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

{
  "checkpoint": "/net/work/wangq/CycAMP/Scripts/Generation_Model/Models_ckpt/260226_content/GPT/SMALL/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": 256,
    "num_layers": 4,
    "max_len": 256,
    "dropout": 0.25,
    "label_smoothing": 0.0
  },
  "evaluation": {
    "device": "cuda",
    "batch_size": 1024,
    "num_samples": 75413,
    "timestamp": "2026-02-26T23:54:11.147988"
  },
  "metrics": {
    "loss": 2.1092825794792187,
    "perplexity": 8.242325949213935
  }
}