datasets/dataset_stats.json

{
  "generated_at": "2026-01-04T01:00:39.599818",
  "source_counts": {
    "thinking_mode": 10000,
    "tool_calling": 8000,
    "qa_pairs": 34446,
    "uk_benchmark": 63,
    "bsi_benchmark": 40,
    "mcp_generated": 3690
  },
  "total_before_processing": 56239,
  "valid_examples": 56239,
  "unique_examples": 44022,
  "balanced_examples": 28824,
  "train_examples": 27383,
  "val_examples": 1441,
  "mcp_servers_covered": 42
}