HIT-SCIR/Chinese-Mixtral-8x7B

Issues

• TypeError: SFTConfig.__init__() got an unexpected keyword argument 'cache_dir'

  #31 opened a month ago by wormcode
  1

• 请问 SFT 训练的时候 训练 数据的格式 是怎么样的？

  #30 opened 2 months ago by wormcode
  3

• 关于词表扩充

  #29 opened 7 months ago by CrazyBoyM
  2

• 增量预训练出错

  #28 opened 7 months ago by lwj2001
  8

• 微调

  #27 opened 7 months ago by cripsgreen
  1

• 咨询一下，mixtral-8x7B在未增量训中文的情况下，ceval、cmmlu的得分是多少呢？在进行42B token续训之后分数有明显增长么？

  #26 opened 8 months ago by KalsaHT
  0

• 微调的--mode 错了吧

  #25 opened 8 months ago by flyingwaters
  2

• 请问微调需要的最小的GPU是多少？

  #24 opened 8 months ago by chaishenAI
  1

• torchrun: command not found

  #23 opened 8 months ago by lfxuan
  1

• 能否说下硬件需求。

  #16 opened 9 months ago by orderer0001
  3

• 通信量不一致

  #21 opened 9 months ago by longzhang418
  1

• ALP横坐标意义

  #20 opened 9 months ago by Jianwei-Lv
  1

• 关于embedding扩展的细节

  #22 opened 9 months ago by cometyang
  1

• 没有tokenizer文件

  #19 opened 9 months ago by cometyang
  3

• 询问一下data.utils套件要在哪裡下载？

  #18 opened 9 months ago by DanielChen1128
  1

• 请问下MMLU评测的时候有增加什么prompt吗？

  #17 opened 9 months ago by matrixssy
  3

• 有没有考虑过将模型文件上传到modelscope

  #11 opened 10 months ago by Tendo33
  2

• autoawq量化后推理很奇怪

  #8 opened 10 months ago by gptbert
  2

• 希望能量化成q4或者q6的形式，并尽快引入Ollama，这样个人也能用了，感谢！

  #2 opened 10 months ago by ricksuzade-maker
  3

• init_embeddings 模型转换问题

  #14 opened 10 months ago by dachengai
  1

• 请教一下预训练数据使用情况

  #13 opened 10 months ago by CLeafYeah
  4

• 请问训练一次要多长时间？

  #12 opened 10 months ago by gongbudaizhe
  1

• 代码中缺失了调用init_embeddings.py的脚本，纯新手，能提供一下吗？

  #10 opened 10 months ago by whosyourdadds
  2

• 这么好的项目，建议作者持续更新！

  #7 opened 10 months ago by whosyourdadds
  1

• 4bit量化推理速度变慢

  #6 opened 10 months ago by snowlixue
  4

• Mixtral

  #9 opened 10 months ago by 591094733
  0

• 词表扩充数据：12G知乎数据和2G悟道数据上训练中文BPE词表

  #5 opened 10 months ago by LiuChaoXD
  2

• 您好，请问指令微调需要多少显存，有没有可以推荐的指令微调数据集？

  #3 opened 10 months ago by tommy3266
  1

• 您好，我这里使用4bit量化后get_peft_model报错

  #4 opened 10 months ago by Ahu-Fgx
  2

• 请问，训练需要至少多大的显存？

  #1 opened 10 months ago by LiuChaoXD
  2