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
425
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
scripts中不同训练脚本的差异与训练效果的问题
#207
FURYFOR
opened
1 year ago
2
多轮对话导致OOM
#206
hongshuo-wang
opened
1 year ago
1
finetune_deepspeed使用load_in_8bit后,出现 RuntimeError(f"expected there to be only one unique element in {items}")
#205
ryan-gz
opened
1 year ago
1
执行sh chat_7B.sh 的问题,推理结果明显错误
#204
niuhuluzhihao
opened
1 year ago
1
如何部署本地api化的模型而不是使用gradio 的界面
#203
zhiyixu
closed
1 year ago
1
使用finetune_chat.py 微调13B 模型,显存爆了,微调13B需要多少显存
#202
tanglaoya321
opened
1 year ago
6
至死都无法学到三国相关的内容
#201
lucasjinreal
opened
1 year ago
3
关于 finetune_others_continue.sh 的参数问题
#200
Lufffya
closed
1 year ago
1
peft模型无法加载
#199
lucasjinreal
opened
1 year ago
2
结果不是很好
#198
lucasjinreal
opened
1 year ago
8
考虑将BELLE的3.5M 数据加入chat数据吗?
#197
lucasjinreal
opened
1 year ago
3
您好,关于源码中相关py文件的请教
#196
niuhuluzhihao
closed
1 year ago
1
训练数据中 input 是不是对应其他数据格式中的history, 有示例吗
#195
babyyanxi
closed
1 year ago
1
instruct_chat_50k.json 数据问题
#194
alphanlp
opened
1 year ago
3
关于chat训练效果与语料关联不上的疑问
#193
grantchenhuarong
closed
1 year ago
7
为啥Lora模型体积相差如此巨大
#192
lucasjinreal
opened
1 year ago
1
WSL2 docker: NotImplementedError: Cannot copy out of meta tensor; no data!
#191
thusinh1969
closed
1 year ago
3
finetune_chat 运行错误
#190
lucasjinreal
opened
1 year ago
9
请更新代码,LlamaTokenizer目前根本无法加载deepra的llama的tokenizer
#189
lucasjinreal
closed
1 year ago
0
模型性能差
#188
ShouyangDong
opened
1 year ago
4
关于数据集的准备和处理
#187
kostya-sec
closed
1 year ago
2
Create devcontainer.json
#186
XQM1025768945
opened
1 year ago
1
ValueError: Can't find 'adapter_config.json' at './lora-Vicuna/checkpoint-final'
#185
adaaaaaa
closed
1 year ago
2
自建数据集及sample\instruct\data_sample.jsonl的问题
#184
LKk8563
closed
1 year ago
1
generate.py 启动的 127.0.0.1怎么调整成0.0.0.0
#183
alanbeen
closed
1 year ago
0
请问从哪里下载你们最新的模型,之前的checkpoint链接404了
#182
Lufffya
closed
1 year ago
3
为什么我训练的lora无论是启动、还是推理的速度都很慢
#181
greatewei
closed
1 year ago
0
关于运行finetune_deepspeed保存的中间checkpoint的问题,
#180
kleinchueng
opened
1 year ago
0
prompt的这些内容是否是多余的,占用了token空间?
#179
apachemycat
closed
1 year ago
1
你好,训练是正常的,但是一到模型保存的step就会显存爆炸,很奇怪,保存模型为啥会额外占用显存呢
#178
wangfudong65
closed
1 year ago
3
Update prompt.py by correcting the speeling error of maxlen
#177
Facico
closed
1 year ago
0
OutOfMemoryError: CUDA out of memory. Tried to allocate 86.00 MiB (GPU 0; 11.00 GiB total capacity; 10.22 GiB already allocated; 0 bytes free; 10.26 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
#176
superjhung
closed
1 year ago
1
可能是一个参数拼写错误?
#175
apachemycat
closed
1 year ago
1
merge.json无法下载
#174
topshu
closed
1 year ago
1
colab上执行:
#173
topshu
closed
1 year ago
1
Where is the 'merge.json' file
#172
superjhung
closed
1 year ago
1
finetune_chat.py 运行报错,IndexError: string index out of range
#171
reverse-2020
closed
1 year ago
1
finetune_chat.py为什么要限制MICRO_BATCH_SIZE和GRADIENT_ACCUMULATION_STEPS呀?
#170
grantchenhuarong
closed
1 year ago
1
有用A10G GPU finetune Vicuna-13B的吗?
#169
wangrui6
opened
1 year ago
0
ValueError: Attempting to unscale FP16 gradients.
#168
taofennanhai
closed
1 year ago
1
请问是否有计划在Vicuna-1.1上进行训练并发布模型?
#167
aresa7796
opened
1 year ago
0
finetune_deepspeed.py A100 80G 单卡跑不起来,显存不足
#166
greatewei
closed
1 year ago
8
请问可以用 fastchat 中的 vicuna 替代 llama 进行微调吗?
#165
Jingsong-Yan
closed
1 year ago
3
7b推理问题
#164
imjking
closed
1 year ago
1
开炉炼丹6.5万首诗词问答的疑问,怎样确定语料数量与训练数量的关系?
#163
grantchenhuarong
closed
1 year ago
7
运行 generate.py的时候报错,cudnn 已经安装了啊?
#162
hopeforus
closed
1 year ago
3
13B推理问题
#161
imjking
opened
1 year ago
4
web page推理无法结束
#160
mavisyyc
opened
1 year ago
5
训练时遇到报错,Some modules are dispatched on the CPU or the disk. Make sure you have enough GPU RAM to fit the quantized model. If you want to dispatch the model on the CPU or the disk while keeping these modules in 32-bit, you need to set `load_in_8bit_fp32_cpu_offload=True` and pass a custom `device_map` to `from_pretrained`. Check https://huggingface.co/docs/transformers/main/en/main_classes/quantization#offload-between-cpu-and-gpu for more details.
#159
wavetao2010
closed
1 year ago
1
fix error url in readme-zh.md
#158
timczm
opened
1 year ago
0
Previous
Next