dandelionsllm/pandallm

Issues

• 在线体验 pandallm.ai/，该链接无法打开

  #35 opened 9 months ago by Etttz
  0

• 在线体验需要访问密码？

  #34 opened a year ago by wanghx629
  0

• Panda-LLaMA2-13B的模型链接不对吧？

  #33 opened a year ago by sun1092469590
  3

• 模型合并推理结果不正常

  #5 opened a year ago by xxxxuee
  11

• Removed.

  #32 opened a year ago by Anna10071
  0

• RuntimeError: The size of tensor a (32000) must match the size of tensor b (32001) at non-singleton dimension 0

  #31 opened a year ago by jeinlee1991
  2

• 关于推理的问题

  #30 opened a year ago by Jiangchenglin521
  5

• 由于训练数据的预处理存在不确定性，请问可以进一步分享处理代码吗

  #29 opened a year ago by Cascol-Chen
  2

• 如果要重新训练panda，请问数据集应该以什么结构存储呢

  #28 opened a year ago by Cascol-Chen
  5

• 能否提供一下模型量化的方法？其他在mac上进行推理测试

  #27 opened a year ago by doctor1984
  1

• 商用7b模型，直接进行推理说话毫无逻辑，难道不是开箱即用的嘛？

  #26 opened a year ago by Goo-goo-goo
  4

• 单卡A100训练7B模型OOM

  #25 opened a year ago by HuihuiChyan
  12

• 请问如何在COIG数据集上继续训练

  #23 opened 2 years ago by scarydemon2
  8

• 如何将COIG数据集处理为instruction tuning的格式

  #24 opened 2 years ago by HuihuiChyan
  1

• 很值得期待的新项目，加油。

  #7 opened 2 years ago by imgingroot
  5

• 权重转换脚本不支持来自 Facebook 的原版权重

  #22 opened 2 years ago by ShadowPower
  2

• 推理超参数

  #17 opened 2 years ago by suolyer
  1

• 请问有使用的examples吗

  #21 opened 2 years ago by Mr-IT007
  1

• 词表没扩充过是吗?还是32,000吧？

  #19 opened 2 years ago by mellivoraPKU
  1

• linly预训练阶段和panda类似，指令微调阶段使用了500万条，为什么效果比panda差？

  #20 opened 2 years ago by mellivoraPKU
  1

• 预训练和微调阶段使用的是全量微调还是lora?

  #18 opened 2 years ago by mellivoraPKU
  1

• 感觉模型比原始的LlaMa模型大了很多，是什么原因

  #16 opened 2 years ago by uloveqian2021
  1

• 13b的结果,很奇怪

  #15 opened 2 years ago by frog-game
  8

• 模型合并后的MD5？

  #13 opened 2 years ago by hejujie
  1

• response prefixes and renewable energy

  #14 opened 2 years ago by stakodiak
  1

• 请问4090单卡可以完成7B模型的训练吗

  #11 opened 2 years ago by mircop1t
  4

• 输出不理想~

  #12 opened 2 years ago by frog-game
  10

• 65b-model

  #8 opened 2 years ago by tfka
  1

• 推理的效果很奇怪，能否帮忙看一下原因

  #9 opened 2 years ago by ccyhxg
  2

• 有做和ChatGLM对比的评测吗

  #6 opened 2 years ago by lanyuer
  3

• 模型合并时报错

  #2 opened 2 years ago by enbacoo
  7

• 关于评估选用的BELLE baseline问题

  #4 opened 2 years ago by hejujie
  1

• llama.cpp support

  #3 opened 2 years ago by lucasjinreal
  1

• RuntimeError: The size of tensor a (32000) must match the size of tensor b (32001) at non-singleton dimension 0

  #1 opened 2 years ago by gqjia
  1