klas/t6_mem0
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
140 Commits 1 Branch 0 Tags
d2e8f796ca60d41e09e0e69415d1ec5e056b884e
Commit Graph

45 Commits

Author SHA1 Message Date
cachho
40dc28406d feat: add multi-document answers (#63) 2023-07-11 18:20:07 +05:30
cachho
9409e5605a chore: linting (#230) 2023-07-11 17:34:57 +05:30
cachho
1c4f629fc0 chore: linting (#229) 2023-07-11 17:15:53 +05:30
cachho
35b6e14328 fix: initialize embedding function manually (#205) 2023-07-11 17:08:48 +05:30
ahnedeee
4d26d7b6d6 feat: customize OpenAI parameters (#215) 2023-07-11 15:53:54 +05:30
Deshraj Yadav
9ca836520f Resolve conflicts (#208) 2023-07-11 10:20:05 +05:30
Anupam Singh
eda28cc491 featL AddConfig should allow configuring Chunker (#200) 2023-07-11 04:23:56 +05:30
aaishikdutta
ae87dc4a6d feat: add streaming support for OpenSource App (#217) 2023-07-11 01:38:22 +05:30
cachho
65414af098 feat: History API (client) (#204) 2023-07-10 23:45:20 +05:30
cachho
b3cf834186 feat: add logging (#206) 2023-07-10 23:28:58 +05:30
aaishikdutta
c597b1939d feat: Use streaming setup at Query level (#214) 2023-07-10 23:07:19 +05:30
aaishikdutta
66c4d30c60 feat: add streaming support for OpenAI models (#202) 2023-07-10 17:18:24 +05:30
cachho
13bac72e25 feat: Show new chunk count when adding a dataset (#207) 2023-07-10 16:41:12 +05:30
cachho
feda6b347a fix: use custom prompt in dry run (#203) 2023-07-09 01:31:06 +05:30
aaishikdutta
ae1e21833c refactor: add data-format restructuring (#92) 2023-07-07 23:53:35 +05:30
Taranjeet Singh
2a7d43551e feat: Add person bot (#182) 
This commit introduces a new bot type called Person.
It behaves and speaks like a particular Person.

For this two app types: PersonApp and PersonOpenSourceApp
 2023-07-07 19:26:07 +05:30
Anupam Singh
6a61fd38c3 feat: Add config based template to query function (#177) 2023-07-07 17:56:43 +05:30
Sahil Kumar Yadav
0bb3d0afe9 feat: changed doc_file to docx and update readme (#157) 2023-07-07 16:18:05 +05:30
cachho
51adc5c886 refactor: Use src instead of url as argument value (#111) 2023-07-07 16:14:44 +05:30
cachho
5ae2a36305 fix: value error in metadata(#110) 2023-07-07 16:12:02 +05:30
gasolin
5e6aef1e37 feat: pass QueryConfig to dry_run (#173) 2023-07-07 16:08:14 +05:30
cachho
0c2653ccf5 fix: embedding function config (#175) 2023-07-07 14:36:43 +05:30
cachho
41a1dbab0e feat: add count to get document count in vector db (#164) 2023-07-07 14:31:14 +05:30
cachho
f3888cc7a8 feat: reset (#169) 2023-07-07 00:45:01 +05:30
cachho
e50c7e6843 refactor: Add config for init, app and query (#158) 2023-07-06 23:46:40 +05:30
Sahil Kumar Yadav
68e732a426 feat: add google doc support added (#155) 2023-07-06 14:04:27 +05:30
Taranjeet Singh
687e3fb61c feat: Add support for chat endpoint (#150) 
* feat: Add support for chat endpoint

This commit introduces a new endpoint called chat.
It maintains context of the last 5 user messages
and continues to do semantic search to find the
context.
 2023-07-06 13:32:16 +05:30
Rohit Gupta
0e74688145 update: import gpt4all locally (#146) 
Import gpt4all locally to avoid error in case its not present.
 2023-07-06 01:01:32 +05:30
cachho
f8e5ccd007 feat: add dry run (#47) 
This commit adds support to simulate the semantic search and see what the prompt will look lile.
 2023-07-05 23:19:53 +05:30
Taranjeet Singh
200f11a0e0 fix: Fix dependency of openai env variables for OpenSourceApp (#144) 
This commit fixes dependency of initializing openai env variables
for OpenSourceApp.
 2023-07-05 23:03:15 +05:30
Taranjeet Singh
cf1e000fb3 Open source embedding and LLM models (#133) 
* Add open source LLM model: gpt4all
* Add open source embedding model: sentence transformers
 2023-07-05 02:23:23 +05:30
limbo
85008ff153 Fix: Solve IndexError when no relevant documents are found in the vector db. (#59) 2023-07-04 22:18:56 +05:30
cachho
f5f5e7edd1 feat: add local text (#44) 
This commits extends the "add_local" function. It
adds support to take text and index/embed it.
 2023-06-25 23:13:41 +05:30
cachho
cf99dce940 Refactor query endpoint into 3 parts (#42) 
Query endpoint now consists of 3 sub functions
- get data from db
- get prompt
- get answer from the data retrieved above by passing to LLM
 2023-06-23 22:17:52 +05:30
cachho
2d35ade7aa refactor: local indicator in file and class name, renamed 2023-06-23 19:53:57 +05:30
cachho
ff2d5ce7fa feat: add local qna pair 2023-06-23 19:53:57 +05:30
Taranjeet Singh
e40f1e5061 Skip creating embedding if it already exists 
This commit adds support to skip generating embedding
for chunks which are already present in vector db.
 2023-06-23 17:08:49 +05:30
Taranjeet Singh
cff244b894 Add default db loader, fix import 
This commits builds on DumoeDss's PR. It

- adds a default db directory name.
- adds a default db instance (Chroma).

Both points offers flexibility for users who
want to use default and users who want to
customize

Lastly, it fixes an import
 2023-06-23 11:51:51 +05:30
Sayo
85a6a0c161 [feat] Refactor VectorDB class hierarchy for flexibility 2023-06-22 12:15:26 +08:00
Taranjeet Singh
63bd18056f Revert "Rename query endpoint to qna" 
This reverts commit 6bc7e9b7a7.
 2023-06-21 16:06:51 +05:30
Taranjeet Singh
6bc7e9b7a7 Rename query endpoint to qna 
Query endpoint takes a query and returns an answer
on the basis of retrieved similar docs.
Thought about the name and realized that qna makes
much more sense than query, as it takes a query
and returns an answer
 2023-06-21 12:35:38 +05:30
Taranjeet Singh
8582613940 Add doc strings 
Adds doc string for functions in embedchain/embedchain.py
 2023-06-20 21:37:21 +05:30
Taranjeet Singh
ac0ec0bf8f Update log statement, doc and package dependency version 2023-06-20 21:04:33 +05:30
Taranjeet Singh
08f155a551 Update website to web page 
This commit renames the website loader, chunker
to web page, as it is loading and chunking a single
url than the complete website.
 2023-06-20 16:50:57 +05:30
Taranjeet Singh
468db83337 Add simple app functionality 
This commit enables anyone to create a app and add 3 types of data
sources:

* pdf file
* youtube video
* website

It exposes a function called query which first gets similar docs from
vector db and then passes it to LLM to get the final answer.
 2023-06-20 14:42:55 +05:30