amaye15 commited on
Commit ·
b105566
1
Parent(s): 5150064
clear cache
Browse files
app.py
CHANGED
|
@@ -23,10 +23,6 @@ def clear_huggingface_cache():
|
|
| 23 |
print("Cache directory does not exist.")
|
| 24 |
|
| 25 |
|
| 26 |
-
# Example usage
|
| 27 |
-
clear_huggingface_cache()
|
| 28 |
-
|
| 29 |
-
|
| 30 |
# Disable caching globally for Hugging Face datasets
|
| 31 |
disable_caching()
|
| 32 |
|
|
@@ -55,13 +51,13 @@ def get_data():
|
|
| 55 |
Uses streaming to avoid loading the entire dataset into memory at once,
|
| 56 |
which is useful for handling large datasets.
|
| 57 |
"""
|
|
|
|
| 58 |
ds = load_dataset(
|
| 59 |
DS_NAME,
|
| 60 |
streaming=True,
|
| 61 |
)
|
| 62 |
for row in ds["train"]:
|
| 63 |
yield row
|
| 64 |
-
clear_huggingface_cache()
|
| 65 |
|
| 66 |
|
| 67 |
def process_and_push_data():
|
|
@@ -77,7 +73,6 @@ def process_and_push_data():
|
|
| 77 |
ds_processed.push_to_hub(TARGET_REPO)
|
| 78 |
|
| 79 |
logger.info("Data processed and pushed to the hub.")
|
| 80 |
-
clear_huggingface_cache()
|
| 81 |
|
| 82 |
|
| 83 |
# Initialize the WebhooksServer with Gradio interface (if needed)
|
|
|
|
| 23 |
print("Cache directory does not exist.")
|
| 24 |
|
| 25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
# Disable caching globally for Hugging Face datasets
|
| 27 |
disable_caching()
|
| 28 |
|
|
|
|
| 51 |
Uses streaming to avoid loading the entire dataset into memory at once,
|
| 52 |
which is useful for handling large datasets.
|
| 53 |
"""
|
| 54 |
+
clear_huggingface_cache()
|
| 55 |
ds = load_dataset(
|
| 56 |
DS_NAME,
|
| 57 |
streaming=True,
|
| 58 |
)
|
| 59 |
for row in ds["train"]:
|
| 60 |
yield row
|
|
|
|
| 61 |
|
| 62 |
|
| 63 |
def process_and_push_data():
|
|
|
|
| 73 |
ds_processed.push_to_hub(TARGET_REPO)
|
| 74 |
|
| 75 |
logger.info("Data processed and pushed to the hub.")
|
|
|
|
| 76 |
|
| 77 |
|
| 78 |
# Initialize the WebhooksServer with Gradio interface (if needed)
|