issues
search
s-JoL
/
Open-Llama
The complete training code of the open-source high-performance Llama model, including the full process from pre-training to RLHF.
https://huggingface.co/s-JoL/Open-Llama-V2
MIT License
30
stars
4
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
请教一下训练速度问题
#75
ray075hl
closed
2 months ago
0
Cannot download the checkpoints from huggingface
#74
GeorgeMarica
opened
1 year ago
0
触发了重复生成的问题
#72
Expert68
opened
1 year ago
0
模型保存问题
#71
expresschen
closed
1 year ago
0
training benchmarking question: unable to reproduce the training speed
#70
tong-zeng
opened
1 year ago
0
v2的模型使用了RLHF吗
#69
Expert68
opened
1 year ago
0
zero3 如何指定使用的卡数呢
#68
Stick-To
closed
1 year ago
0
Can not send text
#67
lucasjinreal
opened
1 year ago
0
关于多机多卡训练
#66
lemonsis
opened
1 year ago
0
Is multi-node (multiple machines) training with zero3 tested?
#65
lipiji
opened
1 year ago
0
Update README.md
#64
eltociear
closed
1 year ago
1
Exact version of OpenLlama available on home.ustc.edu.cn
#63
Kaelorn
opened
1 year ago
0
Question about RLHF
#62
younesbelkada
opened
1 year ago
0
请问一下指令微调（instruct finetune）可以使用32GB显存的v100吗？
#61
honglianglv
closed
1 year ago
1
training speed questions: multi-node training, CPU offload
#60
jli
opened
1 year ago
2
继续预训练的问题
#59
HunterKruger
closed
1 year ago
2
Thailand example in readme.md is a pure hallucination
#56
mattkanwisher
opened
1 year ago
1
Can this train regular llama models or just openllama?
#55
teknium1
closed
1 year ago
2
基于v2的预训练模型进行指令微调时，LOSS总是忽大忽小，波动范围较大，不知道是否是正常情况？
#54
honglianglv
closed
1 year ago
3
v2的checkpoint好像是pretrain模型，不是instruct tuning后的模型
#53
Expert68
closed
1 year ago
7
Open-Llama-V2-pretrain使用train_lm.py进行Instruction-Tuning报错：RuntimeError: CUDA error: CUBLAS_STATUS_NOT_SUPPORTED when calling `cublasGemmEx
#52
ma0722
closed
1 year ago
0
关于V2版代码数据处理的两个问题
#51
xikaluo
closed
1 year ago
1
基于预训练模型在自有数据进行instruct-tuning效果问题
#49
ScottishFold007
closed
1 year ago
11
中文预训练模型的精度有评测吗？
#48
puppet101
opened
1 year ago
2
train_lm.py一直waiting是为什么？
#47
leiwen83
closed
1 year ago
2
33B和65B模型训练
#46
Stick-To
closed
1 year ago
7
Released model on demo
#45
Manuel030
closed
1 year ago
6
有计划放出13B及以上的模型么？
#44
laoda513
closed
1 year ago
1
zero3 如何训练65b模型
#43
Stick-To
closed
8 months ago
2
wudao数据集的下载预处理脚本问题
#42
skepsun
closed
1 year ago
1
关于83200.pt？
#41
ScottishFold007
closed
1 year ago
4
做微调时，保存权重出了问题
#40
ScottishFold007
closed
1 year ago
2
如何将预训练好的pt文件转化为bin格式的文件？
#39
ScottishFold007
closed
1 year ago
1
Is zero2&zero3 with cpuoffload tested?
#38
lipiji
closed
1 year ago
2
instruct finetune报错，初始化的模型和加载的83200.pt好像结构上对不齐
#37
lierer007
closed
1 year ago
3
v2版训练代码在forward阶段报错
#36
xikaluo
closed
1 year ago
2
关于v2
#35
leiwen83
closed
1 year ago
5
what ‘s difference between 10w_vocab_wudao5_pile10.model and llama_tokenizer_extended.model
#33
wac81
closed
1 year ago
3
where is model implement？
#32
wac81
closed
1 year ago
2
How to do V2.0 pre-training?
#31
mikeda100
closed
1 year ago
1
update readme
#30
s-JoL
closed
1 year ago
0
update readme
#29
s-JoL
closed
1 year ago
0
关于预训练中的eval阶段
#27
xikaluo
closed
1 year ago
10
显存问题
#26
cxj01
closed
1 year ago
1
v2 release
#25
s-JoL
closed
1 year ago
0
可以提供一个dockerfile吗？
#24
leiwen83
closed
1 year ago
9
关于开源协议
#23
MrToy
closed
1 year ago
1
关于训练开销
#22
leiwen83
opened
1 year ago
28
instruction tuning失败
#21
leiwen83
closed
1 year ago
2
dataset download fail
#20
leiwen83
closed
1 year ago
1
Next