bfdbe69fa1
增加了自定义分词器适配 ( #1462 )
...
* 添加了自定义分词器适配和测试文件
---------
Co-authored-by: zR <zRzRzRzRzRzRzR>
2023-09-13 15:42:12 +08:00
c4cb4e19e5
清理不必要的依赖，增加星火API需要的websockets ( #1463 )
2023-09-13 15:35:04 +08:00
99b862dfc8
fix SparkApi import error
2023-09-13 14:03:09 +08:00
512638a3b1
支持讯飞星火大模型在线API ( #1460 )
...
* 支持讯飞星火大模型在线API
2023-09-13 13:51:05 +08:00
16d8809c9a
api和webui知识库操作支持chunk_size/overlap_size/zh_title_enhance参数 ( #1459 )
2023-09-13 11:19:47 +08:00
9defa4332e
fix minimax api history error
2023-09-13 10:12:31 +08:00
8b040620de
对话接口支持temperature参数 ( #1455 )
2023-09-13 10:00:54 +08:00
a03b8d330d
升级到langchain==0.0.287,fschat==0.0.28;处理langchain.Milvus bug( #10492 ) ( #1454 )
...
* 修复milvus_kb_service中一些bug，添加文档后将数据同步到数据库
* 升级到langchain==0.0.287,fschat==0.0.28;处理langchain.Milvus bug(#10492 )
* 修复切换模型BUG: 从在线API切换模型时出错
2023-09-13 08:43:11 +08:00
efd6d4a251
修复milvus_kb_service中一些bug，添加文档后将数据同步到数据库 ( #1452 )
2023-09-12 22:34:03 +08:00
2020f5bb88
merge master
2023-09-12 15:51:28 +08:00
03c021d50f
添加MiniMax在线API支持 ( #1445 )
2023-09-12 15:24:47 +08:00
8064586374
delete .doc from supported formats
2023-09-12 15:22:18 +08:00
60f88ae92e
fix: 由于server.utils导入model_workers时，导入fastchat model_worker，导致startup启动延迟明显 ( #1436 )
2023-09-12 08:52:00 +08:00
6a03611750
fix api and webui: ( #1435 )
...
1. fix #1431 : 优化知识库问答与搜索引擎问答的API接口，避免docs重复返回
2. startup.py根据configs.log_verbose控制log级别
3. 修复/llm_model/list_models的bug:
只有一个参数时，fastapi未返回json导致视图函数出错
2023-09-12 08:31:17 +08:00
22ff073309
修改Embeddings和FAISS缓存加载方式，知识库相关API接口支持多线程并发 ( #1434 )
...
* 修改Embeddings和FAISS缓存加载方式，支持多线程，支持内存FAISS
* 知识库相关API接口支持多线程并发
* 根据新的API接口调整ApiRequest和测试用例
* 删除webui.py失效的启动说明
2023-09-11 20:41:41 +08:00
5550e3c1a9
增加显示详细日志开关
2023-09-08 20:48:31 +08:00
775870a516
改变api视图函数的sync/async，提高api并发能力： ( #1414 )
...
1. 4个chat类接口改为async
2. 知识库操作，涉及向量库修改的使用async，避免FAISS写入错误；涉及向量库读取的改为sync，提高并发
2023-09-08 12:25:02 +08:00
1195eb75eb
merge pr1413
2023-09-08 10:30:07 +08:00
661a0e9d72
新功能:
...
- 知识库管理中的add_docs/delete_docs/update_docs均支持批量操作，并利用多线程提高效率
- API的重建知识库接口支持多线程
- add_docs可提供参数控制上传文件后是否继续进行向量化
- add_docs/update_docs支持传入自定义docs(以json形式)。后续考虑区分完整或补充式自定义docs
- download_doc接口添加`preview`参数，支持下载或预览
- kb_service增加`save_vector_store`方法，便于保存向量库（仅FAISS，其它无操作）
- 将document_loader & text_splitter逻辑从KnowledgeFile中抽离出来，为后续对内存文件进行向量化做准备
- KowledgeFile支持docs & splitted_docs的缓存，方便在中间过程做一些自定义
其它：
- 将部分错误输出由print改为logger.error
2023-09-08 08:55:12 +08:00
1027ab6c06
增加构建文档metadata检索的开关，移动到dev_metadata_search
2023-09-05 20:59:57 +08:00
491c9605d9
增加构建文档metadata检索的开关
2023-09-05 18:35:11 +08:00
3e87530478
使zhipuai变为非必要依赖( close #1369 )
2023-09-05 12:43:51 +08:00
464ccc6d53
Update migrate.py ( #1371 )
...
