train_results.json

{
    "before_init_mem_cpu": 3040526336,
    "before_init_mem_gpu": 512,
    "epoch": 18.0,
    "init_mem_cpu_alloc_delta": 18427904,
    "init_mem_cpu_peaked_delta": 0,
    "init_mem_gpu_alloc_delta": 0,
    "init_mem_gpu_peaked_delta": 0,
    "total_flos": 2.949719748897669e+18,
    "train_loss": 0.22952656600628965,
    "train_mem_cpu_alloc_delta": 826355712,
    "train_mem_cpu_peaked_delta": 403324928,
    "train_mem_gpu_alloc_delta": 690193408,
    "train_mem_gpu_peaked_delta": 19408810496,
    "train_runtime": 33563.6304,
    "train_samples": 502272,
    "train_samples_per_second": 448.943,
    "train_steps_per_second": 0.877
}