RedHatAI/speculator_benchmarks

--- license: apache-2.0 --- This dataset contains dataset splits for evaluating speculative decoding algorithms on different tasks. File: 1. Coding: HumanEval.jsonl 2. Math: math_reasoning.jsonl 3. Question Answering: qa.jsonl 4. MT_bench: question.jsonl 5. Retrieval-Augmented Generation: rag.jsonl 6. Summarization: summarization.jsonl 7. Translation (German to English): translation.jsonl 8. Writing: writing.jsonl The data comes from two sources: 1. https://github.com/openai/human-eval (1). (The MIT License) 2. https://github.com/hemingkx/Spec-Bench (2-8) (Apache 2.0)