Anisha Bhatnagar
commited on
Commit
·
8db24a7
1
Parent(s):
c0e59d3
updated file open mode in cache writing
Browse files- utils/llm_feat_utils.py +4 -2
utils/llm_feat_utils.py
CHANGED
|
@@ -21,7 +21,7 @@ def _feat_hash(feature: str, text: str) -> str:
|
|
| 21 |
blob = json.dumps({
|
| 22 |
"version": CACHE_VERSION,
|
| 23 |
"text": text,
|
| 24 |
-
"features":
|
| 25 |
}, sort_keys=True).encode()
|
| 26 |
return hashlib.md5(blob).hexdigest()
|
| 27 |
|
|
@@ -105,8 +105,10 @@ def generate_feature_spans_cached(client, text: str, features: list[str], role:
|
|
| 105 |
|
| 106 |
h = _feat_hash(feat, text)
|
| 107 |
if h in cache:
|
|
|
|
| 108 |
result[feat] = cache[h]["spans"]
|
| 109 |
else:
|
|
|
|
| 110 |
missing_feats.append(feat)
|
| 111 |
|
| 112 |
if missing_feats:
|
|
@@ -123,7 +125,7 @@ def generate_feature_spans_cached(client, text: str, features: list[str], role:
|
|
| 123 |
result[feat] = spans
|
| 124 |
|
| 125 |
# 5) write back the combined cache
|
| 126 |
-
with open(cache_path, "
|
| 127 |
json.dump(cache, f, indent=2)
|
| 128 |
return result
|
| 129 |
|
|
|
|
| 21 |
blob = json.dumps({
|
| 22 |
"version": CACHE_VERSION,
|
| 23 |
"text": text,
|
| 24 |
+
"features": feature
|
| 25 |
}, sort_keys=True).encode()
|
| 26 |
return hashlib.md5(blob).hexdigest()
|
| 27 |
|
|
|
|
| 105 |
|
| 106 |
h = _feat_hash(feat, text)
|
| 107 |
if h in cache:
|
| 108 |
+
print(f"Found feature: {feat}")
|
| 109 |
result[feat] = cache[h]["spans"]
|
| 110 |
else:
|
| 111 |
+
print(f"Missing feature: {feat}")
|
| 112 |
missing_feats.append(feat)
|
| 113 |
|
| 114 |
if missing_feats:
|
|
|
|
| 125 |
result[feat] = spans
|
| 126 |
|
| 127 |
# 5) write back the combined cache
|
| 128 |
+
with open(cache_path, "a") as f:
|
| 129 |
json.dump(cache, f, indent=2)
|
| 130 |
return result
|
| 131 |
|