Commit Graph

252 Commits

Author SHA1 Message Date
Timothy Carambat
13fb63930b
Improve RAG responses via source backfilling (#1477)
* Improve RAG responses via source backfilling

* Hide irrelevant citations from UI
2024-05-23 09:56:57 -07:00
Sean Hatfield
cc7e7fb3ac
[FEAT] Add support for gemini-1.5-flash-latest model (#1502)
* add support for gemini-1.5-flash-latest

* update comment in gemini LLM provider
2024-05-23 09:42:30 -07:00
timothycarambat
961630f7d5 fix example call param 2024-05-23 08:29:25 -07:00
timothycarambat
7bace20769 Improve VoyageAI error responses and textChunk handler
resolves #1491
2024-05-22 13:42:48 -05:00
Timothy Carambat
8c5a30db9d
Support dynamic context length - VoyageAI (#1489) 2024-05-22 10:53:49 -05:00
Shixian Sheng
a256db132d
Fixed links (#1485)
* Update CHROMA_SETUP.md

* Update ASTRA_SETUP.md
2024-05-22 10:06:39 -05:00
timothycarambat
9f327d015a update error handling for OpenAI providers 2024-05-22 09:58:10 -05:00
Timothy Carambat
b23cb1a90f
Improve RAG results via chunkHeader append (#1473) 2024-05-21 14:43:39 -05:00
Timothy Carambat
28eba636e9
Allow setting of safety thresholds for Gemini (#1466)
* Allow setting of safety thresholds for Gemini

* linting
2024-05-20 13:17:00 -05:00
Sean Hatfield
5bf4b4db58
[FEAT] Add support for Voyage AI embedder (#1401)
* add support for voyageai embedder

* remove unneeded import

* linting

* Add ENV examples
Update how chunks are processed for Voyage
use correct langchain import
Add data handling

---------

Co-authored-by: Timothy Carambat <rambat1010@gmail.com>
2024-05-19 13:20:23 -05:00
Timothy Carambat
9ace0e67e6
Validate max_tokens is number (#1445) 2024-05-17 21:44:55 -07:00
Timothy Carambat
1a5aacb001
Support multi-model whispers (#1444) 2024-05-17 21:31:29 -07:00
timothycarambat
3794ef8dfd patch: use workspace LLM for embed chats and fallback to system
resolves #1439
2024-05-17 17:33:19 -07:00
Timothy Carambat
f140139534
Patch WSS upgrade for manual HTTPS certs (#1429)
* Patch WSS upgrade for manual HTTPS certs

* update comment

* refactor
2024-05-17 14:03:25 -07:00
Timothy Carambat
cae6cee1b5
Do not go through LLM to embed when embedding documents (#1428) 2024-05-16 17:51:04 -07:00
Timothy Carambat
01cf2fed17
Make native embedder the fallback for all LLMs (#1427) 2024-05-16 17:25:05 -07:00
Sean Hatfield
826ef00da3
[FEAT] LiteLLM provider support (#1424)
* litellm LLM provider support

* fix lint error

* change import orders
fix issue with model retrieval

---------

Co-authored-by: Timothy Carambat <rambat1010@gmail.com>
2024-05-16 13:56:28 -07:00
Timothy Carambat
15cf921616
Support SQL Agent skill (#1411)
* Support SQL Agent skill

* add MSSQL agent connector

* Add frontend to agent skills
remove FAKE_DB mock
reset skills to pickup child-skill dynamically

* add prompt examples for tools on untooled

* add better logging on SQL agents

* Wipe toolruns on each chat relay so tools can be used within the same session

* update comments
2024-05-16 10:38:21 -07:00
Timothy Carambat
b6be43be95
Add Speech-to-text and Text-to-speech providers (#1394)
* Add Speech-to-text and Text-to-speech providers

* add files and update comment

* update comments

* patch: bad playerRef check
2024-05-14 11:57:21 -07:00
Timothy Carambat
64b62290d7
Set gpt-4o as default for OpenAI (#1391) 2024-05-13 14:31:49 -07:00
Sean Hatfield
9ed2309757
[FEAT] Add API key support for Oobabooga Web UI (#1354)
* add api key support for oobabooga web ui

