AI-Study-Han/Zero-Chatgpt

Issues

• 请问如果想将模型参数量从0.1B改为0.2B，或者改为0.05B，应该改动哪些参数呢？

  #12 opened 19 days ago by ybdesire
  4

• Would you help provide more detailed steps to reproduce the training process?

  #11 opened a month ago by ybdesire
  5

• 为什么在跑rlhf过程中既没有报错也没输出rlhf结果

  #10 opened 3 months ago by qilele00
  7

• 镜像tar包直接用docker load -i 导入吗，我导入后创建容器失败了

  #9 opened 4 months ago by liaozhihao2757
  1

• max_position_embeddings

  #8 opened 4 months ago by jasonzou
  1

• SkyPile-150B数据是全量随机采样嘛

  #7 opened 4 months ago by hegc
  1

• 训练使用的硬件条件

  #4 opened 4 months ago by hegc
  1

• Sir, about the code to create the pretrain model. What is it？

  #6 opened 4 months ago by mzb98
  3

• 你好，请问大概需要多少显存呢？

  #5 opened 4 months ago by Runinthenight
  2

• Sir, what is the json file format for training?

  #3 opened 4 months ago by mzb98
  4

• Sir, How do I generate a training set?

  #2 opened 4 months ago by mzb98
  1

• 能给个镜像吗？另外训练曲线训练效果耗时能贴出来吗？

  #1 opened 4 months ago by dotsonliu
  2