fix bug at line 85 : for kb_file in kb_file
2023-09-05 11:52:22 +08:00
015ce57691
Update migrate.py ( #1371 )
...
fix bug at line 85 : for kb_file in kb_file
2023-09-05 11:47:25 +08:00
93b133f9ac
将KnowledgeFile的file2text拆分成file2docs、docs2texts和file2text三个部分，在保持接口不变的情况下，实现：
...
1、支持chunk_size和chunk_overlap参数
2、支持自定义text_splitter
3、支持自定义docs
修复：csv文件不使用text_splitter
2023-09-04 21:00:16 +08:00
e21ca447af
milvus/pg kb_service需要实现get_doc_by_id方法
2023-09-04 16:40:05 +08:00
0bc9d5c8ee
Merge branch 'dev' into opt_add_doc
2023-09-03 21:38:45 +08:00
8a73673d6b
update model_config.py.example
2023-09-03 15:57:44 +08:00
25880ebd52
fix bug when config["device"]="auto"
2023-09-02 22:11:54 +08:00
6cb1bdf623
添加切换模型功能，支持智谱AI在线模型 ( #1342 )
...
* 添加LLM模型切换功能，需要在server_config中设置可切换的模型
* add tests for api.py/llm_model/*
* - 支持模型切换
- 支持智普AI线上模型
- startup.py增加参数`--api-worker`，自动运行所有的线上API模型。使用`-a
(--all-webui), --all-api`时默认开启该选项
* 修复被fastchat覆盖的标准输出
* 对fastchat日志进行更细致的控制，startup.py中增加-q(--quiet)开关，可以减少无用的fastchat日志输出
* 修正chatglm api的对话模板
Co-authored-by: liunux4odoo <liunu@qq.com>
2023-09-01 23:58:09 +08:00
55e417a263
升级注意
...
数据库表发生变化，需要重建知识库
新功能
- 增加FileDocModel库表，存储知识文件与向量库Document ID对应关系以及元数据，便于检索向量库
- 增加FileDocModel对应的数据库操作函数（这些函数主要是给KBService调用，用户一般无需使用）：
- list_docs_from_db: 根据知识库名称、文件名称、元数据检索对应的Document IDs
- delete_docs_from_db: 根据知识库名称、文件名称删除对应的file-doc映射
- add_docs_to_db: 添加对应的file-doc映射
- KBService增加list_docs方法，可以根据文件名、元数据检索Document。当前仅支持FAISS，待milvus/pg实现get_doc_by_id方法后即自动支持。
- 去除server.utils对torch的依赖
待完善
- milvus/pg kb_service需要实现get_doc_by_id方法
2023-09-01 22:54:57 +08:00
d6dce95682
bug fix
2023-09-01 14:15:13 +08:00
6c4ef26e9a
add RapidOCRPDFLoader and RapidOCRLoader ( #1275 )
...
* add RapidOCRPDFLoader
* update mypdfloader.py and requirements.txt
* add myimgloader.py
* add test samples
* add TODO to mypdfloader
* add loaders to KnowledgeFile class
* add loaders to KnowledgeFile class
2023-09-01 10:23:57 +08:00
b1201a5f23
优化LLM和Embedding模型运行设备配置，可设为auto自动检测；修复：重建知识库时FAISS未保存 ( #1330 )
...
* 避免configs对torch的依赖；
* webui自动从configs获取api地址(close #1319 )
* bug fix: 重建知识库时FAISS未保存
* 优化LLM和Embedding模型运行设备配置，可设为auto自动检测
2023-08-31 17:44:48 +08:00
34a416b941
优化server_config配置项 ( #1293 )
...
