Synthetic Instructions (LLM Judged)
Generated with Distilabel-style pipeline, curated with LLM Judge.
Total Generated: 25
Passed Judge (>= 3/5): 25
Generator: Qwen/Qwen2.5-3B-Instruct
Judge: Qwen/Qwen2.5-3B-Instruct
Score 4: 10 samples
Score 3: 15 samples
instruction: The question/task
input: Additional context (empty)
output: The response
topic: Subject area
quality_score: LLM Judge rating (1-5)