98 Commits

Author SHA1 Message Date
hagen-danswer
aa84846298
Connector deletion fix (#2293)
---------

Co-authored-by: Weves <chrisweaver101@gmail.com>
2024-09-01 23:32:20 -07:00
pablodanswer
c122be2f6a
More explicit Confluence Connector (#2289) 2024-09-01 20:35:29 -07:00
hagen-danswer
8d443ada5b
Integration tests (#2256)
* initial commit

* almost done

* finished 3 tests

* minor refactor

* built out initial permisison tests

* reworked test_deletion

* removed logging

* all original tests have been converted

* renamed user_groups to user_group

* mypy

* added test for doc set permissions

* unified naming for manager methods

* Refactored models and added new deletion test

* minor additions

* better logging+fixed input variables

* commented out failed tests

* Added readme

* readme update

* Added auth to IT

set auth_type to basic and require_email_verification to false

* Update run-it.yml

* used verify and added to readme

* added api key manager
2024-09-01 22:21:00 +00:00
josvdw
50c17438d5
Litellm bump (#2195)
* ran bump-pydantic

* replace root_validator with model_validator

* mostly working. some alternate assistant error. changed root_validator and typing_extensions

* working generation chat. changed type

* replacing .dict with .model_dump

* argument needed to bring model_dump up to parity with dict()

* fix a fewremaining issues -- working with llama and gpt

* updating requirements file

* more requirement updates

* more requirement updates

* fix to make search work

* return type fix:

* half way tpyes change

* fixes for mypy and pydantic:

* endpoint fix

* fix pydantic protected namespaces

* it works!

* removed unecessary None initializations

* better logging

* changed default values to empty lists

* mypy fixes

* fixed array defaulting

---------

Co-authored-by: hagen-danswer <hagen@danswer.ai>
2024-08-28 00:00:27 +00:00
pablodanswer
97ba71e1b3
Db search (#2235)
* k

* update enum imports

* add functional types + model swaps

* remove a log

* remove kv

* fully functional + robustified for kv swap

* validated with hosted + cloud

* ensure not updating current search settings when reindexing

* add instance check

* revert back to updating search settings (will need a slight refactor for endpoint)

* protect advanced config override1

* run pretty

* fix typing

* update typing

* remove unnecessary function

* update model name

* clearer interface names

* validated foreign key constaint

* proper migration

* squash

---------

Co-authored-by: Yuhong Sun <yuhongsun96@gmail.com>
2024-08-27 04:26:51 +00:00
Chris Weaver
a873fc6483
Fix Confluence freezing (#2239) 2024-08-26 19:44:01 -07:00
Chris Weaver
c0e1a02e8e
Add it on merge queue (#2112)
* Github action to run integration tests

* Improve

* Fix build

* Add pull

* Fix readiness script

* Add IT runner

* Add IT runner

* Add logs

* update

* Fix

* Fix path

* file path

* test

* fix

* fix

* fix

* test

* network

* fix

* cleanup

* fix

* test

* Fix downgrade

* Add OpenAI API key

* Add VESPA_HOST

* test pulling first

* Add API server host

* Cache tweak

* Fix pull/push settings:

* Stop pushing to latest tag

* test cache change

* test

* test

* test

* remove cache temporarily

* Fix

* Enable EE

* test

* Remove duplicate funcs

* add back build

* Update all

* Fix stop cmd

* Add to merge queue

* Cleanup image tag
2024-08-26 07:20:28 +00:00
hagen-danswer
205c3c3fc8
Combined the get document set endpoints (#2234)
* Combined the get document set endpoints

* removed unused function

* fixed permissioning for document sets
2024-08-25 19:02:27 +00:00
Chris Weaver
99db27d989
Add metadata for simple doc (#2212) 2024-08-22 12:30:28 -07:00
hagen-danswer
61b5bd569b
Reworked chunking to support mega chunks (#2032) 2024-08-14 22:18:53 -07:00
rkuo-danswer
e517f47a89
add send-message-simple-with-history endpoint to avoid… (#2101)
* add send-message-simple-with-history endpoint to support ramp. avoids bad json output in models and allows client to pass history in instead of maintaining it in our own session

* slightly better error checking

* addressing code review

* reject on any empty message

* update test naming
2024-08-12 03:33:52 +00:00
Nathan Schwerdfeger
c7e5b11c63
EE Connector Deletion Bugfix + Refactor (#2042)
---------

Co-authored-by: Weves <chrisweaver101@gmail.com>
2024-08-11 20:33:07 -07:00
Yuhong Sun
ce666f3320
Propagate Embedding Enum (#2108) 2024-08-11 12:17:54 -07:00
Yuhong Sun
d60fb15ad3
Allowing users to set Search Settings (#2106) 2024-08-10 20:48:58 -07:00
pablodanswer
0a8d44b44c
quote processing for lengthy intros (#2103) 2024-08-10 11:09:45 -07:00
pablodanswer
cc8a6da8e3
improve llm-generated citations (account for edge case) (#2096)
* improve llm-generated citations (account for edge case)

* additional test case
2024-08-10 02:06:39 +00:00
rkuo-danswer
be9ed319d5
add unit test for quotes (#2085)
* add unit test for quotes