* update server_config.py:
- 在model_config中增加HISTORY_LEN配置参数
- 将server_config中helper function移动到server.utils中
- 统一set_httpx_timeout的定义和调用
* update webui.py:
应用model_config中的配置项：HISTORY_LEN,VECTOR_SEARCH_TOP_K,SEARCH_ENGINE_TOP_K
---------
Co-authored-by: liunux4odoo <liunu@qq.com>
2023-08-29 10:06:09 +08:00
ca0ae29fef
运行startup.py时，如果不加参数直接显示配置和帮助信息后退出 ( #1284 )
...
* 统一XX_kb_service.add_doc/do_add_doc接口，不再需要embeddings参数
* 运行startup.py时，如果不加参数直接显示配置和帮助信息后退出
---------
Co-authored-by: liunux4odoo <liunu@qq.com>
2023-08-28 16:03:53 +08:00
3acbf4d5d1
增加数据库字段，重建知识库使用多线程 ( #1280 )
...
* close #1172 : 给webui_page/utils添加一些log信息，方便定位错误
* 修复：重建知识库时页面未实时显示进度
* skip model_worker running when using online model api such as chatgpt
* 修改知识库管理相关内容:
1.KnowledgeFileModel增加3个字段：file_mtime(文件修改时间),file_size(文件大小),custom_docs(是否使用自定义docs)。为后面比对上传文件做准备。
2.给所有String字段加上长度，防止mysql建表错误(pr#1177)
3.统一[faiss/milvus/pgvector]_kb_service.add_doc接口，使其支持自定义docs
4.为faiss_kb_service增加一些方法，便于调用
5.为KnowledgeFile增加一些方法，便于获取文件信息，缓存file2text的结果。
* 修复/chat/fastchat无法流式输出的问题
* 新增功能：
1、KnowledgeFileModel增加"docs_count"字段，代表该文件加载到向量库中的Document数量，并在WEBUI中进行展示。
2、重建知识库`python init_database.py --recreate-vs`支持多线程。
其它：
统一代码中知识库相关函数用词：file代表一个文件名称或路径，doc代表langchain加载后的Document。部分与API接口有关或含义重叠的函数暂未修改。
---------
Co-authored-by: liunux4odoo <liunux@qq.com>, hongkong9771
2023-08-28 13:50:35 +08:00
8faafa6a32
update format
2023-08-27 20:23:07 +08:00
1fa4e906c7
适配score_threshold
2023-08-27 11:21:10 +08:00
96f69edf7b
修复重建知识库界面无反应的问题，添加一些log信息 ( #1258 )
...
* close #1172 : 给webui_page/utils添加一些log信息，方便定位错误
* 修复：重建知识库时页面未实时显示进度
---------
Co-authored-by: liunux4odoo <liunu@qq.com>
2023-08-26 18:26:38 +08:00
9ca53fa3ad
fix #1224 : failed to load json and pdf files ( #1252 )
...
Co-authored-by: liunux4odoo <liunu@qq.com>
2023-08-26 14:21:59 +08:00
f95d41ef47
[BUG] 修复csv文件读取后，单行数据被分成多段。
2023-08-26 11:45:01 +08:00
4dafca2376
Merge branch 'master' into pre-release
2023-08-25 16:25:35 +08:00
447b370416
在search_engine_chat中检查Bing KEY，并更新tests
2023-08-25 10:58:40 +08:00
0c63839ff6
修正 milvus 更改错误
2023-08-24 22:53:13 +08:00
2604341e9e
调整KBService 接口函数do_search 并适配pg
2023-08-24 22:35:30 +08:00
24e7351d32
Merge branch 'master' into dev
2023-08-24 22:31:37 +08:00
67ac53242d
Merge pull request #1233 from Pickupppp/master
...
修复 milvus 作为默认数据库初始化异常和调用search方法异常
2023-08-24 22:28:38 +08:00
bc3dbb443e
修复搜索引擎聊天接口 ( #1228 )
...
* 修复搜索引擎聊天接口
* 修复知识库聊天接口
* 恢复
2023-08-24 22:18:33 +08:00
050a585a9d
修复 milvus 作为默认数据库初始化异常和调用search方法异常
2023-08-24 21:31:02 +08:00
2a2b95d214
search_engine_chat bug
2023-08-24 17:25:54 +08:00
63ca393731
参照chat，给knowledge_base_chat/search_engine_chat添加chatgpt代理设置
2023-08-24 15:10:51 +08:00
ea4b5ebd16
Merge branch 'master' into dev
2023-08-24 14:16:21 +08:00
37454db3ec
添加chatgpt 调用代理设置
2023-08-24 13:51:17 +08:00
7291e77978
support for bge-large-zh-noinstruct and openai text-embedding-ada-002 ( #1119 )
...
