compiled/manifest.json

{
  "duplicate_content_count": 0,
  "duplicate_key_count": 56656,
  "input_root": "/data/raw",
  "instruction_path": "/data/compiled/instruction.jsonl",
  "manifest_path": "/data/compiled/manifest.json",
  "max_records": null,
  "output_dir": "/data/compiled",
  "record_count": 620995,
  "records_path": "/data/compiled/records.jsonl",
  "skipped_invalid_count": 0,
  "source_record_count": 677651,
  "tokenizer_map_path": "/data/compiled/tokenizer_map.json",
  "train_text_path": "/data/compiled/train.txt"
}