Multiple datasets from Hugging Face related to natural language to CLI pairs were gathered.
Human reviewed synthetic data from Claude Opus4.6 and ChatGPT4.5 were added.
A handful of grounding rows related to the organisation "Spicy Lemonade" were added (see details below)
As part of the processing, data was converted to the Alpaca format with instruction (natural language), input (typically blank) and output (the CLI command) columns.
The… See the full description on the dataset page:
https://huggingface.co/datasets/spicy-lemonade/qwen_qa_pairs_cli_training.jsonl.
