issues
search
ymcui
/
Chinese-LLaMA-Alpaca
中文LLaMA&Alpaca大语言模型+本地CPU/GPU训练部署 (Chinese LLaMA & Alpaca LLMs)
https://github.com/ymcui/Chinese-LLaMA-Alpaca/wiki
Apache License 2.0
18.23k
stars
1.86k
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Resuming pretraining from checkpoint
#827
lathashree01
closed
1 year ago
2
请问如何指定哪一张卡去做推理?
#826
bigmover
closed
1 year ago
6
RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling cublasGemmEx
#825
LiemLin
closed
1 year ago
0
用CPU加载模型做推理时报错
#824
NoobPythoner
closed
1 year ago
2
when I do sft using run_clm_sft_with_peft.py, the eval loss is always Nan, can anyone fine a solution ?
#823
trannhatquy
closed
1 year ago
3
resize token embedding的时候会tied input output embedding吗
#822
NonvolatileMemory
closed
1 year ago
3
麻烦问下,不同模型用alpaca_data_zh_51k数据集微调后,如何测评在alpaca_data_zh_51k数据集上的效果?
#821
StevensPrime
closed
1 year ago
3
后续会有关于原始模型 llama-2-7b-chat-hf 的中文llama和中文alpaca的版本吗
#820
Advancejc
closed
1 year ago
2
ValueError: optimizer got an empty parameter list. resume_from_checkpoint
#819
belle9217
closed
1 year ago
2
卡在Loading checkpoint shards: 100%,不报错
#818
H-Justus
closed
1 year ago
2
是否有基于Llama-2的增量训练模型?
#817
ymcui
opened
1 year ago
0
模型合并之后sha256值不对;使用Transformers推理时,回答完问题之后,重复
#816
Zxlan
closed
1 year ago
3
run_clm_pt_with_peft.py 33B做lora的预训练,需要多少G的显存啊?
#815
wanghao-007
closed
1 year ago
3
预训练从1000 steps断点处加载模型参数报错
#814
zzzhaoguziji
closed
1 year ago
5
二次预训练模型选择和灾难性遗忘问题
#813
Zhou-Zoey
closed
1 year ago
0
关于本项目33B模型使用Deepspeed ZeRO3进行训练的求助
#812
xuexidi
closed
1 year ago
6
为啥自己训练的Chinese-Alpaca-Pro-33B的SFT的LoRA权重只有1.3G大小?本项目里是2.1G。
#811
xuexidi
closed
1 year ago
6
33B不量化预训练需要多大显存的GPU,量化后预训练要多大显存的GPU
#810
yaochao1
closed
1 year ago
3
ceval中“A”encode后的sA_id与“:A”encode后的A_id有何不同呢?各代表什么含义。Debug后id确实不一样
#809
lizhzh8
closed
1 year ago
3
torch.distributed.elastic.multiprocessing.errors.ChildFailedError
#808
MSamiee
closed
1 year ago
2
3090 cudamemory 不够用25GB显存
#807
illumionous
closed
1 year ago
3
请问增量预训练数据量有多少?
#806
peiyingxin
closed
1 year ago
2
请问Alpaca词表(49954)是怎么生成的?详细步骤是什么?
#805
jackielyc
closed
1 year ago
5
Bugfix: Fixed spelling error: tokenzier_vocab_size --> tokenizer_voca…
#804
thelongestusernameofall
opened
1 year ago
0
Expand vocabulary for other language from scratch
#802
MSamiee
closed
1 year ago
4
为什么在Plus版本的sft中改变了Lora Rank?
#800
nuoma
closed
1 year ago
3
关于pretrain阶段的一些疑问
#799
chenhk-chn
closed
1 year ago
5
请问 [模型选择指引] 中文LlaMA的已知问题是否还存在?Chinese-LlaMA-13B 如何优雅的控制终止及断句?
#798
TotalDom1Nation
closed
1 year ago
3
run_pt.sh output couldnt be merged with base model
#797
yusufcakmakk
closed
1 year ago
8
</s>为什么在预训练阶段没有使用?
#796
weilx2267
closed
1 year ago
3
为什么Alpaca新增一个pad token
#795
caihaunqai
closed
1 year ago
4
chinese pro 7b一旦输入中文,推理程序自动关闭
#794
ibmxiang
closed
1 year ago
2
Why combined INSTRUCTION + INPUT + TARGET as INPUT for training ?!?!?
#793
thusinh1969
closed
1 year ago
2
预训练后的权重尺寸问题
#792
Tramac
closed
1 year ago
2
预训练最大长度扩展到5k
#789
chensongcan
closed
1 year ago
6
13Balpaca-plus和pro微调完后,均在首次提问推理回答完成后间隔30S左右时间才能结束此轮对话,进行下一次提问
#788
DRZJ1
closed
1 year ago
2
预训练 多GPU训练时LoRA保存有误
#787
Phinease
closed
1 year ago
2
微调阶段数据配比
#786
caihaunqai
closed
1 year ago
2
gradient_checkpointing 参数有何影响?
#784
Tramac
closed
1 year ago
6
Streaming openai api support
#783
yunhaoli24
opened
1 year ago
4
lora训练LLAMA PLUS的时候,爆显存,我现在是4张A10,一共是90G.怎么量化,或者能否微调
#782
DRZJ1
closed
1 year ago
6
Chinese Alpaca Plus 13B预训练时Embed层无法正常训练与合并
#781
Phinease
closed
1 year ago
4
建议添加可配置的内容审核
#780
zjzjzjzj1874
closed
1 year ago
2
Llama 2出来了请也帮忙支持下
#779
gimlee
closed
1 year ago
1
33bplus模型用llama.cpp进行中文对话回复异常
#778
mafamily2496
closed
1 year ago
11
ChatBot Arena访问不了了
#777
kitty-eu-org
closed
1 year ago
3
使用langchain检索式问答输入问题后报错 电脑带有A100显卡
#775
ai499
closed
1 year ago
3
请教模型大小和gpu显存之间的关系
#774
TCHSDUFH
closed
1 year ago
3
单机多卡训练错误
#773
Geministudents
closed
1 year ago
8
plus和pro模型的区别在什么地方呢?
#769
minlik
closed
1 year ago
9
Previous
Next