* test answer and quotes together
2024-08-08 18:20:07 +00:00
pablodanswer
9eb48ca2c3 account for empty links + fix quote processing 2024-08-07 20:55:18 -07:00
Chris Weaver
bb7e1d6e55
Add integration tests for document set syncing (#1904) 2024-08-06 18:00:19 -07:00
rkuo-danswer
fcc4c30ead
don't skip the start of the json answer value (#2067) 2024-08-06 23:59:13 +00:00
pablodanswer
0261d689dc
Various Admin Page + User Flow Improvements (#1987) 2024-08-03 18:09:46 -07:00
Weves
51731ad0dd Fix issue where large docs/batches break openai embedding 2024-08-02 01:07:09 -07:00
hagen-danswer
5307d38472
Fixed tokenizer logic (#1986) 2024-07-31 09:59:45 -07:00
Yuhong Sun
036d5c737e
No Null Embeddings (#1982) 2024-07-30 19:54:49 -07:00
hagen-danswer
3938a053aa
Rework tokenizer (#1957) 2024-07-29 23:01:49 -07:00
rkuo-danswer
4a0a927a64
fix removed parameter in MediaWikiConnector (#1970) 2024-07-29 18:47:30 +00:00
pablodanswer
3aa0e0124b
Add new admin page (#1947)
* add admin page

* credential + typing fix

* rebase fix

* on add, cleaner buttons

* functional G + Ddrive

* organized auth sections

* update types and remove logs

* ccs -> connectors

* validated formik

* update styling + connector-handling logic

* udpate colors

* separate out hooks + util functions

* update to adhere to rest standards

* remove "todos"

* rebase

* copy + formatting + sidebar

* update statuses + configuration possibilities

* update interfaces to be clearer

* update indexing status page

* formatting

* address backend security + comments

* update font

* fix form routing

* fix hydration error

* add statuses, fix bugs, etc. (squash)

* fix color (squash)

* squash

* add functionality to sidebar

* disblae buttons if deleting

* add color

* minor copy + formatting updates
- on modify credential, close
- update copy for deletion of connectors

* fix build error

* copy

---------

Co-authored-by: Yuhong Sun <yuhongsun96@gmail.com>
2024-07-28 20:57:43 -07:00
Yuhong Sun
9651ea828b
Handling Metadata by Vector and Keyword (#1909) 2024-07-24 11:05:56 -07:00
hagen-danswer
6ff8e6c0ea
Improve eval pipeline qol (#1908) 2024-07-23 17:16:34 -07:00
hagen-danswer
a0b46c60c6
Switched eval api target back to oneshotqa (#1902) 2024-07-22 20:55:18 -07:00
hagen-danswer
6c88c0156c
Added file upload retry logic (#1889) 2024-07-22 13:13:22 -07:00
Yuhong Sun
2ba24b1734
Reenable Search Pipeline (#1886) 2024-07-21 10:33:29 -07:00
Yuhong Sun
44820b4909 k 2024-07-21 10:27:57 -07:00
Yuhong Sun
e95bfa0e0b
Suffix Test (#1880) 2024-07-20 15:54:55 -07:00
Yuhong Sun
4848b5f1de
Suffix Edits (#1878) 2024-07-20 13:59:14 -07:00
Yuhong Sun
38290057f2
Search Eval (#1873) 2024-07-19 16:48:58 -07:00
pablodanswer
87fadb07ea
COMPLETE USER EXPERIENCE OVERHAUL (#1822) 2024-07-17 19:44:21 -07:00
hagen-danswer
e93de602c3
Use SHA instead of branch and save more data (#1850) 2024-07-17 14:56:24 -07:00
hagen-danswer
b83f435bb0
Catch dropped eval questions and added multiprocessing (#1849) 2024-07-16 12:33:02 -07:00
hagen-danswer
a1e638a73d
Improved eval logging and stability (#1843) 2024-07-15 14:58:45 -07:00
Weves
0d52e99bd4 Improve confluence rate limiting 2024-07-14 16:40:45 -07:00
hagen-danswer
1b864a00e4
Added support for multiple Eval Pipeline UIs (#1830) 2024-07-14 15:16:20 -07:00
Yuhong Sun
e90c66c1b6
Include Titles in Chunks (#1817) 2024-07-12 09:42:24 -07:00
hagen-danswer
991ee79e47
some qol improvements for search pipeline (#1809) 2024-07-11 17:42:11 -07:00
Yuhong Sun
08c6e821e7
Merge Sections Logic (#1801) 2024-07-10 20:14:02 -07:00
pablodanswer
09a11b5e1a
Fix citations + unit tests (#1760) 2024-07-10 10:05:20 -07:00
Yuhong Sun
39982e5fdc
Info propagating to allow Chunk Merging (#1794) 2024-07-09 18:15:07 -07:00
Yuhong Sun
f0888f2f61
Eval Script Incremental Write (#1784) 2024-07-06 15:43:40 -07:00
Yuhong Sun
d35d7ee833
Evaluation Pipeline Touchup (#1783) 2024-07-06 13:17:05 -07:00
hagen-danswer
ac14369716
Added search quality testing pipeline (#1774) 2024-07-06 11:51:50 -07:00