mirror of
https://github.com/wassname/lora-lite.git
synced 2026-06-27 15:00:33 +08:00
wassname
2a50373311
CI collection failed with ModuleNotFoundError: No module named '_cost' because exec_module loads the script without its dir on sys.path. Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
39 lines
1.3 KiB
Python
39 lines
1.3 KiB
Python
import importlib.util
|
|
import sys
|
|
from pathlib import Path
|
|
|
|
|
|
SCRIPT_PATH = Path(__file__).parents[1] / "scripts" / "metamath_gsm8k_benchmark.py"
|
|
# the script uses sibling imports (`from _cost import ...`), so scripts/ must be importable
|
|
sys.path.insert(0, str(SCRIPT_PATH.parent))
|
|
SPEC = importlib.util.spec_from_file_location("metamath_gsm8k_benchmark", SCRIPT_PATH)
|
|
benchmark = importlib.util.module_from_spec(SPEC)
|
|
assert SPEC.loader is not None
|
|
sys.modules[SPEC.name] = benchmark
|
|
SPEC.loader.exec_module(benchmark)
|
|
|
|
extract_answer = benchmark.extract_answer
|
|
score_predictions = benchmark.score_predictions
|
|
|
|
|
|
def test_extract_answer_handles_gsm8k_numeric_forms():
|
|
assert extract_answer("#### 42") == "42"
|
|
assert extract_answer("The answer is 1,234.") == "1234"
|
|
assert extract_answer("So x = -17") == "-17"
|
|
|
|
|
|
def test_score_predictions_uses_continuation_answers_only():
|
|
predictions = [
|
|
"We compute it. The answer is 42.",
|
|
"No final number here",
|
|
"Prompt said #### 5. But the continuation answer is 6.",
|
|
]
|
|
references = [
|
|
"reasoning\n#### 42",
|
|
"reasoning\n#### 9",
|
|
"reasoning\n#### 5",
|
|
]
|
|
scored = score_predictions(predictions, references)
|
|
assert scored["correct"] == 1
|
|
assert scored["total"] == 3
|
|
assert scored["accuracy"] == 1 / 3 |