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
Chinese-Vicuna-lora-13b-belle-and-guanaco 如何进行 finetune_continue 训练，我希望能基于13b继续训练新的数据
#106
greatewei
closed
1 year ago
8
Chinese-Vicuna-medical 可以直接使用吗？如何直接使用
#105
rdaim
closed
1 year ago
3
chat.py 执行失败
#104
timiil
closed
1 year ago
2
关于Chinese-Vicuna-medical的continuous-finetune疑问
#103
yuelinan
closed
1 year ago
3
Chinese-Vicuna-lora-7b-belle-and-guanaco是基于merge.json的数据训练的吗？
#102
greatewei
closed
1 year ago
1
运行generate.py没有反应
#101
Data2Me
closed
1 year ago
1
这个lora 对应的大模型是 llama 还是 Vicuna?
#100
Lufffya
closed
1 year ago
1
构建dataset时，将instruction部分的label设置为-100可能是无效的
#99
zhengyanzhao1997
closed
1 year ago
8
关于训练中途意外停止的问题
#98
Tian14267
closed
1 year ago
16
generate_quant.py脚本测试13b量化模型，效果很差，如图：
#97
greatewei
closed
1 year ago
2
关于多卡训练的问题
#96
Tian14267
closed
1 year ago
4
Chinese-Vicuna checkpoint-11600的中英文对照效果
#95
grantchenhuarong
closed
1 year ago
5
使用merge_sample.jsonw做了例子简单训练，生成的checkpoints效果检验不理想
#94
grantchenhuarong
closed
1 year ago
8
请问 continuous-finetune 的实现原理是什么？是语料的不断merge还是合并了各个批次的LoRa Adapter？
#93
valkryhx
closed
1 year ago
3
基于merge.json训练了3轮，效果较checkpoint-final差很多，没有改动参数，求指导。
#92
xienan0326
closed
1 year ago
10
在2080ti上运行 finetune提示错误
#91
grantchenhuarong
closed
1 year ago
7
请教一下stream_output 优势是什么
#90
xienan0326
closed
1 year ago
1
Does it or will it support Vicuna-13b-v1.1 finetuning?
#89
ghost
closed
1 year ago
2
TypeError: init_process_group() got multiple values for keyword argument 'backend' 使用torchrun会报这个错，V100，32G，2卡训练，执行finetune.sh 不起来，一直报着个错
#88
hangzeli08
closed
1 year ago
4
Chinese-Vicuna-lora-7b-3epoch-belle-and-guanaco这个模型是怎么训练出来的
#87
hyb1234hi
closed
1 year ago
6
想知道在推理的时候，Temperature、Beams Number 和 Repetiton Penalty等参数有什么意义，该怎么调整？
#86
AIXiaoBaiDemon
closed
1 year ago
3
RuntimeError: shape '[-1, 32001]' is invalid for input of size 32640000
#85
molyswu
closed
1 year ago
21
长度256
#84
zh25714
closed
1 year ago
3
NotImplementedError: Cannot copy out of meta tensor; no data!
#83
greatewei
closed
1 year ago
5
微调自己的数据文件报错
#82
xiaoyi001yeye
closed
1 year ago
1
model saving error
#81
imrankh46
closed
1 year ago
10
无法安装git+https://github.com/huggingface/peft@e536616888d51b453ed354a6f1e243fecb02ea08
#80
yeshcue
closed
1 year ago
4
小白求解，关于torch库
#79
JImmyHui2017
closed
1 year ago
1
is LLama tokenizer support Chinese?
#78
abdoelsayed2016
closed
1 year ago
1
使用sample/merge_sample.json训练的模型output会带出下一句
#77
Albort-z
closed
1 year ago
5
code refactor for argparse and utils
#76
HUGHNew
closed
1 year ago
0
interaction.py 是加了什么限制吗？当进行描述性对话时，会卡住，一直无反应
#75
ZenXir
closed
1 year ago
1
运行了generate.py，出现以下错误，似乎是bitsandbytes 的问题，找到了许多其他项目的类似问题，但是我并没有找到合适的解决方案
#74
GreatWildFire
closed
1 year ago
5
训练到中途：torch.distributed.elastic.multiprocessing.api.SignalException: Process 17871 got signal: 1
#73
Tian14267
closed
1 year ago
2
更新到最新 interaction 没有正常加载基础模型 推理也有问题
#72
ZenXir
closed
1 year ago
12
关于generate生成的结果的问题
#71
Tian14267
closed
1 year ago
13
generate: AttributeError: 'NoneType' object has no attribute 'eval'
#70
Tian14267
closed
1 year ago
11
新手小白在线求入门
#69
ws1957
closed
1 year ago
1
使用垂直领域数据集进行断点训练后的几个问题
#68
Sowhat007
closed
1 year ago
4
数据集无法下载
#67
Pyjacc
closed
1 year ago
1
chat.py生成结果的时候，GPU的显存会持续增加，最后显存溢出
#66
yuxuan2015
closed
1 year ago
22
运行finetune_continue.sh，日志显示很多权重的key都missing了，没load进去
#65
Ulysses0817
closed
1 year ago
2
bash chat.sh报错，看有人遇到过
#64
xiaoaidafu
opened
1 year ago
7
params.json cannot be found in downloaded huggingface 3epoch model path
#63
Modas-Li
closed
1 year ago
3
不能生成performance里面的结果
#62
greedyint
closed
1 year ago
3
How to alter local url address,thx
#61
Modas-Li
closed
1 year ago
1
finetune.py: error: unrecognized arguments: --OUTPUT_PATH /data/Chinese-Vicuna/to/ --MODEL_PATH /data/Chinese-Vicuna/to/to/llama-7b-hf/
#60
Modas-Li
closed
1 year ago
4
generate和interaction都无法停止，直到达到max_tokens限制才会停止
#59
alisyzhu
opened
1 year ago
25
peft版本问题
#58
cc-doughnut
closed
1 year ago
1
关于使用纯C++推理问题
#57
BUPTccy
closed
1 year ago
5
Previous
Next