b51ba11f45
支持通过配置项同时启动多个模型，将Wiki纳入samples知识库 ( #2002 )
...
新功能：
- 将 LLM_MODEL 配置项改为 LLM_MODELS 列表，同时启动多个模型
- 将 wiki 纳入 samples 知识库
依赖变化：
- 指定 streamlit~=1.27.0。1.26.0会报rerun错误，1.28.0会有无限刷新错误
修复优化：
- 优化 get_default_llm_model 逻辑
- 适配 Qwen 在线 API 做 Embeddings 时最大 25 行的限制
- 列出知识库磁盘文件时跳过 . 开头的文件
2023-11-09 22:15:52 +08:00
8d0f8a5d67
fix: langchain warnings for import from root
2023-09-27 21:55:02 +08:00
ec85cd1954
move wrap_done & get_ChatOpenAI from server.chat.utils to server.utils ( #1506 )
2023-09-17 16:19:50 +08:00
f7c73b842a
优化configs ( #1474 )
...
* remove llm_model_dict
* optimize configs
* fix get_model_path
* 更改一些默认参数，添加千帆的默认配置
* Update server_config.py.example
2023-09-15 17:52:22 +08:00
8a4d9168fa
update import pkgs and format
2023-08-10 21:26:05 +08:00
717ddf4d10
add history to chat apis
2023-08-08 23:56:27 +08:00
dcf49a59ef
v0.2.0 first commit
2023-07-27 23:22:07 +08:00
dd3617fcdf
将paddle相关loader改为动态引入，可以在不上传pdf/image知识文件的前提下使用protobuf=4.x。
...
这样可以使用最新版streamlit和chainlit。
2023-07-20 12:32:37 +08:00
33b978b5c4
update README.md
2023-07-17 21:39:10 +08:00
c5bc21781c
修改模型生成的调用方式，兼容Chain调用
...
修改模型切换的bug
2023-07-12 23:09:28 +08:00
64f22a9e53
修改默认的多卡部署方案，基本保证针对新模型也不会失败 ( #788 )
...
* 修复 bing_search.py的typo;更新model_config.py中Bing Subscription Key申请方式及注意事项
* 更新FAQ，增加了[Errno 110] Connection timed out的原因与解决方案
* 修改loader.py中load_in_8bit失败的原因和详细解决方案
* update loader.py
* stream_chat_bing
* 修改stream_chat的接口，在请求体中选择knowledge_base_id;增加stream_chat_bing接口
* 优化cli_demo.py的逻辑：支持 输入提示；多输入；重新输入
* update cli_demo.py
* 按照review建议进行修改
* 修改默认的多卡部署方案，基本保证针对新模型也不会失败
---------
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
2023-07-11 20:24:49 +08:00
a5ca4bf26a
1.增加对llama-cpp模型的支持；2.增加对bloom/chatyuan/baichuan模型的支持；3. 修复多GPU部署的bug;4. 修复了moss_llm.py的bug；5. 增加对openai支持（没有api,未测试);6. 支持在多卡情况自定义设备GPU ( #664 )
...
* 修复 bing_search.py的typo;更新model_config.py中Bing Subscription Key申请方式及注意事项
* 更新FAQ，增加了[Errno 110] Connection timed out的原因与解决方案
* 修改loader.py中load_in_8bit失败的原因和详细解决方案
* update loader.py
* stream_chat_bing
* 修改stream_chat的接口，在请求体中选择knowledge_base_id;增加stream_chat_bing接口
* 优化cli_demo.py的逻辑：支持 输入提示；多输入；重新输入
* update cli_demo.py
* add bloom-3b,bloom-7b1,ggml-vicuna-13b-1.1
* 1.增加对llama-cpp模型的支持；2.增加对bloom模型的支持；3. 修复多GPU部署的bug;4. 增加对openai支持（没有api,未测试)；5.增加了llama-cpp模型部署的说明
* llama模型兼容性说明
* modified: ../configs/model_config.py
modified: ../docs/INSTALL.md
在install.md里增加对llama-cpp模型调用的说明
* 修改llama_llm.py以适应llama-cpp模型
* 完成llama-cpp模型的支持；
* make fastchat and openapi compatiable
* 1. 修复/增加对chatyuan,bloom,baichuan-7等模型的支持；2. 修复了moss_llm.py的bug;
* set default model be chatglm-6b
* 在多卡情况下也支持自定义GPU设备
---------
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
2023-07-11 19:36:50 +08:00
c52308d605
1. 修改stream_chat的接口，在请求体中选择knowledge_base_id;2. 增加stream_chat_bing接口；3. 增加了调用流式接口的方法说明；4. 优化cli_demo.py的逻辑：支持 输入提示；多输入；重新输入 ( #630 )
...
