GCYZSL/MoLA

Issues

• 您好，能否适配下较新版的transformers、peft和pytorch框架

  #15 opened 7 months ago by FrankMinions
  4

• attention_mask 为全 1 时 causal_mask 为 None，导致训练出现问题

  #21 opened 6 months ago by zyc140345
  3

• 作者您好，想问一下在推理时，是把每一个LoRA的权重都融入进pretrained weights中吗

  #19 opened 6 months ago by suntea233
  4

• 关于commonsense_qa和openbook_qa的preparation和evaluation代码

  #17 opened 7 months ago by Yang-Yan-Yang-Yan
  3

• Training on ScienceQA data:出现问题

  #18 opened 7 months ago by today-still-sleep-early
  5

• 关于辅助损失处理部分

  #16 opened 7 months ago by lhyscau
  1

• 您好请问训练需要几张卡多大的显存啊

  #14 opened 7 months ago by sunrise224
  4

• Can not reproduce the results on GLUE

  #12 opened 9 months ago by Car-pe
  10

• mathqa上训练完之后，想再在scienceqa上进行训练

  #13 opened 9 months ago by hy010227
  0

• 请问可以提供instruction数据集的评估代码吗？

  #11 opened 9 months ago by hy010227
  8

• How to get the eval results claimed in the paper

  #10 opened 9 months ago by Car-pe
  1

• 关于实验中的NLP部分

  #8 opened 9 months ago by login256
  3

• Where can I get the "Training of Quick Start"

  #9 opened 9 months ago by Car-pe
  7

• 怎么处理除了scienceqa之外的数据集？

  #1 opened 10 months ago by 2018211801
  3

• 请问load_balancing_loss_func为什么从原版改成现在这种分层累加计算的呢

  #6 opened 9 months ago by 2018211801
  1

• Questions about experimental results

  #7 opened 10 months ago by JH-Yin
  0

• 无

  #5 opened 10 months ago by 2018211801
  0

• 按照原参数配置跑的模型，评测出的scienqa只有50多分

  #3 opened 10 months ago by 2018211801
  4

• 工作是好工作，就是代码太屎山了

  #4 opened 10 months ago by wkml
  1

• 示例中的传参有问题！命令行中的参数没有正确的传到程序

  #2 opened 10 months ago by 2018211801
  1