BigCodeBench_OLMES

NAME = BigCodeBench_OLMES
DATASET_PATH = bigcode/bigcodebench
SAMPLE_SPLIT = v0.1.2
FEWSHOT_SPLIT = v0.1.2
RESPONSE_TYPE = COMPLETION
METRICS = [CodeExecutionPassAtOneWithCodebench]
SUBJECTS = ['original', 'calibrated']
LANGUAGE = <Language.ENG: 'English'>

• Module: eval_framework.tasks.benchmarks.bigcodebench

• File: src/eval_framework/tasks/benchmarks/bigcodebench.py | View on GitHub

• Link to dataset: https://huggingface.co/datasets/bigcode/bigcodebench

More detailed documentation, with prompt examples and ground truth completions, can be generated with uv run -m eval_framework.utils.generate_task_docs --add-prompt-examples --only-tasks "BigCodeBench_OLMES".