minghaochen/chatglm3-base-tuning

Issues

• 微调

  #12 opened 10 months ago by sevenandseven
  1

• 训练过程中报错显存不够，但是明显有剩余显存。

  #10 opened a year ago by DeMoth-1
  2

• LoRa合并报错

  #11 opened a year ago by DeMoth-1
  2

• sft相关问题

  #9 opened a year ago by zbrnwpu
  0

• RuntimeError: Expected all tensors to be on the same device

  #8 opened a year ago by anfernees-iphone
  1

• 训练报错

  #7 opened a year ago by makeukus
  2

• RuntimeError: Subtraction, the `-` operator, with a bool tensor is not supported. If you are trying to invert a mask, use the `~` or `logical_not()` operator instead.

  #5 opened a year ago by hly666666
  2

• user 特殊token

  #6 opened a year ago by bravelll
  0

• 按照您的步骤，训练的过程中loss一直为零，请问下作者您有遇到过吗？

  #4 opened a year ago by Rocky77JHxu
  4

• 请问SFT需要多少显存？

  #3 opened a year ago by tangsipeng
  1

• 关于微调数据的问题

  #2 opened a year ago by 948024326
  1

• torch报错不支持bool类型tensor的-操作

  #1 opened a year ago by Yeuoly
  1