Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / hiyouga/ChatGLM-Efficient-Tuning issues and pull requests

#100 - Issue about: IndexError: Invalid key: 0 is out of bounds for size 0

Issue - State: closed - Opened by mvllwong over 1 year ago - 3 comments
Labels: solved

#99 - 关于绘制loss函数曲线的问题

Issue - State: closed - Opened by CCzzzzzzz over 1 year ago - 3 comments
Labels: solved

#98 - 微调后的推理生成<image_-100>

Issue - State: closed - Opened by Cheung-Z over 1 year ago - 1 comment
Labels: solved

#97 - 微调后的推理生成<image_-100>

Issue - State: closed - Opened by Cheung-Z over 1 year ago

#96 - ValueError: 130000 is not in list

Issue - State: closed - Opened by taofennanhai over 1 year ago - 1 comment
Labels: solved

#95 - 请问是否支持GLM-10-chinese模型

Issue - State: closed - Opened by taofennanhai over 1 year ago - 1 comment
Labels: solved

#94 - 生成文本长度无法修改。

Issue - State: closed - Opened by cc2017111 over 1 year ago - 1 comment
Labels: solved

#93 - 无法输出验证集的loss

Issue - State: closed - Opened by Logosxxw over 1 year ago - 2 comments
Labels: solved

#92 - 额外定义Accelerator

Issue - State: closed - Opened by gugugu-469 over 1 year ago - 1 comment
Labels: solved

#91 - RLHF的效果是否有提升

Issue - State: closed - Opened by reborm over 1 year ago - 3 comments
Labels: solved

#90 - 想用长文本数据微调模型

Issue - State: closed - Opened by tu2022 over 1 year ago - 3 comments
Labels: invalid

#89 - 请教一下epoch和学习率的设置。

Issue - State: closed - Opened by CCzzzzzzz over 1 year ago - 1 comment
Labels: solved

#88 - 有支持DeepSpeed加速训练的计划吗？

Issue - State: closed - Opened by yt7589 over 1 year ago - 3 comments
Labels: solved

#87 - 有没有试过trlx库的代码训练ppo？

Issue - State: closed - Opened by taofennanhai over 1 year ago - 2 comments
Labels: pending

#86 - expect to support the moss model

Issue - State: closed - Opened by kxleee over 1 year ago - 1 comment
Labels: solved

#85 - 修改认知的例子对原始模型污染很严重

Issue - State: closed - Opened by podanshu over 1 year ago - 9 comments
Labels: solved

#84 - 预测/微调读取的样本数与实际不一致？

Issue - State: closed - Opened by CCzzzzzzz over 1 year ago - 2 comments
Labels: solved

#83 - 函数实现有bug（函数preprocess_function_train_pair），所以训练reward模型是错的（train_rm.py）

Issue - State: closed - Opened by fsggh over 1 year ago - 7 comments
Labels: enhancement, solved

#82 - 环境问题：报sklearn错误

Issue - State: closed - Opened by xuehui0725 over 1 year ago - 3 comments
Labels: invalid

#80 - 请问如何使用 int8 推理

Issue - State: closed - Opened by liuzhx2016 over 1 year ago - 10 comments
Labels: solved

#79 - RTX3060,量化Quantized to 8 bit断言报错

Issue - State: closed - Opened by LeXwDeX over 1 year ago - 1 comment
Labels: solved

#78 - 大佬，怎么融合lora和chatglm模型的参数的，有例子吗

Issue - State: closed - Opened by JeffWb over 1 year ago - 2 comments
Labels: solved

#77 - lora微调后的模型在jetson平台推理超时

Issue - State: closed - Opened by liaoweiguo over 1 year ago - 5 comments
Labels: solved

#76 - 请问一机多卡ptuning V2微调支持吗

Issue - State: closed - Opened by pyy1988 over 1 year ago - 1 comment
Labels: solved

#75 - Accelerate多卡Lora微调报错

Issue - State: closed - Opened by YueWen1024 over 1 year ago - 13 comments
Labels: solved

#74 - While I tried to train RLHF by deepspeeed, it happens to some problem

Issue - State: closed - Opened by OceannTwT over 1 year ago - 3 comments
Labels: solved

#73 - RLHF训练后推理的代码有吗，如何加载checkpoint

Issue - State: closed - Opened by liaoweiguo over 1 year ago - 1 comment
Labels: solved

#72 - 多卡lora训练报错

Issue - State: closed - Opened by ZTurboX over 1 year ago - 8 comments
Labels: solved

#71 - 支持全参数+8bit量化模式么？

Issue - State: closed - Opened by cc2017111 over 1 year ago - 1 comment
Labels: solved

#70 - 请问，我想使用原模型（未经微调过的模型）进行eval，该如何加载？我使用

Issue - State: closed - Opened by cc2017111 over 1 year ago - 1 comment
Labels: solved

#69 - what's the different about quantize lora?

