124 Commits

Author SHA1 Message Date
evan-danswer
0c29743538
use max_tokens to do better rate limit handling (#4224)
* use max_tokens to do better rate limit handling

* fix unti tests

* address greptile comment, thanks greptile
2025-03-06 18:12:05 -08:00
evan-danswer
b7da91e3ae
improved basic search latency (#4186)
* improved basic search latency

* address PR comments + minor cleanup
2025-03-06 22:22:59 +00:00
Weves
fe246aecbb Attempt to address tool happy claude 2025-03-05 09:47:27 -08:00
evan-danswer
5ca898bde2
Force use tool overrides (#4024)
* initial rename + timeout bump

* querry override
2025-02-17 21:01:24 +00:00
joachim-danswer
86bd121806
no reranking if local model w/o GPU for Agent Search (#4011)
* no reranking if locql model w/o GPU

* more efficient gpu status calling

* fix unit tests

---------

Co-authored-by: Evan Lohn <evan@danswer.ai>
2025-02-17 14:13:24 +00:00
evan-danswer
9324f426c0
added timeouts for agent llm calls (#4019)
* added timeouts for agent llm calls

* timing suggestions in agent config

* improved timeout that actually exits early

* added new global timeout and connection timeout distinction

* fixed error raising bug and made entity extraction recoverable

* warnings and refactor

* mypy

---------

Co-authored-by: joachim-danswer <joachim@danswer.ai>
2025-02-17 07:02:19 +00:00
joachim-danswer
20d3efc86e
By default, use primary LLM for initial & refined answer (#4012)
* By default, use primary LLM for initial & refined answer

Use of new env variable

* simplification
2025-02-16 23:20:07 +00:00
evan-danswer
3ba65214b8
bump version and fix related issues (#3996) 2025-02-14 19:57:12 +00:00
joachim-danswer
6687d5d499
major Agent Search Updates (#3994) 2025-02-14 19:40:21 +00:00
Yuhong Sun
1a7aca06b9
Fix Agent Slowness (#3979) 2025-02-13 15:54:34 -08:00
joachim-danswer
667b9e04c5
updated rerank function arguments (#3988) 2025-02-13 14:13:14 -08:00
pablonyx
3bc4e0d12f
Very minor robustification (#3926)
* very minor robustification

* robust
2025-02-06 19:55:38 +00:00
pablodanswer
f137fc78a6 various UX improvements 2025-02-05 21:12:55 -08:00
evan-danswer
29f5f4edfa
fixed citations when sections selected (#3914)
* removed some dead code and fixed citations when a search request is made with sections selected

* fix black formatting issue
2025-02-05 22:16:07 +00:00
Yuhong Sun
49fd76b336
Tool Call Error Display (#3897) 2025-02-04 16:12:50 -08:00
Evan Lohn
b928201397 fixed rebase issue and some cleanup 2025-02-03 20:49:45 -08:00
Yuhong Sun
b500c914b0 cleanup 2025-02-03 20:10:51 -08:00
joachim-danswer
b46c09ac6c EL comments 2025-02-03 20:10:51 -08:00
joachim-danswer
3ce8923086 fix for citation update 2025-02-03 20:10:51 -08:00
joachim-danswer
7ac6d3ed50 logging level changes 2025-02-03 20:10:51 -08:00
joachim-danswer
3cd057d7a2 LangGraph comments 2025-02-03 20:10:51 -08:00
joachim-danswer
eb227c0acc nit update 2025-02-03 20:10:51 -08:00
joachim-danswer
25a57e2292 add title and meta-data to doc 2025-02-03 20:10:51 -08:00
Evan Lohn
3f6de7968a prompt improvements for wekaer models 2025-02-03 20:10:51 -08:00
Yuhong Sun
59c65a4192 prompts 2025-02-03 20:10:51 -08:00
Yuhong Sun
16265d27f5 k 2025-02-03 20:10:51 -08:00
Yuhong Sun
570fe43efb log level changes 2025-02-03 20:10:51 -08:00
Yuhong Sun
506a9f1b94 Yuhong 2025-02-03 20:10:51 -08:00
Yuhong Sun
a067b32467 Partial Prompt Updates (#3880) 2025-02-03 20:10:51 -08:00
joachim-danswer
e23dd0a3fa renames + fix of refined answer generation prompt 2025-02-03 20:10:51 -08:00
Evan Lohn
71304e4228 always persist in agent search 2025-02-03 20:10:51 -08:00
Evan Lohn
2adeaaeded loading object into model instead of json 2025-02-03 20:10:51 -08:00
Evan Lohn
a96728ff4d prompt piece optimizations 2025-02-03 20:10:51 -08:00
joachim-danswer
9438f9df05 removal of sone unused states/models 2025-02-03 20:10:51 -08:00
joachim-danswer
b90e0834a5 major renaming 2025-02-03 20:10:51 -08:00
Evan Lohn
29440f5482 alembic heads, basic citations, search pipeline state 2025-02-03 20:10:51 -08:00
Evan Lohn
5a95a5c9fd large number of PR comments addressed 2025-02-03 20:10:51 -08:00
Evan Lohn
118e8afbef reworked config to have logical structure 2025-02-03 20:10:51 -08:00
joachim-danswer
8342168658 initial variable renaming 2025-02-03 20:10:51 -08:00
joachim-danswer
d5661baf98 history summary fix
- adjusted prompt
 - adjusted citation removal
 - length cutoff by words, not characters
2025-02-03 20:10:51 -08:00
joachim-danswer
95fcc0019c history summary update 2025-02-03 20:10:51 -08:00
joachim-danswer
0ccd83e809 deep_search_a and agent_a_config renaming 2025-02-03 20:10:51 -08:00
joachim-danswer
732861a940 rename of documents to verified_reranked_documents 2025-02-03 20:10:51 -08:00
joachim-danswer
d53dd1e356 cited_docs -> cited_documents 2025-02-03 20:10:51 -08:00
joachim-danswer
1a2760edee improved logging through agent_state plus some default fixes 2025-02-03 20:10:51 -08:00
joachim-danswer
23ae4547ca default values of number of strings and other things 2025-02-03 20:10:51 -08:00
Evan Lohn
385b344a43 addressed TODOs 2025-02-03 20:10:51 -08:00
Evan Lohn
a340529de3 sync streaming impl 2025-02-03 20:10:51 -08:00
joachim-danswer
4a0b2a6c09 additional naming fixes 2025-02-03 20:10:51 -08:00
joachim-danswer
756a1cbf8f answer_refined_question_subgraphs 2025-02-03 20:10:51 -08:00