THUDM/GLM

Issues

• ds_finetune_superglue.sh 脚本如何配置流水线并行

  #209 opened 3 months ago by dreamstick
  0

• 110M的模型有huggingface版本吗？

  #208 opened 5 months ago by WzjCoder
  0

• 建议上架ollama

  #207 opened 5 months ago by heimy2000
  0

• ms-swift支持了glm-4v-9b多模态大模型的微调(finetune)🚀😊

  #206 opened 5 months ago by Jintao-Huang
  2

• 模型的分词逻辑

  #205 opened 6 months ago by loki-keroro
  0

• 运行bash scripts/generate_block.sh config_tasks/model_blocklm_10B_chinese.sh报错

  #198 opened a year ago by XiaozhuLove
  1

• Add special_token

  #204 opened 7 months ago by chenzebiaohub
  0

• 在没有InfiniBand情况下能训练glm-large吗

  #173 opened 2 years ago by allendred
  3

• Few-shot tests on GLM-10B

  #203 opened 8 months ago by Vispstar-V
  0

• Eligibility for Commercial Use

  #174 opened 2 years ago by Hegelim
  1

• What is the license of Pretrained Models?

  #202 opened 10 months ago by phuchm
  0

• bug report！rouge-1 = 0.0000 rouge-2 = 0.0000 rouge-l = 0.0000

  #201 opened 10 months ago by runningabcd
  1

• 如果用glm-chinese-large 版本进行微调，相关的配置需要更改吗？

  #200 opened 10 months ago by runningabcd
  0

• 请教一下大家，glm0.3b有什么可用的推理加速的方法吗？目前我的推理任务要3秒钟一个，耗时太长

  #199 opened a year ago by mechigonft
  1

• mpi4py库

  #197 opened a year ago by XiaozhuLove
  0

• ImportError: cannot import name 'container_abcs' from 'torch._six' (/root/anaconda3/envs/lss/lib/python3.8/site-packages/torch/_six.py)

  #189 opened a year ago by LssTry
  1

• 使用glm-large-chinese微调分类任务

  #196 opened a year ago by mechigonft
  0

• 微调glm-large-chinese，不能使用deepspeed吗？

  #195 opened a year ago by mechigonft
  0

• 在使用glm-large-chinese微调分类任务时报错

  #194 opened a year ago by mechigonft
  0

• 使用glm-2b时候，跟随readme提供的例子，得到很糟糕的输出

  #178 opened a year ago by leekum2018
  2

• 运行 GLM-10B 的最低配置是多少？

  #191 opened a year ago by nguyenvanhoangphuc
  1

• 使用GLM-2b推理时生成无意义内容

  #192 opened a year ago by ChristLBUPT
  0

• 请发布一个小参数版本的ChatGLM，与ChatGLM-6B共享Tokenizer，让RLHF最后一步PPO能够最大可能提速

  #159 opened 2 years ago by yynil
  4

• 使用glm-10b-chinese调用generate方法有时时会出错

  #187 opened a year ago by adzhua
  1

• 我基于10B模型做继续训练，loss只从11下降到5

  #160 opened 2 years ago by TccccD
  6

• 使用Zero-1+cpu_offload=true时，出现错误？

  #190 opened a year ago by SkrDrag
  2

• MP_size大于1 continue pretrained后的模型 怎么转换成transformer模型进行测试

  #188 opened a year ago by wangerxiao001
  0

• 调用glm模型，遇到modeling_glm.py的bug：attention_mask初始化device设置遗漏

  #186 opened a year ago by luo-li-ba-suo
  1

• 请问有人使用GLM跑通过Continual Pre-training么？

  #185 opened a year ago by wjn1996
  0

• glm-10b-chinese原始模型推理报错

  #184 opened a year ago by Mryangkaitong
  0

• 在预训练Pretrain时报no valid `self._rcvd_idx` is found错误

  #170 opened 2 years ago by yt7589
  3

• 有对glm-10b-chinese这个模型做过评测的吗？

  #183 opened a year ago by hegang1-tal
  1

• 如何将GLM10B封装成对话式API

  #182 opened a year ago by yihuaxiang
  3

• 用transformers包，下载文件到本地后无法加载AutoTokenizer

  #181 opened a year ago by PolarisRisingWar
  0

• glm-10b / tokenization_glm.py

  #180 opened a year ago by chenhaoenen
  0

• 预训练的数据格式可以给个示例吗，可以不显示数据，就想看下格式

  #179 opened a year ago by gyh123wqe
  1

• block_lm_ratio参数

  #176 opened a year ago by chenhaoenen
  1

• parameter SCB

  #169 opened 2 years ago by zhaoqf123
  1

• 求问glm-10b-chinese推理所需最低配置

  #177 opened a year ago by TianYangCai
  0

• 用"THUDM/glm-10b-chinese"做分类任务出错

  #162 opened 2 years ago by 18335100284
  2

• 请问微调模型的 参考资料哪里可以学习借鉴

  #175 opened 2 years ago by thurdaypeng
  0

• GLM-10B中文版预训练权重下载后解压失败

  #172 opened 2 years ago by echosyy
  0

• 数据集格式是怎么样的？能否把一篇一万字的文档整体塞进去训练？另外对显卡要求是多高

  #171 opened 2 years ago by dizhenx
  0

• GLM-10B-Chinese模型文件太大无法解压

  #161 opened 2 years ago by wusi1590
  1

• .

  #165 opened 2 years ago by xxzcc
  0

• 请问 glm-10b-chinese 模型初始loss是多少，我的是1.7左右合理吗

  #168 opened 2 years ago by shouwangzhe
  0

• glm-10b-chinese模型的预训练数据量

  #166 opened 2 years ago by wlike
  0

• GLM 10B和ChatGLM 6B模型架构的差别

  #163 opened 2 years ago by ccsquare
  3

• 环境问题：python 版本号与 requirements.txt 中的版本号，以及一些依赖

  #164 opened 2 years ago by LucienShui
  0

• 关于GLM的有以下两个问题？1.为什么predict的时候没有加linear映射到词表维度，而是直接与word_embeddings相乘映射到词表维度了。 2.GLM加载使用AutoModelForSeq2SeqLM，而没有使用AutoModelForCausualLM，原因是什么？

  #158 opened 2 years ago by macheng6
  0