* 修复 bing_search.py的typo;更新model_config.py中Bing Subscription Key申请方式及注意事项
* 更新FAQ，增加了[Errno 110] Connection timed out的原因与解决方案
* 修改loader.py中load_in_8bit失败的原因和详细解决方案
* update loader.py
* stream_chat_bing
* 修改stream_chat的接口，在请求体中选择knowledge_base_id;增加stream_chat_bing接口
* 优化cli_demo.py的逻辑：支持 输入提示；多输入；重新输入
* update cli_demo.py
* 按照review建议进行修改
2023-07-02 21:54:12 +08:00
25b46a7b9e
标题增强 ( #631 )
...
* Add files via upload
* Update local_doc_qa.py
* Update model_config.py
* Update zh_title_enhance.py
* Add files via upload
* Update README.md
* fix bugs in MyFAISS.delete_doc
* fix:前端知识库获取失败.
* update zh_title_enhance.py
* update zh_title_enhance.py
* Update zh_title_enhance.py
* add test/textsplitter
* add test_zh_title_enhance.py
---------
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
Co-authored-by: JZF <jiangzhifeng_jzf@163.com>
Co-authored-by: fxjhello <127916299+fxjhello@users.noreply.github.com>
2023-06-18 21:45:06 +08:00
ed7c54858a
add delete knowledge base and delete files from knowledge base to webui
2023-06-15 00:21:41 +08:00
11dd2b5b84
update api.py
2023-06-14 00:35:33 +08:00
a887df1715
add knowledge_base folder and move vector_store and content inside
2023-06-13 23:54:29 +08:00
7863e0fea8
updata MyFAISS
2023-06-12 00:06:06 +08:00
f1f742ce44
add self-defined class MyFAISS
2023-06-07 23:18:47 +08:00
8b7c2e417c
add support for csv file
2023-06-07 22:14:08 +08:00
57b4f9306d
修复BUG ( #570 )
...
一、修复模型历史记录仅一条问题，修改了模型加载代码，已修复。
二、修复模型历史记录仅一条问题，修改了webui有一个地方有问题，已修复。
三、知识库单条数据入库因知识库名称缓存问题导致的BUG，也已修复。
2023-06-07 18:12:51 +08:00
b1ba2003be
update webui.py
2023-05-31 20:55:45 +08:00
88fee48d8c
use lru_cache to optimize loading of local vector store ( #496 )
...
* Add files via upload
* Update README.md
* use lru_cache to optimize loading of local vector store
---------
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
Co-authored-by: liunux <liunux@office>
2023-05-30 23:05:39 +08:00
f7e120fe56
add tree func for reading files in dirs ( #471 )
2023-05-27 22:00:43 +08:00
5f010823aa
Build an abstract dialogue model using classes and methods to represent different dialogue elements
...
Fixes # None
Before submitting
If you want to review, please refer to the quick start example provided in langchain/chains/dialogue_answering/main.py. You may need to set the openaikey and the following startup parameters: --dialogue-path: the location of the dialogue file, --embedding-model: the HuggingFaceEmbeddings model to use (defaults to GanymedeNil/text2vec-large-chinese) if not specified.
Regarding the format of the dialogue file, please refer to the following information:
```text
sun:
Has the offline model been run?
glide-the:
Yes, it has been run, but the results are not very satisfactory.
glide-the:
It lacks chat intelligence and falls far behind in terms of logic and reasoning.
sun:
Are you available for voice chat?
glide-the:
I'm considering using this offline model: https://huggingface.co/chat
glide-the:
voice chat okay.
glide-the:
You can take a look at the dev_agent branch of the langchain-chatglm project.
glide-the:
There's a dialogue model question-answering example under the agent.
sun:
Alright.
glide-the:
The specified chat record file is exported from WeChat.
