| global_step,perplexity,eval_loss,train_loss | |
| 10000,tensor(1.9330),0.659087556891259,0.5109515190124512 | |
| 20000,tensor(1.9278),0.6563766614771557,0.6348448395729065 | |
| 30000,tensor(1.9368),0.6610437813159591,0.7273242473602295 | |
| 40000,tensor(1.9303),0.6576739994996246,0.6494269371032715 | |
| 50000,tensor(1.9290),0.6570122954839838,0.6384057998657227 | |
| 52144,tensor(1.9377),0.6615094625095226,0.7903710007667542 | |