Giters
mymusise
/
ChatGLM-Tuning
基于ChatGLM-6B + LoRA的Fintune方案
Geek Repo:
Geek Repo
Github PK Tool:
Github PK Tool
Stargazers:
3707
Watchers:
31
Issues:
247
Forks:
444
mymusise/ChatGLM-Tuning Issues
咨询一下,训练后的模型,怎么在Chatglm-6b上使用?
Updated
2 days ago
Comments count
2
小白,求大神解答,ImportError: cannot import name 'soft_unicode' from 'markupsafe
Closed
3 months ago
Comments count
1
关于保存的adapter_model.bin无实际推理效果的问题
Updated
4 months ago
Comments count
2
基于3af1bfd提交在3090上跑起来的requirements.txt
Updated
4 months ago
根据infer的代码,lora微调之后的answer和###answer结果一致,没有任何变化
Updated
4 months ago
Comments count
2
在colab上运行finetune.ipynb的时候会报一个huggingface登录的错误,有人遇到同样的错误吗?
Updated
5 months ago
Comments count
1
请问在训练过程中输出的日志中loss、learning_rate和epoch分别代表什么含义
Updated
5 months ago
请问这个项目支持chatglm3吗
Updated
6 months ago
[数据预处理-tokenization时报错] datasets.builder.DatasetGenerationError
Updated
7 months ago
AttributeError: 'ChatGLMModel' object has no attribute 'lm_head'
Updated
7 months ago
Comments count
3
问题请教:将prompt token设置为-100即可不计算loss
Updated
7 months ago
请问大佬什么时候能更新支持chatglm2呢?
Updated
8 months ago
Comments count
13
error:raise DatasetGenerationError("An error occured while generating the dataset)
Updated
8 months ago
Comments count
1
问题请教
Updated
8 months ago
请教一个问题,data_collator中不需要实现attention mask么?
Closed
10 months ago
Comments count
2
data_collator有问题?
Updated
10 months ago
Comments count
2
使用默认代码在单卡上进行推理的时候为什么其他显卡上也会有进程(占用)
Updated
10 months ago
Comments count
4
尝试在自己的数据集上使用LORA微调 但是报错
Updated
10 months ago
Comments count
1
请问如何读取checkpoint继续训练?
Updated
10 months ago
Comments count
1
训练后加载模型好像没有效果,这是什么情况?
Updated
10 months ago
Comments count
5
微调后的模型如何加载运行?用官方的web_demo跑起来似乎有问题
Updated
a year ago
Comments count
1
这个项目停更了吗
Updated
a year ago
finetune数据使用data_collator时报错 KeyError:seq_len
Closed
a year ago
Comments count
2
修改max_seq_length好像并没有生效?
Closed
a year ago
Comments count
1
请问下如果想让模型学到某个领域的数据集,大概需要多大的数据量呢?
Updated
a year ago
训练结果改变不大,用infer代码能知道是加载了正确答案。但是永远不是正确答案。怎么就能让他回答正确答案呢。
Updated
a year ago
Comments count
1
ChatGLM LoRA微调之后,量化quantize=8显存、推理耗时都反向增加
Updated
a year ago
Comments count
1
model.is_parallelizable = True,model.model_parallel = True,这两句的意思是 就可以多卡load吗?
Closed
a year ago
Comments count
1
微调语料格式转换出现乱码
Closed
a year ago
Comments count
1
请问大佬是否有计划可以支持下qlora?
Updated
a year ago
Comments count
1
如何支持多卡跑
Updated
a year ago
怎么做evaluate,计算bleu 和rougue值之类的
Updated
a year ago
Lora微调chatglm-6b后chekpoint里面缺乏adapter_config.json文件
Updated
a year ago
Comments count
1
tokenize_dataset_rows.py运行报错
Closed
a year ago
Comments count
2
好像是bitsandbytes的错
Updated
a year ago
Comments count
1
启用fp16, load_in_8bit设置为False, 报错: RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!
Updated
a year ago
Comments count
1
有没有用macos来finetune的啊
Updated
a year ago
训练数据一点也没拟合,预测结果跟chatglm基础模型一模一样
Updated
a year ago
Comments count
1
微调的时候可以使用负样本吗?
Updated
a year ago
咨询一下,支持多显卡么
Closed
a year ago
Comments count
1
怎样强化模型在某些方面的能力?
Updated
a year ago
Comments count
1
examples/infer_pretrain.ipynb 执行报错
Updated
a year ago
Comments count
1
微调完多卡推理时报精度不对的问题 expected scalar type Half but found Float ,单卡推理就没有这个问题
Updated
a year ago
Comments count
1
用cpu推理的情况下越大的ram可以支持越长的记忆吗
Updated
a year ago
期待支持RLHF
Updated
a year ago
如何做batchsize>1的infer
Updated
a year ago
No model named 'torch._six'
Updated
a year ago
能用中文训练吗?
Updated
a year ago
ValueError: Input None is not valid. Should be a string, a list/tuple of strings or a list/tuple of integers.
Updated
a year ago
infer报错,RuntimeError: mixed dtype (CPU): expect input to have scalar type of BFloat16
Updated
a year ago
Comments count
1
Previous
Next