```
2023-05-26 01:49:07 +08:00
c4ee36b8ac
删除 AnswerResultStream 、generate_with_callback收集器
2023-05-25 21:07:40 +08:00
bca32cae83
update local_doc_qa.py
2023-05-22 18:09:23 +08:00
c79058f064
更新 local_doc_qa.py
2023-05-22 13:14:23 +08:00
4634a578f4
local_doc_qa.py
2023-05-22 10:39:55 +08:00
9140a5e70c
local_doc_qa.py
2023-05-22 10:36:41 +08:00
037b5e4502
local_doc_qa.py
...
添加精排
2023-05-22 10:32:58 +08:00
e352a04cbb
merge master
2023-05-22 00:18:48 +08:00
9c422cc6bc
update bing_search.py
2023-05-21 22:08:38 +08:00
2681728329
update local_doc_qa.py
2023-05-20 01:54:08 +08:00
e8a37ff4c7
update loader.py
2023-05-20 01:24:35 +08:00
9f5d1afc31
fix bug : 1、历史最后一条丢失；2、上下两行连读造成歧义 ( #415 )
...
* fix _call bug
* fix
* fix bug : 上下两行连读造成歧义
2023-05-20 01:19:22 +08:00
33bbb4779e
llm_model_dict 处理了loader的一些预设行为，如加载位置，模型名称，模型处理器实例, 定义checkpoint名称和远程路径
...
loader.py: 模型重载
定义 generatorAnswer 增加 AnswerResultStream
定义generate_with_callback收集器，在每次响应时将队列数据同步到AnswerResult
requirements.txt 变更项目依赖
2023-05-18 22:54:41 +08:00
621a0fe686
update cli_demo.py
2023-05-13 11:45:57 +08:00
dd93837343
使用paddleocr实现 ( #342 )
...
* jpg and png ocr
* fix
* write docs to tmp file
* fix
* [BUGFIX] local_doc_qa.py line 172: logging have no end args. (#323 )
* image loader
* fix
* fix
* update api.py
* update api.py
* update api.py
* update README.md
* update api.py
* add pdf_loader
* fix
---------
Co-authored-by: RainGather <3255329+RainGather@users.noreply.github.com>
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
2023-05-13 08:45:17 +08:00
64406cee45
Add MOSS ( #317 )
...
* Add MOSS
* Update local_doc_qa.py
---------
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
2023-05-11 18:42:19 +08:00
5852b4c62e
update pdf read strategy
2023-05-11 10:54:45 +08:00
fbdc62d910
merge master
2023-05-10 17:59:28 +08:00
55504fcd4d
新增加知识库测试能力 ( #302 )
2023-05-10 17:18:20 +08:00
466bfb7a3f
logger bug introduced by https://github.com/imClumsyPanda/langchain-ChatGLM/pull/268 , fixed #295 ( #297 )
2023-05-10 13:28:44 +08:00
c613c41d0e
update local_doc_qa.py
2023-05-08 23:49:57 +08:00
e2d7452c7b
update chinese_text_splitter.py
2023-05-06 23:26:49 +08:00
2c72a00954
Update local_doc_qa.py
...
增加命令行导入的进度条显示
增加文本提示
2023-05-06 18:39:58 +08:00
14d998b8e6
可选择lora权重加载 ( #231 )
...
* Add files via upload
增加lora权重使用
* Update model_config.py
* Add files via upload
修复一个小错误，少写了模型加载
* 使用lora微调的权重
使用lora微调的权重
* Update model_config.py
2023-05-04 21:01:55 +08:00
47922d2ee3
fix: 扩展文档的代码逻辑 ( #227 )
...
* fix: 扩展文档的代码逻辑
* Update local_doc_qa.py
---------
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
2023-05-04 20:58:15 +08:00
07ff81a119
update torch_gc
2023-05-04 20:48:36 +08:00
liunux4odoo
liunux4odoo
imClumsyPanda
glide-the
Zhi-guo Huang
kiddog99
shrimp
Ding Junyao
yawudede
zhenkaivip
Tyler Luan
JohnSaxon
keenzhu
Winter