File size: 3,074 Bytes
0b32ad6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 |
import tempfile
from pathlib import Path
import pytest
import yaml
from dotenv import dotenv_values
from torch.utils.data import Subset
from tqdm import tqdm
from s3prl.downstream.emotion.expert import DownstreamExpert
from s3prl.problem import SuperbER
@pytest.mark.corpus
@pytest.mark.parametrize("fold_id", [0, 1, 2, 3, 4])
def test_er_dataset(fold_id):
v3_er_folder = Path(__file__).parent.parent / "s3prl" / "downstream" / "emotion"
IEMOCAP = dotenv_values()["IEMOCAP"]
with (v3_er_folder / "config.yaml").open() as file:
config = yaml.load(file, Loader=yaml.FullLoader)["downstream_expert"]
config["datarc"]["root"] = IEMOCAP
config["datarc"]["meta_data"] = v3_er_folder / "meta_data"
config["datarc"]["test_fold"] = f"fold{fold_id + 1}"
with tempfile.TemporaryDirectory() as tempdir:
expert = DownstreamExpert(320, config, tempdir)
train_dataset_v3 = expert.get_dataloader("train").dataset
valid_dataset_v3 = expert.get_dataloader("dev").dataset
test_dataset_v3 = expert.get_dataloader("test").dataset
with tempfile.TemporaryDirectory() as tempdir:
default_config = SuperbER().default_config()
train_csv, valid_csv, test_csvs = SuperbER().prepare_data(
{"iemocap": IEMOCAP, "test_fold": fold_id}, tempdir, tempdir
)
encoder_path = SuperbER().build_encoder(
default_config["build_encoder"],
tempdir,
tempdir,
train_csv,
valid_csv,
test_csvs,
)
train_dataset_v4 = SuperbER().build_dataset(
default_config["build_dataset"],
tempdir,
tempdir,
"train",
train_csv,
encoder_path,
None,
)
valid_dataset_v4 = SuperbER().build_dataset(
default_config["build_dataset"],
tempdir,
tempdir,
"valid",
valid_csv,
encoder_path,
None,
)
test_dataset_v4 = SuperbER().build_dataset(
default_config["build_dataset"],
tempdir,
tempdir,
"test",
test_csvs[0],
encoder_path,
None,
)
def compare_dataset(v3, v4):
data_v3 = {}
for wav, label, name in tqdm(v3, desc="v3"):
if isinstance(v3, Subset):
v3 = v3.dataset
label_name = [k for k, v in v3.class_dict.items() if v == label][0]
data_v3[name] = label_name
data_v4 = {}
for batch in tqdm(v4, desc="v4"):
data_v4[batch["unique_name"]] = batch["label"]
assert sorted(data_v3.keys()) == sorted(data_v4.keys())
for key in data_v3:
value_v3 = data_v3[key]
value_v4 = data_v4[key]
assert value_v3 == value_v4
compare_dataset(train_dataset_v3, train_dataset_v4)
compare_dataset(valid_dataset_v3, valid_dataset_v4)
compare_dataset(test_dataset_v3, test_dataset_v4)
|