Upload emotions_dataset.py
Browse files- emotions_dataset.py +40 -3
emotions_dataset.py
CHANGED
|
@@ -10,6 +10,39 @@ _URLS = [
|
|
| 10 |
"https://storage.googleapis.com/gresearch/goemotions/data/full_dataset/goemotions_3.csv",
|
| 11 |
]
|
| 12 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
|
| 14 |
class EmotionsDatasetConfig(datasets.BuilderConfig):
|
| 15 |
|
|
@@ -23,7 +56,7 @@ class EmotionsDataset(datasets.GeneratorBasedBuilder):
|
|
| 23 |
BUILDER_CONFIGS = [
|
| 24 |
EmotionsDatasetConfig(
|
| 25 |
name="all",
|
| 26 |
-
label_classes=
|
| 27 |
features=["text", "label", "dataset"]
|
| 28 |
)
|
| 29 |
]
|
|
@@ -35,7 +68,7 @@ class EmotionsDataset(datasets.GeneratorBasedBuilder):
|
|
| 35 |
features=datasets.Features(
|
| 36 |
{
|
| 37 |
"id": datasets.Value("string"),
|
| 38 |
-
'label': ClassLabel(names=
|
| 39 |
'text': Value(dtype='string', id=None),
|
| 40 |
'dataset': Value(dtype='string', id=None)
|
| 41 |
}
|
|
@@ -49,5 +82,9 @@ class EmotionsDataset(datasets.GeneratorBasedBuilder):
|
|
| 49 |
def _generate_examples(self, filepaths):
|
| 50 |
for i, filepath in enumerate(filepaths):
|
| 51 |
df = pd.read_csv(filepath)
|
|
|
|
|
|
|
|
|
|
| 52 |
for row_idx, row in enumerate(df.iterrows()):
|
| 53 |
-
yield f"{i}_{row_idx}", row
|
|
|
|
|
|
| 10 |
"https://storage.googleapis.com/gresearch/goemotions/data/full_dataset/goemotions_3.csv",
|
| 11 |
]
|
| 12 |
|
| 13 |
+
_CLASS_NAMES = [
|
| 14 |
+
"no emotion",
|
| 15 |
+
"happiness",
|
| 16 |
+
"admiration",
|
| 17 |
+
"amusement",
|
| 18 |
+
"anger",
|
| 19 |
+
"annoyance",
|
| 20 |
+
"approval",
|
| 21 |
+
"caring",
|
| 22 |
+
"confusion",
|
| 23 |
+
"curiosity",
|
| 24 |
+
"desire",
|
| 25 |
+
"disappointment",
|
| 26 |
+
"disapproval",
|
| 27 |
+
"disgust",
|
| 28 |
+
"embarrassment",
|
| 29 |
+
"excitement",
|
| 30 |
+
"fear",
|
| 31 |
+
"gratitude",
|
| 32 |
+
"grief",
|
| 33 |
+
"joy",
|
| 34 |
+
"love",
|
| 35 |
+
"nervousness",
|
| 36 |
+
"optimism",
|
| 37 |
+
"pride",
|
| 38 |
+
"realization",
|
| 39 |
+
"relief",
|
| 40 |
+
"remorse",
|
| 41 |
+
"sadness",
|
| 42 |
+
"surprise",
|
| 43 |
+
"neutral",
|
| 44 |
+
]
|
| 45 |
+
|
| 46 |
|
| 47 |
class EmotionsDatasetConfig(datasets.BuilderConfig):
|
| 48 |
|
|
|
|
| 56 |
BUILDER_CONFIGS = [
|
| 57 |
EmotionsDatasetConfig(
|
| 58 |
name="all",
|
| 59 |
+
label_classes=_CLASS_NAMES,
|
| 60 |
features=["text", "label", "dataset"]
|
| 61 |
)
|
| 62 |
]
|
|
|
|
| 68 |
features=datasets.Features(
|
| 69 |
{
|
| 70 |
"id": datasets.Value("string"),
|
| 71 |
+
'label': ClassLabel(names=_CLASS_NAMES, id=None),
|
| 72 |
'text': Value(dtype='string', id=None),
|
| 73 |
'dataset': Value(dtype='string', id=None)
|
| 74 |
}
|
|
|
|
| 82 |
def _generate_examples(self, filepaths):
|
| 83 |
for i, filepath in enumerate(filepaths):
|
| 84 |
df = pd.read_csv(filepath)
|
| 85 |
+
current_classes = list(set(df.columns).intersection(set(_CLASS_NAMES)))
|
| 86 |
+
df = df[["text"] + current_classes]
|
| 87 |
+
df = df[df[current_classes].sum(axis=1) == 1].reset_index(drop=True)
|
| 88 |
for row_idx, row in enumerate(df.iterrows()):
|
| 89 |
+
yield f"{i}_{row_idx}", {"text": row["text"],
|
| 90 |
+
"label": row[current_classes][row == 1].index.item()}
|