Skip to content

euroeval.benchmark_config_factory

source module euroeval.benchmark_config_factory

Factory class for creating dataset configurations.

Functions

source build_benchmark_config(progress_bar: bool, save_results: bool, task: str | list[str] | None, dataset: str | list[str] | None, language: str | list[str], model_language: str | list[str] | None, dataset_language: str | list[str] | None, device: Device | None, batch_size: int, raise_errors: bool, cache_dir: str, api_key: str | None, force: bool, verbose: bool, trust_remote_code: bool, use_flash_attention: bool | None, clear_model_cache: bool, evaluate_test_split: bool, few_shot: bool, num_iterations: int, api_base: str | None, api_version: str | None, debug: bool, run_with_cli: bool, only_allow_safetensors: bool, first_time: bool = False) → BenchmarkConfig

Create a benchmark configuration.

Parameters

  • progress_bar : bool — Whether to show a progress bar when running the benchmark.

  • save_results : bool — Whether to save the benchmark results to a file.

  • task : str | list[str] | None — The tasks to include for dataset. If None then datasets will not be filtered based on their task.

  • dataset : str | list[str] | None — The datasets to include for task. If None then all datasets will be included, limited by the task parameter.

  • language : str | list[str] — The language codes of the languages to include, both for models and datasets. Here 'no' means both Bokmål (nb) and Nynorsk (nn). Set this to 'all' if all languages should be considered.

  • model_language : str | list[str] | None — The language codes of the languages to include for models. If None then the language parameter will be used.

  • dataset_language : str | list[str] | None — The language codes of the languages to include for datasets. If None then the language parameter will be used.

  • device : Device | None — The device to use for running the models. If None then the device will be set automatically.

  • batch_size : int — The batch size to use for running the models.

  • raise_errors : bool — Whether to raise errors when running the benchmark.

  • cache_dir : str — The directory to use for caching the models.

  • api_key : str | None — The API key to use for a given inference server.

  • force : bool — Whether to force the benchmark to run even if the results are already cached.

  • verbose : bool — Whether to print verbose output when running the benchmark. This is automatically set if debug is True.

  • trust_remote_code : bool — Whether to trust remote code when running the benchmark.

  • use_flash_attention : bool | None — Whether to use Flash Attention for the models. If None then it will be used if it is available.

  • clear_model_cache : bool — Whether to clear the model cache before running the benchmark.

  • evaluate_test_split : bool — Whether to use the test split for the datasets.

  • few_shot : bool — Whether to use few-shot learning for the models.

  • num_iterations : int — The number of iterations each model should be evaluated for.

  • api_base : str | None — The base URL for a given inference API. Only relevant if model refers to a model on an inference API.

  • api_version : str | None — The version of the API to use for a given inference API.

  • debug : bool — Whether to run the benchmark in debug mode.

  • run_with_cli : bool — Whether the benchmark is being run with the CLI.

  • only_allow_safetensors : bool — Whether to only allow evaluations of models stored as safetensors.

  • first_time : bool — Whether this is the first time the benchmark configuration is being created. Defaults to False.

Returns

source get_correct_language_codes(language_codes: str | list[str]) → list[str]

Get correct language code(s).

Parameters

  • language_codes : str | list[str] — The language codes of the languages to include, both for models and datasets. Here 'no' means both Bokmål (nb) and Nynorsk (nn). Set this to 'all' if all languages should be considered.

Returns

  • list[str] — The correct language codes.

source prepare_languages(language_codes: str | list[str] | None, default_language_codes: list[str]) → list['Language']

Prepare language(s) for benchmarking.

Parameters

  • language_codes : str | list[str] | None — The language codes of the languages to include for models or datasets. If specified then this overrides the language parameter for model or dataset languages.

  • default_language_codes : list[str] — The default language codes of the languages to include.

Returns

  • list['Language'] — The prepared model or dataset languages.

source prepare_tasks_and_datasets(task: str | list[str] | None, dataset_languages: list['Language'], dataset: str | list[str] | None) → tuple[list['Task'], list[str]]

Prepare task(s) and dataset(s) for benchmarking.

Parameters

  • task : str | list[str] | None — The tasks to include for dataset. If None then datasets will not be filtered based on their task.

  • dataset_languages : list['Language'] — The languages of the datasets in the benchmark.

  • dataset : str | list[str] | None — The datasets to include for task. If None then all datasets will be included, limited by the task and dataset_languages parameters.

Returns

  • tuple[list['Task'], list[str]] — The prepared tasks and datasets.

Raises

  • InvalidBenchmark — If the task or dataset is not found in the benchmark tasks or datasets.

source prepare_device(device: Device | None) → torch.device

Prepare device for benchmarking.

Parameters

  • device : Device | None — The device to use for running the models. If None then the device will be set automatically.

Returns

  • torch.device — The prepared device.