clue-ai/PromptCLUE

Issues

• fine-tune out of memory

  #11 opened 2 years ago by jiangix-paper
  3

• 使用 pCLUE-main 项目里面的datasets里面的120万数据训练后,效果不佳

  #20 opened a year ago by yzhzx
  1

• ClueAI/PromptCLUE-base-v1-5 分类任务

  #19 opened 2 years ago by zzfer1
  3

• base模型使用问题

  #18 opened 2 years ago by wzg-zhuo
  7

• pCLUE数据集只有5类标签，与PromptCLUE中的16类任务对应不上

  #15 opened 2 years ago by plmsmile
  5

• 针对t5-large模型的训练问题

  #17 opened 2 years ago by Tian14267
  1

• 基于大模型的优化

  #16 opened 2 years ago by xxllp
  1

• 任务的完成情况有评价分数吗？clueai里可以返回likelihood，promptclue里没有找到

  #14 opened 2 years ago by munian08
  1

• 能否开源各个任务的数据格式（demo）？

  #13 opened 2 years ago by skykiseki
  2

• 请问“营销文案生成”的任务不在base或base-1.5的版本中吗？在线体验中的该功能属于哪个版本的模型呢

  #12 opened 2 years ago by sunny0315
  1

• 建议发布small 版本模型 用于下游自定义单任务 finetune

  #10 opened 2 years ago by svjack
  1

• 用于分类任务时，怎么保证每次输出的预测类别为单个类别标签

  #9 opened 2 years ago by jinsongpan
  1

• large模型

  #8 opened 2 years ago by jiangix-paper
  1

• 为什么相同参数,每次运行结果都会不一样呢?

  #7 opened 2 years ago by thouger
  1

• 咱们这个效果对比上有和现有开源模型的对比吗？

  #6 opened 2 years ago by XiaoqingNLP
  0

• client.py 216行未考虑response为空的情况

  #4 opened 2 years ago by cpg361
  1

• 能够把每个任务的训练代码开源？

  #5 opened 2 years ago by zbbwss
  1

• 预训练代码开源

  #3 opened 2 years ago by blueseasky
  1

• 这边模型训练是怎么训练的 ？

  #1 opened 2 years ago by zbbwss
  1