Anonymous
6/14/2025, 5:46:48 AM
No.105588222
>>105588207
I think so. It was around the same time that he increased the lr for an epoch and it broke the model. I know he rolled the lr change back, but I think he kept that quasi-distillation method.
I think so. It was around the same time that he increased the lr for an epoch and it broke the model. I know he rolled the lr change back, but I think he kept that quasi-distillation method.