Updated dataset
Browse files
app.py
CHANGED
@@ -9,10 +9,10 @@ from difflib import ndiff
|
|
9 |
model = StaticModel.from_pretrained("minishlab/M2V_base_output")
|
10 |
|
11 |
# Default parameters
|
12 |
-
default_dataset_name = "
|
13 |
default_dataset1_split = "train" # Default for the first dataset is "train"
|
14 |
default_dataset2_split = "test" # Default for the second dataset is "test"
|
15 |
-
default_text_column = "
|
16 |
default_threshold = 0.9
|
17 |
|
18 |
def deduplicate_embeddings(
|
|
|
9 |
model = StaticModel.from_pretrained("minishlab/M2V_base_output")
|
10 |
|
11 |
# Default parameters
|
12 |
+
default_dataset_name = "ag_news"
|
13 |
default_dataset1_split = "train" # Default for the first dataset is "train"
|
14 |
default_dataset2_split = "test" # Default for the second dataset is "test"
|
15 |
+
default_text_column = "text"
|
16 |
default_threshold = 0.9
|
17 |
|
18 |
def deduplicate_embeddings(
|