data/final_data/longbench_blend_metadata.json

{
  "script_version": "qcfuse-longbench-v1-top20-200",
  "datasets": [
    "musique",
    "2wikimqa",
    "hotpotqa"
  ],
  "input_dir": "data/raw_longbench",
  "output_dir": "data/final_data",
  "tokenizer_path": "/public/models/Qwen3-8B",
  "embedding_model": "/public/models/embedding/bge-m3",
  "chunk_size": 512,
  "chunk_overlap": 50,
  "context_topk": 20,
  "max_samples_per_dataset": 200,
  "batch_size": 128,
  "device": "auto",
  "input_sha256": {
    "musique": "4ac69b91281c4ec6b21316cb7282e83fb6b4dda04fc68480bb8d8ed1e19ff7bd",
    "2wikimqa": "dda279cf93a99e1e5bfa3291fb199fd55978d10a1feb31822953cf77a1742e37",
    "hotpotqa": "a0005ab2a1bc2ac3a70352dccbf96cccc4e0aac6bb677f6a55180fa51b92ef6f"
  },
  "output_counts": {
    "musique": 200,
    "2wikimqa": 200,
    "hotpotqa": 200
  }
}