Issue - State: closed - Opened by jiayi37u over 1 year ago - 3 comments
Labels: solved

#68 - About using acelerate when uses lora tuning with 4-v100 on one machine

Issue - State: closed - Opened by 1719930244 over 1 year ago - 2 comments
Labels: solved

#67 - 支持全参数微调么？

Issue - State: closed - Opened by cc2017111 over 1 year ago

#66 - RM数据准备的技巧

Issue - State: closed - Opened by SmileSmith over 1 year ago - 1 comment
Labels: solved

#65 - 使用8bit量化LoRA训练时，出现了如下错误，是否有人遇到过？

Issue - State: closed - Opened by jimmy-kuo over 1 year ago - 3 comments
Labels: solved

#64 - 训练中的 prompt query 怎么在推理时拼接

Issue - State: closed - Opened by liaoweiguo over 1 year ago - 3 comments
Labels: solved

#63 - 断点续训

Issue - State: closed - Opened by www516717402 over 1 year ago - 1 comment
Labels: solved

#62 - 可以支持一次加载多个数据集吗

Issue - State: closed - Opened by cangyi071 over 1 year ago - 2 comments
Labels: solved

#61 - 提供一个单独的模型合并脚本

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

#59 - 使用ads_generation的例子进行单卡微调无法复现结果

Issue - State: closed - Opened by fangd123 over 1 year ago - 5 comments
Labels: invalid

#58 - accelerate

Issue - State: closed - Opened by www516717402 over 1 year ago - 1 comment

#57 - RLHF 训练时报错

Issue - State: closed - Opened by CCzzzzzzz over 1 year ago - 1 comment
Labels: solved

#56 - lora训练完毕后参数加回原模型

Issue - State: closed - Opened by nghuyong over 1 year ago - 1 comment
Labels: solved

#55 - python3.7是否可以运行呢

Issue - State: closed - Opened by jimmy-kuo over 1 year ago - 1 comment
Labels: solved

#54 - 训练语料长度超出2048token

Issue - State: closed - Opened by clintlong over 1 year ago - 5 comments
Labels: invalid

#53 - about train speed.

Issue - State: closed - Opened by www516717402 over 1 year ago - 1 comment
Labels: solved

#52 - 全参数微调

Issue - State: closed - Opened by nghuyong over 1 year ago - 8 comments
Labels: solved

#51 - 多卡如何进行分布式部署推理

Issue - State: closed - Opened by ZTurboX over 1 year ago - 2 comments
Labels: solved

#50 - The latest code has some problems while training

Issue - State: closed - Opened by OceannTwT over 1 year ago - 1 comment
Labels: solved

#49 - Occurs some loss = nan steps, is it rational?

Issue - State: closed - Opened by OceannTwT over 1 year ago - 2 comments
Labels: solved

#48 - 支持其他llm吗

Issue - State: closed - Opened by Chan-0312 over 1 year ago - 1 comment
Labels: solved

#47 - 0505 commit

Pull Request - State: closed - Opened by janglichao over 1 year ago - 1 comment

#46 - commit docker and docker-compose yml file

Pull Request - State: closed - Opened by janglichao over 1 year ago - 1 comment

#45 - TypeError: argument of type 'NoneType' is not iterable

Issue - State: closed - Opened by WangRongsheng over 1 year ago - 5 comments
Labels: solved

#44 - Is train_rm not supported by multiple GPUs?

Issue - State: closed - Opened by OceannTwT over 1 year ago - 4 comments
Labels: solved

#43 - 我在尝试lora时报错

Issue - State: closed - Opened by WangShunzhiDQ over 1 year ago - 7 comments
Labels: solved

#42 - 第三步出现错误：RuntimeError: probability tensor contains either inf, nan or element < 0

Issue - State: closed - Opened by Lauorie over 1 year ago - 16 comments
Labels: solved

#41 - add google translate for comparison_gpt4_data_en.json

Pull Request - State: closed - Opened by zhongpei over 1 year ago - 1 comment

#40 - fix a compliance bug when enabling quant during rm

Pull Request - State: closed - Opened by codemayq over 1 year ago - 1 comment

#39 - 怎么使用int4量化版本模型进行finetune,使用quantization_bit 4 后loss 一直为0

Issue - State: closed - Opened by Resign-ljy over 1 year ago - 1 comment
Labels: solved

#37 - 训练的时候出现RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'

Issue - State: closed - Opened by jiuyuanzhimeng over 1 year ago - 1 comment
Labels: solved

#36 - 怎么不支持边训练边验证保存最佳模型再做测试？

Issue - State: closed - Opened by xujunrt over 1 year ago - 2 comments
Labels: solved

#35 - 为什么我在不联网的情况下不能lora呢？

Issue - State: closed - Opened by WangShunzhiDQ over 1 year ago - 9 comments
Labels: solved

#34 - AssertionError: No inf checks were recorded for this optimizer.

