Issues

• Problems in train_deepspeed.py with ZeRO stage 1|2|3

  yanqiangmiffy/InstructGLM#34 · Filter by author zjhJOJO opened on Sep 7, 2023

• 用tokenizer_dataset_rows.py转换自己的数据报错datasets.builder.datasetgeneraationerror

  yanqiangmiffy/InstructGLM#33 · Filter by author cat1222 opened on May 9, 2023

• 使用lora训练，使用web_demo加载lora权重后，结果跟原生chatglm结果一样，lora权重没生效，这个是什么原因呢

  yanqiangmiffy/InstructGLM#32 · Filter by author AnddyWang opened on May 6, 2023
  1

• 用BelleGroup/train_1M_CN训练后，为什么用数据集里的问题测，回答不一样

  yanqiangmiffy/InstructGLM#31 · Filter by author czhcc opened on May 5, 2023

• 调教后的逻辑能力如何？

  yanqiangmiffy/InstructGLM#30 · Filter by author daiaji opened on Apr 29, 2023
  1

• train_lora最低需要多大显存GPU可以训练？除了batch size 还有别的参数可以降低显存使用吗？

  yanqiangmiffy/InstructGLM#29 · Filter by author twosnowman opened on Apr 27, 2023
  2

• 用peft加载lora后，generate时报错ValueError: 130000 is not in list，加载lora之前推理是正常的

  yanqiangmiffy/InstructGLM#28 · Filter by author BIGPPWONG opened on Apr 25, 2023

• 请问下train_deepspeed.py 怎么引入lora.pt

  yanqiangmiffy/InstructGLM#27 · Filter by author AlexXx-Wu opened on Apr 20, 2023
  1

• ValueError: ChatGLMForConditionalGeneration does not support gradient checkpointing.

  yanqiangmiffy/InstructGLM#25 · Filter by author deepeye opened on Apr 19, 2023
  7

• datasets.builder.InvalidConfigName: Bad characters from black list '<>:/\|?*' found in 'data/belle_data.json'. They could create issues when creating a directory for this config on Windows filesystem.

  yanqiangmiffy/InstructGLM#23 · Filter by author deepeye opened on Apr 19, 2023
  1

• 4张32G的可以吗，作者可以用你写的其他开源数据集finetune看看效果吗，再放出转换和训练代码

  yanqiangmiffy/InstructGLM#21 · Filter by author hangzeli08 opened on Apr 15, 2023
  2

• 4张 12G的 3060能训练吗

  yanqiangmiffy/InstructGLM#20 · Filter by author zlszhonglongshen opened on Apr 14, 2023
  2

• 运行web_demo_alpaca_lora.py报错，是单纯的显存不够嘛

  yanqiangmiffy/InstructGLM#19 · Filter by author tianmala opened on Apr 14, 2023
  1

• 训练python train_lora.py的时候显示 ModuleNotFoundError: No module named 'configuration_chatglm'

  yanqiangmiffy/InstructGLM#18 · Filter by author dragononly opened on Apr 13, 2023
  3

• 测试数据打不开https://huggingface.co/datasets/BelleGroup/generated_train_0.5M_CN

  yanqiangmiffy/InstructGLM#17 · Filter by author dragononly opened on Apr 12, 2023
  2

• 微调2:BELLE中文指令数据的问题

  yanqiangmiffy/InstructGLM#16 · Filter by author czhcc opened on Apr 12, 2023
  1

• 运行 finetune.py 遇到问题：OSError: /data/pretrained-chatglm-6b/ does not appear to have a file named config.json

  yanqiangmiffy/InstructGLM#13 · Filter by author xubuvd opened on Apr 4, 2023
  1

• 最新update的代码中，web_demo推理时报错

  yanqiangmiffy/InstructGLM#12 · Filter by author feyxong opened on Apr 3, 2023
  2

• 关于训练完成后，生成的答案总是带一些莫名奇妙的Q，A数据，真的不造是哪里出了问题，还望大佬赐教！谢谢！

  yanqiangmiffy/InstructGLM#11 · Filter by author UMU689 opened on Apr 2, 2023
  1

• 关于多轮对话的疑问

  yanqiangmiffy/InstructGLM#9 · Filter by author ZeyuTeng96 opened on Mar 29, 2023

• web_demo_belle生成结果时有大段重复的问题

  yanqiangmiffy/InstructGLM#8 · Filter by author JiayiFu opened on Mar 29, 2023
  10

• ValueError: 150000 is not in list

  yanqiangmiffy/InstructGLM#5 · Filter by author superhg opened on Mar 25, 2023
  5