sail-sg/regmix

Issues

• 1B mixture and validation data

  #9 opened 3 months ago by thompsonb
  7

• step1-mixture, 由于占比过小，类别在随机后变成了0

  #10 opened 3 months ago by feifei2023
  3

• Can you share the script for generating the prior distribution of the token？

  #8 opened 3 months ago by a154377713
  2

• Parameters and sizes of models mismatch

  #7 opened 4 months ago by clarence-lee-sheng
  2

• torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 请问你使用的是哪个版本的torch？我用的2.1.0 2.3.1 2.4的都报这个错

  #6 opened 4 months ago by dotsonliu
  3

• A question regarding data preprocessing.

  #5 opened 4 months ago by Zhang-Mozhi
  2

• Loss计算咨询

  #4 opened 4 months ago by zhaochs1995
  2

• So many data files after tokenizing

  #3 opened 5 months ago by shuxiaobo
  4

• Could you please upload the requirements.txt?

  #2 opened 5 months ago by shuxiaobo
  4

• 如何得到 test_pile_loss_1m.csv？

  #1 opened 5 months ago by YWMditto
  2