| iter,train_loss,val_loss,lr |
| 0,9.819828033447266,9.82157039642334,0.0 |
| 500,5.613026142120361,5.6414713859558105,6.25e-05 |
| 1000,4.421712875366211,4.406324863433838,81.96766690215053,0.000125 |
| 1500,3.852750301361084,3.8415539264678955,46.59782780729977,0.0001875 |
| 2000,3.4628515243530273,3.4731228351593018,32.23725705756281,0.00025 |
| 2500,3.2453742027282715,3.247955083847046,25.737654721269692,0.00024993976609109114 |
| 3000,3.108496904373169,3.132746934890747,22.936899253546873,0.00024975912886434287 |
| 3500,3.0051283836364746,3.0273146629333496,20.641728101023848,0.00024945828175062216 |
|
|
