ARC_EU20_DE

NAME = ARC_EU20_DE
DATASET_PATH = openGPT-X/arcx
SAMPLE_SPLIT = test
FEWSHOT_SPLIT = train
RESPONSE_TYPE = LOGLIKELIHOODS
METRICS = [AccuracyLoglikelihood, AccuracyNormLoglikelihood]
SUBJECTS = ['challenge_DE', 'easy_DE']
LANGUAGE = <Language.DEU: 'German'>

• Module: eval_framework.tasks.benchmarks.opengptx_eu20

• File: src/eval_framework/tasks/benchmarks/opengptx_eu20.py | View on GitHub

• Link to dataset: https://huggingface.co/datasets/openGPT-X/arcx

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 "ARC_EU20_DE".