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

{
  "checkpoint": "/net/work/wangq/CycAMP/Scripts/Generation_Model/Models_ckpt/260226_content/GPT/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": "gpt",
    "d_model": 768,
    "num_layers": 12,
    "max_len": 512,
    "dropout": 0.4,
    "label_smoothing": 0.0
  },
  "evaluation": {
    "device": "cuda",
    "batch_size": 128,
    "num_samples": 75413,
    "timestamp": "2026-02-28T09:43:11.043200"
  },
  "metrics": {
    "loss": 1.8888031629637274,
    "perplexity": 6.611451114412459
  }
}