HarderThenHarder/transformers_tasks issues and pull requests

#90 - 文本匹配任务Sentence Transformer疑问求解

Issue - State: open - Opened by magicleo 16 days ago

#89 - 文本匹配任务的一些问题

Issue - State: open - Opened by EvanSong77 17 days ago

#88 - ssl连接失败，无法下载 bert-base-chinese

Issue - State: open - Opened by YKT-668 about 1 month ago

#87 - 大佬们看到help！！TypeError: tuple indices must be integers or slices, not str

Issue - State: open - Opened by 11pomelo 4 months ago

#86 - 求解🙇🏻‍♀️标注平台打开网页失败运行streamlit run playground_local.py --server.port 8001

Issue - State: open - Opened by keyilizzie 8 months ago - 1 comment

#85 - 运行UIE train.sh时报错，求解答

Issue - State: closed - Opened by keyilizzie 8 months ago

#84 - 关于zero-shot

Issue - State: open - Opened by pengcheng-yan 9 months ago

#83 - mbti_questions.json 这个测试文件可以分享一下吗

Issue - State: closed - Opened by zhongxing-mt 12 months ago - 1 comment

#82 - 发现错误的代码

Issue - State: open - Opened by Victoria-aaaaaa about 1 year ago - 1 comment

#81 - 大佬您好，请问在做词表扩充的这个环节，为什么在llama的词表基础上扩充，而不是选择ChatGLM2的分词器

Issue - State: open - Opened by GromZhang about 1 year ago

#80 - 基于lora的方式finetune模型后，推理时间很长

Issue - State: open - Opened by wangjiangyue0226 about 1 year ago

#79 - 请问目前的对于chatglm的微调方法，对于新发布的chatglm2适用吗？

Issue - State: closed - Opened by GromZhang about 1 year ago - 1 comment

#78 - 关于ptuing

Issue - State: open - Opened by init-neok about 1 year ago - 4 comments

#77 - entity linking基于大模型的方案什么时候放出呢？有计划么

Issue - State: open - Opened by amelieyu about 1 year ago

#76 - fix typo in core.py

Pull Request - State: closed - Opened by eltociear about 1 year ago

#75 - 您这里的uie是生成式还是抽取式，cpu速度如何

Issue - State: open - Opened by BEILOP about 1 year ago

#74 - UIE抽取为什么加载不了模型

Issue - State: open - Opened by Smile-shirley about 1 year ago

#73 - PPO中critic模型不是应当使用reward model模型吗？

Issue - State: open - Opened by zhangjian94cn about 1 year ago

#72 - UIE支持uie-m-large吗

Issue - State: open - Opened by liaoweiguo about 1 year ago

#71 - p-tuning 换成英文数据集会报错， raise ValueError(f'Lable Error: "{label}" not in label_dict {list(self.label_dict)}.')

Issue - State: open - Opened by chengturbo over 1 year ago

#70 - where can I download checkpoints/model_1000？

Issue - State: open - Opened by hsauod over 1 year ago - 3 comments

#69 - 多卡训练后，推理时提示找不到tokenization_chatglm.py

Issue - State: open - Opened by GromZhang over 1 year ago - 5 comments

#68 - chatglm 微调

Issue - State: open - Opened by xiaogogonuo over 1 year ago

#67 - 请问，在您的zeroshot当中，仅仅是通过反馈给模型的history为我们设定的prompt，就完成了某一类的任务问题是吗？

Issue - State: open - Opened by GromZhang over 1 year ago - 1 comment

#66 - 在finetune中，请问save_model的时候，是把glm6b的model和lora训练后的model合并成一个了是吗？

Issue - State: closed - Opened by GromZhang over 1 year ago - 7 comments

#65 - --mixed_precision=fp16

Issue - State: open - Opened by WGS-note over 1 year ago - 2 comments

#64 - UIE模型train文件运行报错

Issue - State: open - Opened by White-bbb over 1 year ago - 2 comments

#63 - 单卡lora微调需要多少显存？

Issue - State: open - Opened by CCzzzzzzz over 1 year ago - 4 comments

#62 - 训练时带参数--quantization_bit 4 报错RuntimeError: self and mat2 must have the same dtype

Issue - State: open - Opened by shangzhensen over 1 year ago - 3 comments

#61 - 请问数据格式为什么要是把instruct，input，answer这种放在context里面的格式，直接传统的instruct，input，output三元组不行吗，差异在哪里

Issue - State: open - Opened by hangzeli08 over 1 year ago - 1 comment

#60 - RuntimeError: CUDA out of memory. Tried to allocate 42.00 MiB (GPU 0; 22.02 GiB total capacity; 20.

