Datasets:
Tasks:
Text Generation
Formats:
parquet
Sub-tasks:
language-modeling
Languages:
English
Size:
< 1K
License:
Update dataset (part 2)
Browse files
harbor/sources/druid-v9/tasks.py
CHANGED
|
@@ -122,6 +122,7 @@ async def _judge(name: str, weight: float, diff: str) -> SubScore:
|
|
| 122 |
weight, name=name, answer=diff[:100_000],
|
| 123 |
criteria=[tuple(c) if isinstance(c, list) else c for c in JUDGE["criteria"]],
|
| 124 |
question=JUDGE["question"],
|
|
|
|
| 125 |
)
|
| 126 |
except Exception as exc: # noqa: BLE001 - a judge outage must not kill grading
|
| 127 |
# Keep the judge's weight on a failure: a zero-weight subscore makes combine() renormalise the other
|
|
|
|
| 122 |
weight, name=name, answer=diff[:100_000],
|
| 123 |
criteria=[tuple(c) if isinstance(c, list) else c for c in JUDGE["criteria"]],
|
| 124 |
question=JUDGE["question"],
|
| 125 |
+
model=os.environ.get("JUDGE_MODEL", "claude-sonnet-5-5"),
|
| 126 |
)
|
| 127 |
except Exception as exc: # noqa: BLE001 - a judge outage must not kill grading
|
| 128 |
# Keep the judge's weight on a failure: a zero-weight subscore makes combine() renormalise the other
|