issues
search
DLLXW
/
baby-llama2-chinese
用于从头预训练+SFT一个小参数量的中文LLaMa2的仓库;24G单卡即可运行得到一个具备简单中文问答能力的chat-llama2.
MIT License
2.34k
stars
288
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
没有SFT的话 推理会抱错,麻烦看看
#30
hopeforus
opened
10 months ago
2
'../track1/train_valid.json。这个文件在哪里下载?
#29
hopeforus
opened
10 months ago
2
sft dataset
#28
paopao0226
opened
10 months ago
2
Data process modifications
#27
jh01231230
closed
10 months ago
0
要训练几个epoch,会有比较好的效果?
#26
binwang672012
closed
10 months ago
4
处理百度数据集的时间报错
#25
hopeforus
opened
10 months ago
6
交个作业
#24
AClolinta
opened
10 months ago
13
数据集问题
#23
zhihui-shao
opened
10 months ago
3
sft.py运行报错 CUDA out of memory,请问咋解决?
#22
qxj
closed
10 months ago
6
您好,请问显存为24G 3090预训练这个参数量大小的模型大概需要多久呀?
#21
LePanda026
opened
10 months ago
3
可以提供一个训练好的模型吗?
#20
PeterouZh
opened
10 months ago
5
fix: remove redundant pkg
#19
jianhu-chen
closed
10 months ago
0
单卡下运行pretrain.py 报错 Default process group has not been initialized, please make sure to call init_process_group.
#18
TristanShao
opened
10 months ago
10
大家在预训练的时候有遇到过loss为nan吗
#17
ZK-Zhou
opened
10 months ago
15
Where to fetch medical_qa_144w.csv?
#16
qxj
closed
10 months ago
1
在处理百度563baike时Memory error
#15
ZK-Zhou
opened
10 months ago
5
百度云垃圾
#14
KKIverson
closed
10 months ago
2
dataset_sft.py中loss_mask的切片为什么和X一致?
#13
BigaGrayWolf
opened
10 months ago
2
Question about tokenizer
#12
IshootLaser
closed
10 months ago
1
预训练完后执行python sft.py报错找不到文件
#11
xamofb-xsk
closed
10 months ago
2
sft使用的checkpoint问题
#10
Deep1994
closed
10 months ago
1
medical_qa.bin 没有用上
#9
Deep1994
closed
10 months ago
3
CUDA_VISIBLE_DEVICES=0,1 torchrun pretrain.py 只利用了一块GPU
#8
BigaGrayWolf
closed
10 months ago
6
关于 GBK 编码的问题
#7
lavinal712
closed
10 months ago
4
上下文长度32K
#6
CanvaChen
closed
11 months ago
1
请教下参数大小如何计算
#5
CanvaChen
closed
11 months ago
1
运行预训练报错
#4
1633232731
closed
11 months ago
2
chore: add requirements.txt
#3
jiey2
closed
6 months ago
2
没有找到medical_qa_144w
#2
Yuiceee
closed
11 months ago
5
可以给个测评结果吗?
#1
linonetwo
closed
11 months ago
3
Previous