Issue - State: open - Opened by chenchaoac over 1 year ago - 5 comments

#59 - 请问标注平台是否支持chatglm或者chatrwkv等其他的语言模型

Issue - State: open - Opened by wzsage over 1 year ago

#58 - 请问这个repo里的peft代码是做过什么定制化的改动吗？是否可以直接把peft安装到python环境里呢？

Issue - State: open - Opened by zhoujz10 over 1 year ago

#57 - Hi,后面会考虑复现基于大模型如bloom、llama的rlhf流程代码？

Issue - State: open - Opened by nieallen over 1 year ago

#56 - 关于多卡训练总iter数设置

Issue - State: closed - Opened by HXZhong1997 over 1 year ago - 1 comment

#55 - 考虑支持MOSS吗

Issue - State: open - Opened by BestSongEver over 1 year ago

#54 - 关于GLM finetune的OOM

Issue - State: open - Opened by nuoma over 1 year ago

#53 - PET损失函数代码

Issue - State: closed - Opened by wuguangshuo over 1 year ago

#52 - 多卡训练，无法运行train_multi_gpu.sh

Issue - State: open - Opened by cheney369 over 1 year ago - 1 comment

#51 - 请问大模型微调需要多少显存？

Issue - State: open - Opened by Crayon-s7 over 1 year ago - 3 comments

#50 - 模型多卡训练

Issue - State: closed - Opened by HocherCN over 1 year ago

#49 - 关于词表问题

Issue - State: open - Opened by magnificent1208 over 1 year ago

#48 - LLM的finetune需要多大的显存啊？为什么我batch__size调到了1,24g的显存还是不够。。。。。。。

Issue - State: open - Opened by SlothRan over 1 year ago - 2 comments

#47 - 关于UIE模型微调，如何进行一机多卡训练？

Issue - State: open - Opened by zhougz2018 over 1 year ago - 2 comments

#46 - 大模型阅读理解微调

Issue - State: open - Opened by tianchiguaixia over 1 year ago

#45 - ValueError: Attempting to unscale FP16 gradients

Issue - State: open - Opened by magnificent1208 over 1 year ago - 1 comment

#44 - 能去掉网上的关联吗，每次都要重新下载

Issue - State: open - Opened by magnificent1208 over 1 year ago - 1 comment

#43 - UI界面打开提示模型未被定义

Issue - State: open - Opened by kbwzy over 1 year ago - 6 comments

#42 - playground推理过慢

Issue - State: open - Opened by nieallen over 1 year ago - 2 comments

#41 - 关于UIE训练效果

Issue - State: open - Opened by shenw8023 over 1 year ago - 1 comment

#40 - chatglm做finetune的时候报错

Issue - State: open - Opened by tianmala over 1 year ago

#39 - 最新更新代码后，模型训练保存出错

Issue - State: closed - Opened by nieallen over 1 year ago - 4 comments

#38 - 请教关于模型原理--关于response构成的问题

Issue - State: open - Opened by magnificent1208 over 1 year ago

#37 - 请教关于Tokenizer的问题

Issue - State: open - Opened by magnificent1208 over 1 year ago - 1 comment

#36 - 关于模型训练的问题

Issue - State: closed - Opened by chenchongyuan over 1 year ago - 1 comment

#35 - 关于训练的问题

Issue - State: open - Opened by chenchongyuan over 1 year ago - 2 comments

#34 - The hostname of the rendezvous endpoint ':8080' must be a dot-separated list of labels, an IPv4 address, or an IPv6 address.