Issue - State: closed - Opened by km1994 over 1 year ago - 2 comments
Labels: solved

#33 - 为什么v100不能用lora int8量化训练呢

Issue - State: closed - Opened by ZTurboX over 1 year ago - 1 comment
Labels: solved

#32 - lora训练后如何加速预测

Issue - State: closed - Opened by ZTurboX over 1 year ago - 1 comment
Labels: solved

#31 - 請問要怎麼加入--lora_rank 8這個參數

Issue - State: closed - Opened by TimLeeGee over 1 year ago - 3 comments
Labels: solved

#30 - 用lora微调，为什么没有readme的效果

Issue - State: closed - Opened by liaoweiguo over 1 year ago - 12 comments
Labels: solved

#29 - 微信群人数超200了，无法直接加入

Issue - State: closed - Opened by imndszy over 1 year ago - 2 comments
Labels: solved

#27 - 支持多卡+lora+8bit量化训练吗？

Issue - State: closed - Opened by Dreamerzp over 1 year ago - 1 comment
Labels: in-progress

#25 - Using Accelerate train

Issue - State: closed - Opened by www516717402 over 1 year ago - 2 comments
Labels: solved

#24 - 微调时报错了，能帮我看下吗

Issue - State: closed - Opened by jeffreychen567 over 1 year ago - 2 comments
Labels: solved

#22 - 请问在多卡运行文件fine_tuning_chatglm6b.py时报错是什么原因呢？

Issue - State: closed - Opened by LebronJames0423 over 1 year ago - 10 comments
Labels: solved

#21 - Temporarily deploy the ChatGLM fintune model running on colab through cloudfare tunnel

Pull Request - State: closed - Opened by michaeloo0 over 1 year ago - 1 comment

#20 - add web_demo adapted from ChatGLM-6B

Pull Request - State: closed - Opened by codemayq over 1 year ago - 1 comment

#18 - Dataset doesn't exist.

Issue - State: closed - Opened by michaeloo0 over 1 year ago - 2 comments
Labels: solved

#17 - Ryb

Issue - State: closed - Opened by michaeloo0 over 1 year ago - 3 comments
Labels: solved

#16 - 自己内部的知识库数据，必须整成问答集吗？

Issue - State: closed - Opened by huangxd- over 1 year ago - 1 comment
Labels: solved

#15 - 找不到arguments模块

Issue - State: closed - Opened by cutelitchi over 1 year ago - 1 comment
Labels: solved

#14 - added quick-start of fine-tuning base on our framework (chinese guide）

Pull Request - State: closed - Opened by janglichao over 1 year ago - 1 comment

#13 - OOM Error when saving models trained in INT8 mode

Issue - State: closed - Opened by hiyouga over 1 year ago - 1 comment
Labels: solved

#11 - mode怎么改成int4,显存不够，只有12G

Issue - State: closed - Opened by paulqi over 1 year ago - 1 comment
Labels: solved

#10 - 哥要不出个懒人包吧，折腾了两天环境都没弄好总是报错

Issue - State: closed - Opened by shmily91 over 1 year ago - 1 comment
Labels: solved

#8 - 请问作者是用什么机器跑的呢，V100 or A100 显存需要多大呢～ V100能跑起来吗

Issue - State: closed - Opened by hangzeli08 over 1 year ago - 1 comment
Labels: solved

#7 - ValueError: ChatGLMForConditionalGeneration does not support gradient checkpointing.

Issue - State: closed - Opened by shmily91 over 1 year ago - 2 comments
Labels: solved

#6 - 如何本地加载模型

Issue - State: closed - Opened by shmily91 over 1 year ago - 1 comment
Labels: solved

#5 - 请问--checkpoint_dir是只支持一个epoch的断点么？每1000次保存的没法继续吗

Issue - State: closed - Opened by LainNetWork over 1 year ago - 1 comment
Labels: solved

#4 - 4.14最新代码，尝试跑demo报数据集sum校验异常

Issue - State: closed - Opened by LainNetWork over 1 year ago - 2 comments
Labels: solved

#3 - 大神，能出个懒人包吗？

Issue - State: closed - Opened by shmily91 over 1 year ago - 2 comments
Labels: in-progress

#2 - 为啥照着步骤来，加载dataset example，始终无法加载自定义内容

Issue - State: closed - Opened by poppysix over 1 year ago - 9 comments
Labels: solved

#1 - 关于数据集中单个example的长度

Issue - State: closed - Opened by li-aolong over 1 year ago - 1 comment
Labels: solved