This dataset contains task-level metadata annotations for large language model (LLM) requests.
Each example includes a user prompt and structured metadata describing the task characteristics.
instruction: a system-level instruction specifying the annotation format
input: the user prompt
output: a JSON object with the following fields:
task_type: categorical task label (e.g., QA, CodeGeneration… See the full description on the dataset page:
https://huggingface.co/datasets/dadadada1/task-metadata-slo.