* dont expose API Key for TextWebGenUi

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-13 12:58:16 -07:00
timothycarambat
a87978d1d9 Make LanceDB the vector database default provider in backend to prevent issues where somehow this key is not set by the user resulting in a Pinecone error even though they never said they wanted Pinecone to be their vector db 2024-05-13 12:22:53 -07:00
Sean Hatfield
948ac8a3dd
[FIX] Validate messages schema for gemini provider (#1351)
validate messages schema for gemini provider
2024-05-10 17:33:25 -07:00
Timothy Carambat
7b18a36288
prevent accidental lockout from restricted chars in single pass mode (#1352)
* prevent accidental lockout from restrict chars in single pass mode

* update error message
2024-05-10 17:29:49 -07:00
Timothy Carambat
98953d70b7
Fix typo leading to agent history in threads not being unique in same workspace (#1350) 2024-05-10 16:09:34 -07:00
Sean Hatfield
0a6a9e40c1
[FIX] Add max tokens field to generic OpenAI LLM connector (#1345)
* add max tokens field to generic openai llm connector

* add max_tokens property to generic openai agent provider
2024-05-10 14:49:02 -07:00
Sean Hatfield
d36c3ff8b2
[FEAT] Slash templates (#1314)
* WIP slash presets

* WIP slash command customization CRUD + validations complete

* backend slash command support

* fix permission setting on new slash commands
rework form submit and pattern on frontend

* Add field updates for hooks,
required=true to field
add user<>command constraint to keep them unique
enforce uniquness via teritary uid field on table for multi and non-multi user

* reset migration

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-10 12:35:33 -07:00
Timothy Carambat
81bc16cc39
More agent providers (#1316)
* add OpenRouter support

* add mistral agents
add perplexity agents
add textwebgenui agents
2024-05-08 16:04:18 -07:00
Sean Hatfield
8422f92542
Agent support for LLMs with no function calling (#1295)
* add LMStudio agent support (generic) support
"work" with non-tool callable LLMs, highly dependent on system specs

* add comments

* enable few-shot prompting per function for OSS models

* Add Agent support for Ollama models

* azure, groq, koboldcpp agent support complete + WIP togetherai

* WIP gemini agent support

* WIP gemini blocked and will not fix for now

* azure fix

* merge fix

* add localai agent support

* azure untooled agent support

* merge fix

* refactor implementation of several agent provideers

* update bad merge comment

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-08 15:17:54 -07:00
Sean Hatfield
977a07db86
[FEAT] Text Generation Web UI LLM provider support (#1279)
* add text gen web ui LLM provider support

* update README

* README typo

* update TextWebUI display name
patch workspace<>model support for provider

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-08 11:56:30 -07:00
Timothy Carambat
331d3741c9
Ollama agents (#1270)
* add LMStudio agent support (generic) support
"work" with non-tool callable LLMs, highly dependent on system specs

* add comments

* enable few-shot prompting per function for OSS models

* Add Agent support for Ollama models

* improve json parsing for ollama text responses
2024-05-07 18:06:31 -07:00
Timothy Carambat
1b4559f57f
Add LMStudio agent support (generic) support (#1246)
* add LMStudio agent support (generic) support
"work" with non-tool callable LLMs, highly dependent on system specs

* add comments

* enable few-shot prompting per function for OSS models
2024-05-07 16:35:47 -07:00
Sean Hatfield
fc77b46800
[FEAT] KoboldCPP LLM Support (#1268)
* koboldcpp LLM support

* update .env.examples for koboldcpp support

* update LLM preference order
update koboldcpp comments

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-02 12:12:44 -07:00
Sean Hatfield
3caebc47b4
[FEAT] Cohere LLM and embedder support (#1233)
* getChatCompletion working WIP streaming

* WIP

* working streaming WIP abort stream

* implement cohere embedder support

* remove inputType option from cohere embedder

* fix cohere LLM from not aborting stream when canceled by user

* Patch Cohere implemention

* add cohere to onboarding

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-02 10:35:50 -07:00
Sean Hatfield
d02013fd71
[FIX] Document pinning does not count in query mode (#1250)
* if document is pinned, do not give queryRefusalResponse message

* forgot embed.js patch

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-02 10:27:09 -07:00
Timothy Carambat
244ce2e307
Prevent concurrent downloads on first-doc upload (#1267) 2024-05-02 10:15:11 -07:00
Timothy Carambat
47b7df4fc3
Clear chat window on /reset (#1261)
clear chat window on /reset
2024-05-01 20:39:58 -07:00
Sean Hatfield
9feaad79cc
[CHORE] Remove sendChat and streamChat in all LLM providers (#1260)
* remove sendChat and streamChat functions/references in all LLM providers