* support for bge-large-zh-noinstruct and openai text-embedding-ada-002
* 完善了readme,修改了BGE模型的载入方式
* 补充了readme中bge-large-zh-noinstruct的信息
* Update faiss_kb_service.py
* Update utils.py
---------
Co-authored-by: zR <zRzRzRzRzRzRzR>
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>
2023-08-23 22:55:53 +08:00
351b62f7a7
fix part of #1206 : add/delete/update_doc() get unexpected keyword argument 'not_refresh_vs_cache`
2023-08-23 15:26:52 +08:00
44edce6bcf
fix #1142 : 在History中使用jinja2模板代替f-string，避免消息中含有{ }时出错
2023-08-23 08:35:26 +08:00
16542f20b4
修复pg和milvus kbservice代码
2023-08-22 16:52:04 +08:00
11856cc542
Merge pull request #1160 from shutter-cp/master
...
fix bug 1159
2023-08-22 16:38:06 +08:00
c571585ffd
optimize recreate vector store: save vector store once after all docs parsed for FAISS
2023-08-21 08:50:15 +08:00
f40bb69224
优化FAISS向量库多文件操作；修复recreate_vector_store，大量文件时不再超时。
2023-08-20 19:10:29 +08:00
150a78bfd9
update kb_doc_api:make faiss cache working; delete vector store docs before add duplicate docs
2023-08-20 16:52:49 +08:00
62aa97bb32
Merge branch 'master' into dev
2023-08-20 10:39:30 +08:00
cdea240c31
update README.md and llm_api_stale.py
2023-08-20 10:38:41 +08:00
69627a2fa3
update chat and knowledge base api: unify exception processing and return types
2023-08-19 15:14:45 +08:00
95d9fb0ee9
fix bug 1159
...
修复使用使用pg存储向量时，向量检索错误 {TypeError}TypeError("'Document' object is not subscriptable")
2023-08-18 14:41:11 +08:00
4fb2e2198b
merge dev_allinone
2023-08-17 22:29:26 +08:00
52837429d2
update configs.__init__, webui.py and api.py
2023-08-17 22:19:37 +08:00
6d1ea9bc99
Update utils.py ( #1128 )
...
不区分文件扩展名的大小写。
2023-08-17 21:09:05 +08:00
44d4bacf1c
update llm_api_launch,api_allinone,webui_allinone,readme:1.llm_api_launch,api_allinone_webui_allinone更名;2. 更新readme关于startup的说明
2023-08-17 16:59:16 +08:00
f92b002342
add startup.py: start specified servers with one command. see python startup.py --help
2023-08-17 13:24:53 +08:00
8396b57101
update format of api_allinone.py and llm_api_launch.py
2023-08-16 22:24:29 +08:00
6c00c03faa
llm_api can be terminated by Ctrl+c. success on windows
2023-08-16 17:48:55 +08:00
e7d07a7923
Merge remote-tracking branch 'refs/remotes/origin/dev' into dev
2023-08-16 16:16:56 +08:00
8ed1f8efca
update fastchat to 0.2.24
2023-08-16 16:15:45 +08:00
84b491b8b2
Merge branch 'dev' of github.com:chatchat-space/Langchain-Chatchat into dev
2023-08-16 16:00:40 +08:00
5f1471232d
update llm_api_launch.py:修复本地加载问题
2023-08-16 16:00:24 +08:00
f64affc930
update llm_api and api server:
...
1. fastchat's controller/model_worker/api_server use swagger UI offline.
2. add custom title and icon.
3. remove fastapi-offline dependence
2023-08-16 14:20:09 +08:00
67b8ebef52
update api and webui:
...
