Spaces:
Sleeping
Sleeping
Sami Ali
commited on
Commit
·
c48531b
1
Parent(s):
b098829
fix dir issue
Browse files- src/download_data.py +6 -3
- src/embedding.py +1 -1
src/download_data.py
CHANGED
|
@@ -17,11 +17,14 @@ def download_pmc_docs(
|
|
| 17 |
target_dir=TARGET_DIR,
|
| 18 |
limit=1000
|
| 19 |
):
|
| 20 |
-
|
| 21 |
-
return
|
| 22 |
-
|
| 23 |
os.makedirs(target_dir, exist_ok=True)
|
| 24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
s3 = boto3.client("s3", config=Config(signature_version=UNSIGNED))
|
| 26 |
paginator = s3.get_paginator('list_objects_v2')
|
| 27 |
|
|
|
|
| 17 |
target_dir=TARGET_DIR,
|
| 18 |
limit=1000
|
| 19 |
):
|
| 20 |
+
|
|
|
|
|
|
|
| 21 |
os.makedirs(target_dir, exist_ok=True)
|
| 22 |
|
| 23 |
+
existing_files = os.listdir(target_dir)
|
| 24 |
+
if len(existing_files) >= limit:
|
| 25 |
+
print(f"✅ Found {len(existing_files)} existing files. Skipping download.")
|
| 26 |
+
return True
|
| 27 |
+
|
| 28 |
s3 = boto3.client("s3", config=Config(signature_version=UNSIGNED))
|
| 29 |
paginator = s3.get_paginator('list_objects_v2')
|
| 30 |
|
src/embedding.py
CHANGED
|
@@ -14,7 +14,7 @@ class EmbeddingManager:
|
|
| 14 |
def load_model(self):
|
| 15 |
print("Loading embedding model:", self.model_name)
|
| 16 |
print('Using device', self.device)
|
| 17 |
-
self.model = SentenceTransformer(
|
| 18 |
print("Model loaded.")
|
| 19 |
|
| 20 |
def get_model(self):
|
|
|
|
| 14 |
def load_model(self):
|
| 15 |
print("Loading embedding model:", self.model_name)
|
| 16 |
print('Using device', self.device)
|
| 17 |
+
self.model = SentenceTransformer(model_name_or_path=self.model_name, device=self.device)
|
| 18 |
print("Model loaded.")
|
| 19 |
|
| 20 |
def get_model(self):
|