = None, None

    evaluate(H, ema_sampler if H.ema else sampler, True)