1. 增加search_docs接口，返回原始知识库检索文档，close #1103
2. 为FAISS检索增加score_threshold参数。milvus和PG暂不支持
2023-08-16 13:18:58 +08:00
b97358ea33
Merge branch 'dev' of github.com:chatchat-space/Langchain-Chatchat into dev
2023-08-16 11:15:31 +08:00
8e00f8305b
1. 更新allinone脚本提示，增加中文提示;2. 新增服务一键停止脚本shutdown_all.sh;3. 更新readme说明;3. 修复命令行bug
2023-08-16 11:15:10 +08:00
a47240e871
fix KnowledgeFile: set text_splitter_name in database to SpacyTextSplitter if it is not set
2023-08-16 10:14:47 +08:00
348edb9a0d
更新提示
2023-08-16 10:13:51 +08:00
7c0e9c46db
fix api & webui: 1.add logs, 2. check json response, 3. set timeout=False in recreate_vector_stoore
2023-08-15 14:24:54 +08:00
80f44ba24e
git pull dev
2023-08-15 11:22:19 +08:00
5c0274efce
1. 增加api的一键启动脚本;2. 增加webui的一键启动脚本;3. 更新readme;4. 调整llm_api_launch
2023-08-15 11:02:32 +08:00
8477adfa33
update api and webui:
...
1. add download_doc to api
2. return local path or http url in kowledge_base_chat depends on
no_remote_api
3. change assistant avater in webui
2023-08-14 19:32:28 +08:00
087bb37e08
fix chat and knowledge_base_chat
2023-08-14 19:32:28 +08:00
ec984205ae
fix knowledge base management:
...
1. docs in database were note deleted when clear vector store
2. diable buttons when local doc file not exist.
2023-08-14 19:09:02 +08:00
8b18cf2a5e
git rebase and pull
2023-08-14 16:48:35 +08:00
0704eabe23
fix chat and knowledge_base_chat
2023-08-14 16:46:26 +08:00
90c92f8c93
解决多卡启动问题
2023-08-14 16:43:01 +08:00
cc08e2cb96
update api and webui:
...
1. add download_doc to api
2. return local path or http url in kowledge_base_chat depends on
no_remote_api
3. change assistant avater in webui
2023-08-14 11:46:36 +08:00
0d6a9cf8f3
Merge branch 'master' into dev
2023-08-14 11:39:12 +08:00
6590ca32db
更新多卡部署
2023-08-14 11:24:52 +08:00
126bd51232
fix chat and knowledge_base_chat
2023-08-14 10:35:47 +08:00
afc6aa968d
update webui layout
2023-08-11 23:40:41 +08:00
2a57f49d80
update knowledge base management ui.
...
remove numpy/pandas from requirements, move them to webui requirements
2023-08-11 13:53:20 +08:00
a08fe994c2
更新API与ApiReuest：
...
1. 重新整理webui_pages/utils与server/knowledge_base间的工具依赖
2.
将delete_knowledge_base与delete_doc接口从delete改为post.delete不支持body参数
3. 修复update_doc
4. 修复部分bug
2023-08-11 08:45:26 +08:00
a261fda20b
add tools to migrate knowledge base
2023-08-11 08:33:05 +08:00
ef098111dc
update webui and utils
2023-08-10 23:04:05 +08:00
8a4d9168fa
update import pkgs and format
2023-08-10 21:26:05 +08:00
18e7e7307b
patch fastapi-offline to use local swagger-ui assests
2023-08-10 17:00:39 +08:00
0364215987
Merge branch 'pr1037_pg_vs' into dev_fastchat
2023-08-10 15:07:58 +08:00
17ddee38a4
fix conflict between def_fastchat and pr1037
2023-08-10 15:02:41 +08:00
9525915a85
update llm_api_launch,llm_api_shutdown
2023-08-10 14:43:18 +08:00
fd247f4657
Merge branch 'dev_fastchat' into pr1037_pg_vs
2023-08-10 14:21:37 +08:00
9ed91f0e8a
automatically init vector store info to database when run api.py & webui.py
2023-08-10 14:12:02 +08:00
f5024245a7
make api.py and webui.py working offline
2023-08-10 13:33:40 +08:00
d4f728dfa2
添加向量数据库pg支持，和向量库docker-compose.ym环境文件
2023-08-10 11:16:52 +08:00
6c2b2b7ba2
测试llm_api_shutdown.py,通过
2023-08-10 09:38:45 +08:00
8df00d0b04
add SCORE_THRESHOLD to faiss and milvus kb service
2023-08-10 00:36:59 +08:00
1a112c6908
update fastch_chat: disable streaming in swagger besides streaming in ApiRequest
