all_results.json

{
    "epoch": 1.0,
    "eval_loss": 3.290411949157715,
    "eval_mem_cpu_alloc_delta": 544768,
    "eval_mem_cpu_peaked_delta": 73728,
    "eval_mem_gpu_alloc_delta": 0,
    "eval_mem_gpu_peaked_delta": 2645388800,
    "eval_runtime": 803.942,
    "eval_samples": 28999,
    "eval_samples_per_second": 36.071,
    "init_mem_cpu_alloc_delta": -36622336,
    "init_mem_cpu_peaked_delta": 1313382400,
    "init_mem_gpu_alloc_delta": 1444470784,
    "init_mem_gpu_peaked_delta": 0,
    "perplexity": 26.853923828930338,
    "train_mem_cpu_alloc_delta": 596549632,
    "train_mem_cpu_peaked_delta": 0,
    "train_mem_gpu_alloc_delta": 4257896960,
    "train_mem_gpu_peaked_delta": 2961890816,
    "train_runtime": 4429.0691,
    "train_samples": 29667,
    "train_samples_per_second": 1.675
}