issues
search
Facico
/
Chinese-Vicuna
Chinese-Vicuna: A Chinese Instruction-following LLaMA-based Model —— 一个中文低资源的llama+lora方案，结构参考alpaca
https://github.com/Facico/Chinese-Vicuna
Apache License 2.0
4.14k
stars
421
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
用finetune_others_continue.sh自己训练的数据，行4-bit量化后在llama回答不了，其他的问题也比较乱
#157
qqshuqain
closed
1 year ago
3
RuntimeError: Error(s) in loading state_dict for PeftModelForCausalLM: size mismatch for base_model.model.model.layers.0
#156
zhiyixu
closed
1 year ago
3
requeirements中给的https翻墙也会出现404问题
#155
153zml
closed
1 year ago
5
构造古诗词（65263首）单指令语料直接使用finetune从头开始训练
#154
grantchenhuarong
closed
1 year ago
3
使用需要遵循的协议
#153
yuanjing-jane
closed
1 year ago
1
关于古诗词（347首）训练的后继问题
#152
grantchenhuarong
closed
1 year ago
3
使用定时器方式监测断点执行finetune
#150
grantchenhuarong
closed
1 year ago
2
checkpoint-11600这个在哪里下载
#149
jialongguo
closed
1 year ago
0
instruct_chat_50k.jsonl 数据
#148
ethan-funny
closed
1 year ago
0
cuda out of memory in saving steps.
#147
jzsbioinfo
closed
1 year ago
2
请问用4090单卡是否可以lora微调13b模型？
#146
mircop1t
closed
1 year ago
7
finetune.py中关于选择lora模型的代码段逻辑有些问题
#145
grantchenhuarong
closed
1 year ago
0
finetune_others_continue.sh 报AttributeError: 'NoneType' object has no attribute 'print_trainable_parameters'
#144
qqshuqain
closed
1 year ago
4
使用您提供的simple_test.py脚本测试，发现token第一个起始字符id是0，识别也是乱码
#143
leizhu1989
closed
1 year ago
15
我用这个模型发现已经支持中文，请问是他自己实现的还是组合了本模型
#142
mushuanli
closed
1 year ago
1
python finetune.py --data_path ./sample/merge_sample.json --test_size 9 训练报错
#141
jackywei1228
closed
1 year ago
5
微调之后加载权重发现输出停不下来
#140
wilson9x1
closed
1 year ago
11
使用Lora和不使用Lora速度会差很多
#139
cxj01
closed
1 year ago
3
效果炸裂: Langchain + vicuna-13b + 本地知识库
#138
csunny
opened
1 year ago
2
guanaco_belle_merge_v1.0 直接merge可能重复的情况？
#137
wilson9x1
closed
1 year ago
1
训练的3.0轮是够的么？
#136
wilson9x1
closed
1 year ago
1
提示 ：ERROR: Could not find a version that satisfies the requirement triton==2.0.0 (from versions: none) 是什么情况？
#135
davedat
closed
1 year ago
2
README.md mis-inforamtion: `LLAMA13B is numerically superior to 175B GPT3）` -> 17.5B
#134
huide9
closed
1 year ago
2
希望作者提供 Chinese-Vicuna/Chinese-Vicuna-lora-13b-belle-and-guanaco 的训练过程的优化器参数
#133
greatewei
closed
1 year ago
7
interaction.py 第158行，返回了两次history
#132
Rogerspy
closed
1 year ago
5
If you tried to load a PyTorch model from a TF 2.0 checkpoint, please set from_tf=True
#131
longkeyy
closed
1 year ago
3
流式输出结果和最终结果长度/内容相差较大
#130
Aiemu
closed
1 year ago
1
How to run vicuna on a CPU with 16GB of RAM?
#129
DimIsaev
closed
1 year ago
0
if it is possible to fine tune your model
#128
TY-hue123
closed
1 year ago
3
我构建了一个在共享GPU机器上也能提供公开服务测试的接口
#127
youkpan
closed
1 year ago
1
bash generate.sh FAILED with AttributeError: 'NoneType' object has no attribute 'eval'
#126
SeekPoint
closed
1 year ago
1
what the pytorch version you used ?
#125
SeekPoint
closed
1 year ago
1
OSError: Can't load tokenizer for './llama-hf/llama-7b/'
#124
adaaaaaa
closed
1 year ago
2
bash finetune_continue.sh failed with 'RuntimeError: Trainer requires either a model or model_init argument'
#123
SeekPoint
opened
1 year ago
2
用7B的模型，以及官方checkpoint-final运行generate.sh，效果很差
#122
dizhenx
closed
1 year ago
4
llama 7b lora微调效果比展示的差很多。https://github.com/Facico/Chinese-Vicuna/blob/master/docs/performance.md
#121
jzsbioinfo
closed
1 year ago
4
两张3090多卡训练，运行finetune.sh后无任何报错，但是程序直接中断
#120
JupyterChu
closed
1 year ago
2
这是什么情况，后面的回答都是回答的第一个问题？
#119
lichao4Java
closed
1 year ago
2
请问13b模型具体是怎么微调的？用的什么参数和数据？还是直接用的finetune.sh么？
#118
jzsbioinfo
closed
1 year ago
1
The tokenizer class you load from this checkpoint is not the same type as the class this function is called from. It may result in unexpected tokenization.
#117
dizhenx
closed
1 year ago
1
May I ask how much training data you have for each language?
#116
god-ts
closed
1 year ago
1
使用checkpoint-11600训练效果出不来，求指教。
#115
grantchenhuarong
closed
1 year ago
31
纯CPP 推理，make chat 失败
#114
shikeno
closed
1 year ago
3
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
#113
SHITIANYU-hue
closed
1 year ago
1
更新代码后，重新执行finetune.sh出错， TypeError: init_process_group() got multiple values for keyword argument 'backend'
#112
alisyzhu
opened
1 year ago
7
运行generate.sh后一直报model为NoneType
#111
JupyterChu
closed
1 year ago
2
Something went wrong Expecting value: line 1 column 1 (char 0)
#110
FantDing
closed
1 year ago
2
The server console prints an error message when I submit a request from the browser
#109
ImGoodBai
closed
1 year ago
2
请问这个的效果和vicuna相比怎么样
#108
rayvzn119
closed
1 year ago
6
需要最低配置是什么？
#107
dravinbox
opened
1 year ago
4
Previous
Next