2023-08-09 23:48:41 +08:00
02b9d57072
update textsplitter and pdfloader
2023-08-09 23:36:43 +08:00
222689ed5b
update kb and search chat: disable streaming in swagger besides streaming in ApiRequest
2023-08-09 23:35:36 +08:00
66ff913bb4
Merge branch 'dev_fastchat' of https://github.com/chatchat-space/langchain-ChatGLM into dev_fastchat
2023-08-09 23:18:54 +08:00
3e9d088cd9
update server
2023-08-09 23:18:44 +08:00
34bb681db1
删除shell脚本
2023-08-09 23:17:51 +08:00
8a34c3f163
增加停止服务的python脚本
2023-08-09 23:16:02 +08:00
24a280ce8c
re-add zh_title_enhance.py
2023-08-09 23:09:24 +08:00
1b70fb5f9b
update faiss_kb_service.py
2023-08-09 22:57:36 +08:00
db29a2fea7
update knowledge_base_chat and search_engine_chat: yield chinese correctly
2023-08-09 22:54:37 +08:00
ba3335efb8
update llm_api: move fastchat logs to LOG_PATH
2023-08-09 22:43:45 +08:00
997f8b2e3f
update knowledge base api:
...
1. list_kbs_from_db return all kbs by default instead of return kbs with
file_count > 0 only.
2. KBServiceFactory.get_service_by_name could return a FaissKBService
that not in the db
2023-08-09 21:57:40 +08:00
323fc13d4c
make parameter examples available in openapi docs
2023-08-09 18:15:14 +08:00
c7b91bfaf1
update api/kb_doc_api and knowledge base management ui:
...
1. add update_doc to api which can udpate vector store from existed
content file
2. add parameter `delete_content` to delete_doc api. user can decide
whether delete local content file when delete doc.
3. fix bug in ApiReqeust.upload_doc
4. support listing docs existed in local folder bu not in db
2023-08-09 16:52:04 +08:00
4da7c04859
fix: parse history parameter to correct type
2023-08-09 12:09:45 +08:00
9883341924
update server.chat.*: set default value [] to history parameter.
2023-08-09 10:48:37 +08:00
b98f5fd0b9
update knowledge base kbservice and api:
...
1. make huggingfaceembeddings hashable
2. unify embeddings' loading method for all kbservie
3. make ApiRequest skip empty content when streaming json to avoid dict
KeyError
2023-08-09 10:46:01 +08:00
532ed81ecf
add functions to get kb and doc details from db
2023-08-08 23:57:39 +08:00
717ddf4d10
add history to chat apis
2023-08-08 23:56:27 +08:00
c17572b6a6
add history to chat apis
2023-08-08 23:55:27 +08:00
2d49746a8d
add history to chat apis
2023-08-08 23:54:51 +08:00
c8b6d53ede
add llmchain_with_history.py as example
2023-08-08 22:15:16 +08:00
135af5f3ff
update knowledge base db orm:
...
1. set default values for file_count, file_version, create_time.
2. fix bug: add_doc_to_db
3. make add_kb_to_db more flexiable with existing kb
2023-08-08 21:47:20 +08:00
2b0f8caa62
improve recreate_vector_store:
...
1. allow empty knowledge base
2. skip unspported files instead of exception
2023-08-08 17:59:41 +08:00
e918244159
move KBServiceFactory to server.knowledge_base.kb_service.base
2023-08-08 17:41:58 +08:00
444e1afad5
remove server/knowledge_base/knowledge_file.py
2023-08-08 16:51:37 +08:00
c4a3995310
move commonto tools from kbservice.base to server.knowledge_base.utils
2023-08-08 16:40:18 +08:00
584e3a9234
remove server/knowledge_base/knowledge_base_factory.py
2023-08-08 16:21:00 +08:00
7e07f0bbf4
move configs to configs.model_config
2023-08-08 15:56:32 +08:00
b61e0772c9
使用orm操作数据库
2023-08-08 14:25:55 +08:00
41fd1acc9c
Merge remote-tracking branch 'origin/dev_fastchat' into dev_fastchat
...
