Spaces:
Sleeping
Sleeping
Update utils.py
Browse files
utils.py
CHANGED
|
@@ -49,17 +49,17 @@ def update_db_hub(texts, topics, dates):
|
|
| 49 |
dataset = load_dataset(dataset_name, download_config=DownloadConfig(token=api_token))
|
| 50 |
# print("Dataset loaded successfully!", dataset)
|
| 51 |
# print(dataset)
|
| 52 |
-
deduplicated_new_dataset = remove_duplicates(
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
)
|
| 58 |
|
| 59 |
-
updated_dataset = concatenate_datasets([dataset['train'],
|
| 60 |
|
| 61 |
# updated_dataset = new_dataset
|
| 62 |
-
del dataset
|
| 63 |
except Exception as e:
|
| 64 |
updated_dataset = new_dataset
|
| 65 |
print(f"Failed to load dataset: {e}")
|
|
|
|
| 49 |
dataset = load_dataset(dataset_name, download_config=DownloadConfig(token=api_token))
|
| 50 |
# print("Dataset loaded successfully!", dataset)
|
| 51 |
# print(dataset)
|
| 52 |
+
# deduplicated_new_dataset = remove_duplicates(
|
| 53 |
+
# dataset['train'],
|
| 54 |
+
# new_dataset,
|
| 55 |
+
# unique_key="text",
|
| 56 |
+
# batch_size=1000 # Adjust batch size based on available memory
|
| 57 |
+
# )
|
| 58 |
|
| 59 |
+
updated_dataset = concatenate_datasets([dataset['train'], new_dataset])
|
| 60 |
|
| 61 |
# updated_dataset = new_dataset
|
| 62 |
+
# del dataset
|
| 63 |
except Exception as e:
|
| 64 |
updated_dataset = new_dataset
|
| 65 |
print(f"Failed to load dataset: {e}")
|