Buckets:

download
raw
1.94 kB
"""Helpers for loading/saving model_weights.yaml and performing single updates.
These helpers are intentionally small and file-system focused so they can be
unit-tested independently of Streamlit UI code.
"""
from pathlib import Path
import os
import yaml
from typing import Dict, Optional, Any
def load_model_weights(path: Path) -> Dict[str, float]:
"""Load model weights mapping from path. Returns empty dict on missing file.
Non-dict YAML content is coerced to empty dict.
"""
try:
if not path.exists():
return {}
with open(path, "r") as f:
data = yaml.safe_load(f) or {}
if not isinstance(data, dict):
return {}
# Coerce numeric values to floats where possible
out: Dict[str, float] = {}
for k, v in data.items():
try:
out[k] = float(v)
except Exception:
# skip invalid values
continue
return out
except Exception:
return {}
def save_model_weights(mapping: Dict[str, Any], path: Path) -> None:
"""Write mapping to YAML file, ensuring parent dir exists.
Values are not further validated here; callers should clamp/validate as
appropriate.
"""
os.makedirs(path.parent, exist_ok=True)
with open(path, "w") as f:
yaml.dump(mapping or {}, f, default_flow_style=False, sort_keys=False)
def set_model_weight(path: Path, doc_id: str, new_value: Optional[float]) -> Optional[float]:
"""Set a single doc's model weight. Returns previous value (or None).
If new_value is None, the doc_id is removed from the mapping if present.
"""
mapping = load_model_weights(path)
prev = mapping.get(doc_id)
if new_value is None:
if doc_id in mapping:
mapping.pop(doc_id)
else:
mapping[doc_id] = float(new_value)
save_model_weights(mapping, path)
return prev

Xet Storage Details

Size:
1.94 kB
·
Xet hash:
5e7ed8473e12d48bc0b79871cff4f8d4213d066ac558f6e77ebdda2de0fea2c1

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.