# Conflicts:
# server/knowledge_base/kb_service/default_kb_service.py
# server/knowledge_base/kb_service/milvus_kb_service.py
# server/knowledge_base/knowledge_base_factory.py
2023-08-08 14:06:13 +08:00
82494aec00
Merge branch 'dev_fastchat' of github.com:chatchat-space/langchain-ChatGLM into dev_fastchat
2023-08-08 13:39:31 +08:00
360bd0a559
update llm_api_sh.py
2023-08-08 13:39:19 +08:00
0746272525
remove server/knowledge_base/knowledge_base.py
2023-08-08 13:36:20 +08:00
de8db40f4b
BaseChatOpenAIChain,支持基础的ChatOpenAI对话的Chain接入
2023-08-07 23:05:54 +08:00
823eb06c5d
BaseChatOpenAIChain,支持基础的ChatOpenAI对话的Chain接入
2023-08-07 22:57:13 +08:00
08493bffbb
add a basic knowledge base management ui.
2023-08-07 21:00:55 +08:00
035a199c32
update kb_server.base
2023-08-07 20:48:06 +08:00
44c713ef98
use KBServiceFactory to replace all the KnowledgeBase.
...
make KBServiceFactory support embed_model parameter.
rewrite api: recreate_vector_store.
fix some bugs.
2023-08-07 20:37:16 +08:00
0f46185cfb
添加Milvus库 ( #1011 )
2023-08-07 16:56:57 +08:00
72951d95fb
添加milvus库支持
2023-08-07 16:32:34 +08:00
18d31f5116
add KBService and KBServiceFactory class
2023-08-06 23:44:02 +08:00
06af3f4c5e
BaseChatOpenAIChain,支持基础的ChatOpenAI对话的Chain接入
2023-08-06 22:57:57 +08:00
b91d96ab0c
update kb_doc_api.py
2023-08-06 18:32:10 +08:00
a447529c2e
update kb_doc_api.py
2023-08-05 23:35:20 +08:00
8773149a3e
update api.py
2023-08-05 23:12:17 +08:00
7f89efee4b
fix kb_doc_api.py
2023-08-05 23:01:56 +08:00
3f045cedb9
1. add add_doc and list_docs to KnowledgeBase class
...
2. add DB_ROOT_PATH to model_config.py.example
2023-08-05 22:57:19 +08:00
313e590961
update DocumentLoader in knowledge_file.py
2023-08-05 21:51:07 +08:00
f68e432570
update table_name and column_name in SQL
2023-08-05 21:02:14 +08:00
cd0f2a2c41
update class method of KnowledgeBase and KnowledgeFile
2023-08-05 20:52:14 +08:00
14a5de6a29
update class method of KnowledgeBase and KnowledgeFile
2023-08-05 13:51:05 +08:00
7beeaacea6
update class method of KnowledgeBase and KnowledgeFile
2023-08-05 13:48:55 +08:00
e8a88f174d
update class method of KnowledgeBase and KnowledgeFile
2023-08-05 13:48:30 +08:00
206261cd0c
update class method of KnowledgeBase and KnowledgeFile
2023-08-05 13:46:00 +08:00
6c7adfbaeb
change kb_api functions with KnowledgeBase class method
2023-08-05 03:15:41 +08:00
7e1472a95b
add KnowledgeBase class
2023-08-05 00:10:10 +08:00
590367a5b5
add KnowledgeFile class
2023-08-04 23:54:26 +08:00
775d46fecf
format update in knowledge_base/utils.py
2023-08-04 23:14:43 +08:00
bc3ee83910
update kb_doc_api.py
2023-08-04 22:22:48 +08:00
5d1a0ec15d
add recreate_vector_store to kb_doc_api, and the router in api.py.
2023-08-04 20:26:14 +08:00
46c7d8d169
allow kb_doc_api.upload_doc to override existed file by parameter.
...
update_doc is not needed.
2023-08-04 15:53:44 +08:00
b4f80ca370
update server.knowledge_base.utils
2023-08-04 15:32:42 +08:00
7bfbe18011
update requirements.txt and kb_doc_api.py
2023-08-04 15:12:14 +08:00
c4994e85df
fix: knowledge_base_chat and search_engine_chat should stream json text
...
instead of dict. client should use json.loads.
