all_results.json

{
    "epoch": 3.0,
    "eval_loss": 4.299047470092773,
    "eval_mem_cpu_alloc_delta": -192512,
    "eval_mem_cpu_peaked_delta": 0,
    "eval_mem_gpu_alloc_delta": 0,
    "eval_mem_gpu_peaked_delta": 1846225408,
    "eval_runtime": 5.1104,
    "eval_samples": 95,
    "eval_samples_per_second": 18.59,
    "init_mem_cpu_alloc_delta": -145973248,
    "init_mem_cpu_peaked_delta": 1313857536,
    "init_mem_gpu_alloc_delta": 1444470784,
    "init_mem_gpu_peaked_delta": 0,
    "perplexity": 73.629625865726,
    "train_mem_cpu_alloc_delta": 90374144,
    "train_mem_cpu_peaked_delta": 10436608,
    "train_mem_gpu_alloc_delta": 4257890816,
    "train_mem_gpu_peaked_delta": 2657059328,
    "train_runtime": 3295.7162,
    "train_samples": 4268,
    "train_samples_per_second": 1.943
}