* remove unused imports

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-05-01 16:52:28 -07:00
Timothy Carambat
894f727903
Remove restrictions on pinned documents to use more context (#1248)
* Remove restrictions on pinned documents to use more contet

* update comment
2024-05-01 13:32:52 -07:00
Timothy Carambat
42e1d8e8ce
Customize refusal response for query mode (#1243)
* Customize refusal response for `query` mode

* remove border for desktop
2024-04-30 16:14:30 -07:00
Timothy Carambat
547d4859ef
Bump openai package to latest (#1234)
* Bump `openai` package to latest
Tested all except localai

* bump LocalAI support with latest image

* add deprecation notice

* linting
2024-04-30 12:33:42 -07:00
Timothy Carambat
94017e2b51
bump langchain deps (#1231)
* bump langchain deps

* patch native and ollama providers remove deprecated deps

---------

Co-authored-by: shatfield4 <seanhatfield5@gmail.com>
2024-04-30 12:04:24 -07:00
Timothy Carambat
bf435b2861
Adjust how text is split depending on input type (#1238)
resolves #1230
2024-04-30 10:11:56 -07:00
Timothy Carambat
ca63012c0f
bump lancedb dep (#1229) 2024-04-29 09:52:22 -07:00
Timothy Carambat
1b35bcbeab
Strengthen field validations on user Updates (#1201)
* Strengthen field validations on user Updates

* update writables
2024-04-26 16:46:04 -07:00
timothycarambat
df2c01b176 patch OpenRouter model fetcher when key is not present 2024-04-26 15:58:30 -07:00
Timothy Carambat
2e813846dc
Agent skill: chart generation (#1103)
* WIP agent support

* move agent folder

* wip frontend socket

* checkpoint

* fix schema

* Checkpoint for plugins and AgentHandler

* refactor plugins and agent arch

* agent error reporting and handling

* add frontend elements for agents in prompt input

* WIP integrations for agents

* enable web-search agent config from frontend

* persist chat history

* update alert

* update migration
remove console logs
update close state for agent invocations

* add examples to dockerignore
Extract statusResponse to its own component

* update close method

* wrap scraping rejections

* add RAG search as funciton

* Add telem and link highlight

* chat support

* patch memory

* Add rechart as a plugin option

* Toggles for abilites of default agent (system wide)
Validate values for agent skills
Enable dynamic loading of skills
UI for toggle of skills

* add UI for toggle of configs for agent

* toggle WS or WSS protocol

* update NGNIX proxy pass

* move components around and capture failed websocket creation

* fix name

* tmp docker image

* reset workflow

* safety mark functions

* telem on tool calls

* remove hardcode short circuit

* separate web-browser from scrape

* extract summarizer to util
add abort handlers and controller for langchain stuff so socket close kills process

* langchain summarize verbose when in dev

* chart styling improvements + add title to chart

* fix legend from being cutoff in chart downloads

* remove cursor blink

---------

Co-authored-by: shatfield4 <seanhatfield5@gmail.com>
2024-04-26 11:18:55 -07:00
Sean Hatfield
11f6419c3c
[FEAT] Implement new login screen UI & multi-user password reset (#1074)
* WIP new login screen UI

* update prisma schema/create new models for pw recovery

* WIP password recovery backend

* WIP reset password flow

* WIP pw reset flow

* password reset logic complete & functional UI

* WIP login screen redesign for single and multi user

* create placeholder modal to display recovery codes

* implement UI for recovery code modals/download recovery codes

* multiuser desktop password reset UI/functionality complete

* support single user mode for pw reset

* mobile styles for all password reset/login flows complete

* lint

* remove single user password recovery

* create PasswordRecovery util file to make more readable

* do not drop-replace users table in migration

* review pr

---------

Co-authored-by: timothycarambat <rambat1010@gmail.com>
2024-04-25 16:52:30 -07:00
timothycarambat
dfaaf1680f update perplexity models
resolves #1188
2024-04-25 07:34:28 -07:00
Timothy Carambat
df17fbda36
Add generic OpenAI endpoint support (#1178)
* Add generic OpenAI endpoint support

* allow any input for model in case provider does not support models endpoint
2024-04-23 13:06:07 -07:00