Upload src/xscript/_yaml.py with huggingface_hub
Browse files- src/xscript/_yaml.py +35 -0
src/xscript/_yaml.py
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""YAML loading with a corrected float resolver.
|
| 2 |
+
|
| 3 |
+
PyYAML's default implicit float resolver requires a signed exponent, so plain
|
| 4 |
+
scientific notation like `1.0e15` or `2.0e9` is loaded as a *string*. Our
|
| 5 |
+
configs are full of token budgets in that form, so we install the fixed
|
| 6 |
+
resolver everywhere configs are read.
|
| 7 |
+
"""
|
| 8 |
+
import re
|
| 9 |
+
|
| 10 |
+
import yaml
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
class _Loader(yaml.SafeLoader):
|
| 14 |
+
pass
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
_Loader.add_implicit_resolver(
|
| 18 |
+
"tag:yaml.org,2002:float",
|
| 19 |
+
re.compile(r"""^(?:
|
| 20 |
+
[-+]?(?:[0-9][0-9_]*)\.[0-9_]*(?:[eE][-+]?[0-9]+)?
|
| 21 |
+
|[-+]?\.[0-9_]+(?:[eE][-+]?[0-9]+)?
|
| 22 |
+
|[-+]?[0-9][0-9_]*(?:[eE][-+]?[0-9]+)
|
| 23 |
+
|[-+]?[0-9][0-9_]*(?::[0-5]?[0-9])+\.[0-9_]*
|
| 24 |
+
|[-+]?\.(?:inf|Inf|INF)
|
| 25 |
+
|\.(?:nan|NaN|NAN))$""", re.X),
|
| 26 |
+
list("-+0123456789."))
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
def load(path) -> dict:
|
| 30 |
+
from pathlib import Path
|
| 31 |
+
return yaml.load(Path(path).read_text(), Loader=_Loader)
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
def loads(text: str) -> dict:
|
| 35 |
+
return yaml.load(text, Loader=_Loader)
|