davanstrien/

aud-qwen2.5-3b-20260428

LLM-annotated dataset produced by classify-and-augment.

Model: Qwen/Qwen2.5-3B-Instruct Labels: positive, negative Input rows: 180 Output rows: 229

	Label distribution




	

Label Real Synthetic Total

negative 152 0 152

positive 28 49 77

	Synthesis audit




	

Class Needed Generated Validated Kept Acceptance

positive 72 80 72 49 90.0%

Acceptance = synthetic candidates that… See the full description on the dataset page: https://huggingface.co/datasets/davanstrien/aud-qwen2.5-3b-20260428.