- 11 4月, 2023 6 次提交
-
-
liang tongtong authored
-
liang tongtong authored
-
1、自动读取knowledge_based_chatglm.py中LLM及embedding模型枚举,选择后点击setting进行模型加载,可随时切换模型进行测试 2、可手动调节保留对话历史长度,可根据显存大小自行调节 3、添加上传文件功能,通过下拉框选择已上传的文件,点击loading加载文件,过程中可随时更换加载的文件 4、底部添加use via API可对接到自己系统 TODO: 1、添加模型加载进度条 2、添加输出内容及错误提示 3、国家化语言切换 4、引用标注 5、添加插件系统(可基础lora训练等)
liang tongtong authored -
imClumsyPanda authored
-
Update README
imClumsyPanda authored -
About how to solve "Resource not found".
Calcitem authored
-
- 10 4月, 2023 3 次提交
-
-
imClumsyPanda authored
-
imClumsyPanda authored
-
imClumsyPanda authored
-
- 09 4月, 2023 7 次提交
-
-
imClumsyPanda authored
-
imClumsyPanda authored
-
imClumsyPanda authored
-
imClumsyPanda authored
-
imClumsyPanda authored
-
Update README.md
imClumsyPanda authored -
Calcitem authored
-
- 08 4月, 2023 3 次提交
-
-
imClumsyPanda authored
-
imClumsyPanda authored
-
imClumsyPanda authored
-
- 07 4月, 2023 16 次提交
-
-
imClumsyPanda authored
-
imClumsyPanda authored
-
根据4.7日更新代码,解决冲突
imClumsyPanda authored -
新增ernie系列模型作为embedding model,默认依旧使用GanymedeNil/text2vec-large-chinese 显存不足时可尝试使用ernie-tiny作为embedding模型 调用方式为embeddings = HuggingFaceEmbeddings(model_name=embedding_model_dict["ernie-tiny"], )
zhangwei authored -
imClumsyPanda authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
littlepanda0716 authored
-
fix: 修复chatglm模型被复制,显存占用过多
imClumsyPanda authored -
littlepanda0716 authored
-
- 05 4月, 2023 3 次提交
-
-
Update requirements.txt
imClumsyPanda authored -
thomas-yanxin authored
-
model作为类成员会在类实例化时进行一次复制 这导致每询问一个问题显存占用就会翻倍 通过将model改成全局变量修复这个问题
myml authored
-
- 03 4月, 2023 2 次提交
-
-
littlepanda0716 authored
-
littlepanda0716 authored
-