CVI-SZU/Linly

Issues

• llama3增量预训练冻结哪些层训练哪些层效果比较好？

  #137 opened a month ago by CanvaChen
  0

• 请问有没有性别年龄检测模型？

  #136 opened 2 months ago by broadcast98
  0

• 请问70B的模型要如何使用，抱脸上的模型看着文件和其他模型不一样

  #135 opened 5 months ago by afezeriaWrnbbmm
  0

• pretrain.py的示例似乎有点错误

  #134 opened 6 months ago by xinghudamowang
  0

• 请问，deepspeed 微调时，CPU的内存需要多大？

  #126 opened 10 months ago by Batmana
  1

• 在线地址无法使用

  #133 opened 7 months ago by zzSleep
  0

• 服务器最低配置要求是什么？

  #132 opened 7 months ago by jqs1124
  0

• 有人有pile的数据集吗？22个来源，825G的那个版本

  #131 opened 8 months ago by AI-Study-Han
  0

• 问下大佬们有没有训练3B的打算？场景需要时延不能太高

  #130 opened 9 months ago by zjuzhfbloodz
  0

• readme上的加群二维码过期了

  #129 opened 10 months ago by potong
  0

• 关于平行语料的预处理

  #93 opened a year ago by lyy-zz
  3

• Chinese-LLaMA-2-13B-hf样本模板prompt到底是什么样的？

  #128 opened 10 months ago by sunzhaowei
  0

• 关于Chinese-LLaMA-2-13B (hf格式)

  #125 opened a year ago by sun1092469590
  0

• Please clarify the License for Chinese-LLaMA-2

  #124 opened a year ago by JayLiangs
  1

• 微信满员了，请重新上传新的微信图片 我可以免费做管理员

  #123 opened a year ago by ArtificialZeng
  3

• 多轮对话问问题之后直接报错

  #122 opened a year ago by caowenhero
  0

• python3 llama_server.py结果乱码

  #121 opened a year ago by caowenhero
  0

• ChatFlow-13B.bin只有136字节

  #120 opened a year ago by NewEricWang
  1

• 关于33B模型预训练语料长度

  #96 opened a year ago by minlik
  2

• huggingface上openllama-13b的模型大小为26.4G,转换为huggingface那种模型格式之后模型大小为24.7G，这也就是大概是以fp16或者是bf16保存的模型

  #119 opened a year ago by belle9217
  0

• Are the tokenizer.model the same with the one in llama-7b?

  #118 opened a year ago by treya-lin
  0

• Chinese-LLaMA-33B在多少块gpu上训了多长时间？

  #117 opened a year ago by JingxinLee
  0

• 是否考虑通过位置插值来扩展大语言模型的上下文窗口 ，将上下文窗口提升至32K

  #116 opened a year ago by xfg0913
  0

• 请问在指令微调时损失函数与预训练有什么区别吗

  #115 opened a year ago by dazhaxie0526
  0

• open-llama13B做推理时，结果是英文

  #114 opened a year ago by yating0823
  0

• 使用openllama13B + openmodel进行推理时，结果都是数字？这个需要做其他操作？

  #104 opened a year ago by suhaibo1
  1

• 关于openllama的两个相关问题

  #106 opened a year ago by lucasjinreal
  1

• falcon的使用中文预料进行增量训练

  #113 opened a year ago by fengstar7827
  2

• readme上的加群二维码过期了

  #99 opened a year ago by aihaidong
  4

• 额，是我用错了吗?简单推理都不行吗

  #112 opened a year ago by Mousaic
  1

• Multi machine pre-training hung

  #111 opened a year ago by BUPTAnderson
  1

• 请问有中文falcon的下载地址嘛

  #110 opened a year ago by AlexXx-Wu
  2

• Wrong argments

  #109 opened a year ago by jeffchy
  0

• chatflow模型推理的时候，prompt需要加类似“human: {query}\n assistant: \n” 前后缀嘛？

  #108 opened a year ago by jeinlee1991
  0

• openllama 13b base model生成内容比较奇怪

  #107 opened a year ago by lucasjinreal
  0

• 请问是否有增量预训练的基础模型13B的评测结果？

  #105 opened a year ago by caihaunqai
  0

• 如何cite？

  #98 opened a year ago by hackerchenzhuo
  1

• 增量预训练的时候报错exits with return code = -9 ，单卡80G显存的A100

  #103 opened a year ago by pydaxing
  2

• 请问OpenLLaMA-13B在转换为hf模型时，convert_llama_from_tencentpretrain_to_hf.py直接复制了词表tokenizer.model，open_llama.model没有用到，是正常的吗？

  #94 opened a year ago by chk4991
  1

• HF在线崩溃了

  #102 opened a year ago by Batmana
  1

• Pretraining corpus formatting

  #101 opened a year ago by treya-lin
  0

• 7b模型性能和billa对比

  #100 opened a year ago by lucasjinreal
  0

• 请问大佬65B的模型何时能够放出

  #97 opened a year ago by Expert68
  0

• 请问openllama 13b怎么转成HF格式

  #95 opened a year ago by lin1490188
  0

• 博主群二维码过期了,可以更新一个新的二维码吗

  #87 opened a year ago by zhangfan-algo
  1

• Is it possible to support OPT models

  #88 opened a year ago by treya-lin
  1

• openllama性能评估

  #92 opened a year ago by enbacoo
  0

• 请问模型在tencentpretrain框架下预训练时选择的是bpe tokenizer吗？是否有对应的预训练的merge.txt呢？

  #91 opened a year ago by yyqi17
  0

• Chinese-LLaMA-33B (hf格式)的模型如何部署，进行推理？

  #90 opened a year ago by xfg0913
  1

• 33b Huggingface 格式怎么转成TencentPretrain 格式

  #89 opened a year ago by lyy-zz
  2