FlagOpen/FlagScale

Issues

• 异构训练报NCCL错误

  #201 opened 2 months ago by echo-valor
  2

• 从checkpoint中加载模型进行增量预训练

  #200 opened 2 months ago by echo-valor
  0

• [BUG or ENHANCEMENT] about SSH runner

  #193 opened 2 months ago by shh2000
  0

• 异构训练问题

  #179 opened 2 months ago by derekwin
  2

• 请问pp异构并行训练时，划分模型层列表是否只能通过手动设置？

  #153 opened 4 months ago by JinXiaozhao
  2

• [QUESTION] 支持在异构设备上进行训练吗？

  #31 opened 9 months ago by slbqc
  2

• llama2 70B模型在不同PP下loss下降趋势不同

  #81 opened 7 months ago by ZLkanyo009
  1

• 期望新增以下切割模型权重的功能

  #68 opened 7 months ago by helen88
  0

• [QUESTION] 请问是否能公开将HuggingFace版本的Aquila模型转化为Megatron模型的脚本和代码？

  #32 opened 9 months ago by stainswei
  1

• [QUESTION] Support other hardware?

  #16 opened 10 months ago by SueeH
  1

• Support Lora or other peft

  #1 opened a year ago by 2033329616
  3