Issue - State: open - Opened by lmx760581375 over 1 year ago

#33 - ValueError: 150001 is not in list

Issue - State: open - Opened by magnificent1208 over 1 year ago - 7 comments

#32 - UIE信息抽取

Issue - State: open - Opened by xxm1668 over 1 year ago - 3 comments

#31 - LLM微调

Issue - State: open - Opened by shuangshuangguo over 1 year ago - 2 comments

#30 - transformers_tasks\LLM\finetune训练模型的存放路径checkpoints/model_1000文件方便上传吗？

Issue - State: open - Opened by AnliComputer over 1 year ago

#29 - 你好，你使用peft是做了一些修改吗？是在那个版本上进行的修改？

Issue - State: closed - Opened by Ant0082 over 1 year ago - 1 comment

#28 - 纯请教 lora_rank 这个参数咋用，什么含义？求解

Issue - State: open - Opened by magnificent1208 over 1 year ago - 2 comments

#27 - 多卡训练求助？

Issue - State: closed - Opened by Traceve over 1 year ago - 4 comments

#26 - ModuleNotFoundError: No module named 'transformers.generation'

Issue - State: open - Opened by magnificent1208 over 1 year ago - 10 comments

#25 - train error 需要您的解答

Issue - State: open - Opened by zhangnn520 over 1 year ago - 3 comments

#24 - torch.load显示"no module named model"

Issue - State: closed - Opened by Singal0927 over 1 year ago - 6 comments

#23 - 由于 token 长度限制，怎么在 prompting 构造时输入更多的标注样本，提升分类精度呢？

Issue - State: open - Opened by leonall over 1 year ago - 3 comments

#22 - 打开标注平台的时候出现以下bug

Issue - State: open - Opened by YangJoy750 over 1 year ago - 4 comments

#21 - UIE第五部分数据增强，文件格式不对

Issue - State: closed - Opened by AnitaSherry over 1 year ago - 3 comments

#20 - PPO策略之后，更新后的模型是怎么存储的呢

Issue - State: open - Opened by tnickMoxuan over 1 year ago

#19 - 咨询一下RLHF算法的细节

Issue - State: open - Opened by WilfChen over 1 year ago - 2 comments

#18 - 在ppo的训练过程中生成质量越来越差。

Issue - State: open - Opened by tonekko over 1 year ago - 1 comment

#17 - 推理部分

Issue - State: closed - Opened by yhc582825016 over 1 year ago - 2 comments

#16 - “在最后一位加上人工给的reward” 的出处

Issue - State: closed - Opened by PYNing over 1 year ago - 5 comments

#15 - Here seems Wrong

Issue - State: closed - Opened by freesunshine0316 over 1 year ago - 5 comments

#14 - Might it better to change the loss for PET if one label corresponds to multiple label words?

Issue - State: open - Opened by xiningnlp over 1 year ago - 1 comment

#13 - a little bug for RLHF/train_reward_model.py

Issue - State: closed - Opened by archwolf118 over 1 year ago - 1 comment

#12 - 运行RLHF/train_reward_model.sh的疑问

Issue - State: closed - Opened by yaleimeng over 1 year ago - 3 comments

#11 - RuntimeError: The size of tensor a (16) must match the size of tensor b (17)

Issue - State: closed - Opened by archwolf118 over 1 year ago - 4 comments

#10 - 生成式任务里所谓问答模型（Text-Generation, T5 Based）似乎更像是阅读理解任务

Issue - State: closed - Opened by yaleimeng over 1 year ago - 1 comment

#9 - ppo中获取生成tokens的value 代码疑似有误

Issue - State: closed - Opened by wangxicoding over 1 year ago - 3 comments

#8 - 请问streamlit安装的是什么版本，

Issue - State: closed - Opened by answerme11 over 1 year ago - 2 comments

#7 - 您好，我将bert-base-chinese换成了roberta-base，然后报错了

Issue - State: closed - Opened by yumogu over 1 year ago - 1 comment

#6 - 没有dataset文件

Issue - State: closed - Opened by lzf00 over 1 year ago - 1 comment

#5 - transformers==4.22.1

Issue - State: closed - Opened by Victor-xx over 1 year ago - 3 comments

GitHub / HarderThenHarder/transformers_tasks issues and pull requests