Data/Generation_Model/Content/GRU/LARGE/eval_test.json

{
  "checkpoint": "/net/work/wangq/CycAMP/Scripts/Generation_Model/Models_ckpt/260226_content/GRU/LARGE/best.pt",
  "tokenizer_path": "/net/work/wangq/CycAMP/Scripts/Generation_Model/Tokenizer/content_vocab.json",
  "test_data_path": "/net/work/wangq/CycAMP/Scripts/Generation_Model/Data/processed/hierarchical/test_content.jsonl",
  "config": {
    "arch": "gru",
    "d_model": 768,
    "num_layers": 4,
    "max_len": 512,
    "dropout": 0.3,
    "label_smoothing": 0.0
  },
  "evaluation": {
    "device": "cuda",
    "batch_size": 512,
    "num_samples": 75413,
    "timestamp": "2026-02-27T01:39:32.264670"
  },
  "metrics": {
    "loss": 1.9885217692978947,
    "perplexity": 7.3047277045907295
  }
}