InfiniteBench_RetrieveKV2

NAME = InfiniteBench_RetrieveKV2
DATASET_PATH = xinrongzhang2022/InfiniteBench
SAMPLE_SPLIT = kv_retrieval
FEWSHOT_SPLIT = kv_retrieval
RESPONSE_TYPE = COMPLETION
METRICS = [AccuracyCompletion]
SUBJECTS = ['default']
LANGUAGE = <Language.ENG: 'English'>

• Module: eval_framework.tasks.benchmarks.infinitebench

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

• Link to dataset: https://huggingface.co/datasets/xinrongzhang2022/InfiniteBench

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 "InfiniteBench_RetrieveKV2".