A few-shot training dataset containing structured feedback evaluations in XML format, designed to train language models in providing detailed, rubric-based content evaluation and constructive feedback.
The dataset presents examples in a structured XML format where each instance contains multiple evaluation blocks. Each example follows a consistent pattern:
A global instruction for content evaluation
Multiple… See the full description on the dataset page:
https://huggingface.co/datasets/forcemultiplier/feedback.
