musashihinck/

IssueBench_Completions

For the IssueBench dataset, see paul/IssueBench. This dataset provides the completions and annotations used in IssueBench: Millions of Realistic Prompts for Measuring Issue Bias in LLM Writing Assistance.

This dataset is organized into eight splits, one for each of the eight models used to generate completions:

GPT-4o-mini meta-llama/Meta-Llama-3.1-8B-Instruct meta-llama/Meta-Llama-3.1-70B-Instruct Qwen/Qwen2.5-7B-Instruct… See the full description on the dataset page: https://huggingface.co/datasets/musashihinck/IssueBench_Completions.