Benchmark

Constructor

Benchmark(model, tokenizer[, task_name, ...])

Generic interface to compute multiple explanations.

Explaining

Benchmark.explain(text[, target, ...])

Compute explanations using all the explainers stored in the class.

Benchmarking Explanations

Benchmark.evaluate_explanation(explanation)	Evaluate an explanation using all the evaluators stored in the class.
Benchmark.evaluate_explanations(explanations)	Evaluate explanations using all the evaluators stored in the class.

Visualization

Benchmark.show_table(explanations[, ...])
Benchmark.show_evaluation_table(...[, style])
Benchmark.show_samples_evaluation_table(...)	Format average evaluation scores into a colored table.

Datasets Interface

Benchmark.load_dataset(dataset_name, **kwargs)
Benchmark.evaluate_samples(dataset, sample)	Explain a dataset sample, evaluate explanations, and compute average scores.

Inference

Benchmark.score(text[, return_dict])

Compute prediction scores for a single query