2023-08-04 12:12:13 +08:00
88682c87ff
优化server下知识库相关模块的结构：将知识库相关操作从knowledge_base_chat移动到knowledge_base.utils；优化kb_doc_api中embeddings加载。
2023-08-04 10:16:30 +08:00
329c24ee73
1. change bing_search_chat and duckduckgo_search_chat into search_engine_chat
...
2. add knowledge_base not found to knowledge_base_chat and add search_engine not found to search_engine_chat
2023-08-03 18:22:36 +08:00
b62ea6bd2a
add top-k to search_chat apis and add top-k params in model_config
2023-08-03 17:06:43 +08:00
5c804aac75
make api and webui support top_k paramter
2023-08-03 15:47:53 +08:00
df61f31c8e
修改knowledge_base_chat实现:利用缓存避免每次请求重新加载向量库；增加top_k参数。
2023-08-03 15:22:46 +08:00
fdbff8a91f
Merge branch 'dev_fastchat' of github.com:chatchat-space/langchain-ChatGLM into dev_fastchat
2023-08-03 14:39:14 +08:00
6f93a27e53
更新llm_api_sh.py调用说明
2023-08-03 14:39:00 +08:00
f7d465b7d4
bug fix
2023-08-03 12:11:18 +08:00
d0fd6253a3
update llm_api_sh
2023-08-02 09:19:58 +08:00
1d1a8e9339
Merge branch 'dev_fastchat' of github.com:chatchat-space/langchain-ChatGLM into dev_fastchat
2023-08-01 22:07:12 +08:00
ab7a76f380
update llm_api_sh.py and model_config.example
2023-08-01 22:07:05 +08:00
e4fed93989
update format in api.py
2023-08-01 21:53:19 +08:00
18a94fcf45
Merge branch 'dev_fastchat' of github.com:chatchat-space/langchain-ChatGLM into dev_fastchat
2023-08-01 18:02:52 +08:00
15e67a4d3e
1.*在config里将所有fastchat的命令行参数加入;2.*加入启动和停止fastchat的shell脚本；3. **增加了通过命令行启动所有fastchat服务的python脚本llm_api_sh.py;4. 修改了默认的config日志格式
2023-08-01 17:59:20 +08:00
7c01a2a253
add bing_search_chat.py and duckduckgo_search_chat.py
2023-08-01 16:39:17 +08:00
c8a75ab11f
update llm_api.py and webui.py
2023-08-01 14:33:18 +08:00
9e2b411b01
cuda error with multiprocessing, change model_worker to main process
2023-07-31 11:18:57 +08:00
47dfb6cd8b
udpate llm_api
2023-07-31 11:00:33 +08:00
946f10e1f2
split api_start to create_app & run_api. user can run api with uvicorn in console: uvicorn server.api:app --port 7861
2023-07-31 10:05:19 +08:00
463659f0ba
llm_api升级到fschat==0.2.20，支持百川模型
2023-07-30 23:16:47 +08:00
51ea717606
Merge branch 'dev_fastchat_me' into dev_fastchat
2023-07-30 09:01:31 +08:00
179c2a9a92
修改server.chat.openai_chat中的参数定义，使其与openai中/v1/chat/completions接口的入参保持一致，并按照model_config提供默认值。
...
openai_chat中的接口还要修改：openai根据参数stream有不同的返回值，本接口要与其对应。
2023-07-30 08:56:49 +08:00
05ccc0346e
update test code
2023-07-30 00:48:07 +08:00
41444fd4b5
update requirements.txt and llm_api.py
2023-07-30 00:24:34 +08:00
1a7271e966
fix: model_worker need global variable: args
2023-07-29 23:22:25 +08:00
c880412300
use multiprocessing to run fastchat server
2023-07-29 23:01:24 +08:00
829ced398b
make api.py import model_config coreectlly
2023-07-29 23:00:50 +08:00
97ee4686a1
更新会议记录
2023-07-28 16:16:59 +08:00
154cad1b45
会议记录
2023-07-28 16:12:57 +08:00
dcf49a59ef
v0.2.0 first commit
2023-07-27 23:22:07 +08:00
zR
liunux4odoo
liunux4odoo
imClumsyPanda
glide-the
zqt
liunux4odoo
Pickupppp
zqt996
Meiracle
陈鹏
xain
hzg0601