| loss,grad_norm,learning_rate,epoch,step | |
| 1.4885,2.976400136947632,3.336666666666667e-05,1.0,500 | |
| 1.1504,4.292872428894043,1.6700000000000003e-05,2.0,1000 | |
| 1.0006,7.022700309753418,3.3333333333333334e-08,3.0,1500 | |
| 11730.008,4.092,0.128,1020745916282304.0,1.2131712443033855,3.0,1500 | |