rubra-ai / rubra Goto Github PK

Open Weight, tool-calling LLMs

License: Apache License 2.0

Shell 4.82% Dockerfile 1.02% Python 85.04% Jupyter Notebook 0.45% Makefile 0.70% Rust 5.24% HTML 0.44% JavaScript 1.72% CSS 0.55%

ai ai-agents function-calling large-language-models tools

rubra's People

Contributors

Stargazers

Watchers

Forkers

njhale cloudnautique lucj rohitg00 render-ai vmatev dnouv gmarkley-vi thedadams tylerslaton g-linville iwilltry42 funkyfisch mchorfa ganeshwhere debdutdeb zweahtet wahuh rocketchat

rubra's Issues

Not able to delete assistants.

Steps to reproduce the problem:

Created an assistant.
Deleted this assistant.
Assistant continues to be present in the Explore_Assistants page and is also listed as one of the available assistants in the Chat page and continues to be usable.

Expected Behavior:
Deleted this assistant should not be shown in Explore_Assistants and Chat page

Rubra chat with default model (custom) does not return results from browser

Steps to reproduce the problem:

Installed Rubra 0.2.0 using dmg.
Use Chat window , to ask "Who won super bowl 2024?"

Browser is not enabled for "custom" model.

Note - If i create my own assistant with browser option enabled, then i see browser integration working as expected.

Show rubra version info in UI

Improve user experience when attempting to start rubra when Docker Desktop is installed but not running.

Steps to reproduce the problem:

Install Rubra.
Stop Rubra
Quit Docker Desktop
Start Rubra

User gets stuck in this state forever:

Expected Behavior:
Message to prompt user to start Docker Desktop to be able to get rubra started would be good to have.

Can't get the knowledge assistant to work

Create a new assistant, and upload a text version of Google Q4 2023 earnings release. The resulting assistant always responds with "Sorry, I can't find any relevant information." regardless what questions I ask.

I tried uploading the original PDF version of the earnings release. It does not work either.
2023q4-alphabet-earnings-release.pdf
2023q4-alphabet-earnings-release.txt

Notification when the complete set of documents ingested has finished embedding generation

Not getting any meaningful response from Assistant when json file is uploaded.

Steps to reproduce the problem:

Create an assistant with following json file
myjson.txt
Once the assistant is done loading the file completely ( In my case this took a very long time > 1 hour)

024-02-13 11:47:19 [2024-02-13 19:47:19,616: INFO/ForkPoolWorker-9] processing file id : file_af27df
2024-02-13 12:54:33 [2024-02-13 20:54:33,665: WARNING/ForkPoolWorker-9] [447700193805348723, 447700193805348724, 447700193805348725, 447700193805348726, 447700193805348727, 447700193805348728, 447700193805348729, 447700193805348730, 447700193805348731, 447700193805348732, 447700193805348733, 447700193805348734, 447700193805348735, 447700193805348736, 447700193805348737, 447700193805348738, 447700193805348739, 447700193805348740, 447700193805348741, 447700193805348742, 447700193805348743, 447700193805348744, 447700193805348745, 447700193805348746, 447700193805348747, 447700193805348748, 447700193805348749, 447700193805348750, 447700193805348751, 447700193805348752, 447700193805348753, 447700193805348754, 447700193805348755, 447700193805348756, 447700193805348757, 447700193805348758, 447700193805348759, 447700193805348760, 447700193805348761, 447700193805348762, 447700193805348763, 447700193805348764, 447700193805348765, 447700193805348766, 447700193805348767, 447700193805348768, 447700193805348769, 447700193805348770, 447700193805348771, 447700193805348772, 447700193805348773, 447700193805348774, 447700193805348775, 447700193805348776, 447700193805348777, 447700193805348778, 447700193805348779, 447700193805348780, 447700193805348781, 447700193805348782, 447700193805348783, 447700193805348784, 447700193805348785, 447700193805348786, 447700193805348787, 447700193805348788, 447700193805348789, 447700193805348790, 447700193805348791, 447700193805348792, 447700193805348793, 447700193805348794, 447700193805348795, 447700193805348796, 447700193805348797, 447700193805348798, 447700193805348799, 447700193805348800, 447700193805348801, 447700193805348802, 447700193805348803, 447700193805348804, 447700193805348805, 447700193805348806, 447700193805348807, 447700193805348808, 447700193805348809, 447700193805348810, 447700193805348811, 447700193805348812, 447700193805348813, 447700193805348814, 447700193805348815, 447700193805348816, 447700193805348817, 447700193805348818, 447700193805348819, 447700193805348820, 447700193805348821, 447700193805348822, 447700193805348823, 447700193805348824, 447700193805348825, 447700193805348826, 447700193805348827, 447700193805348828, 447700193805348829, 447700193805348830, 447700193805348831, 447700193805348832, 447700193805348833, 447700193805348834, 447700193805348835, 447700193805348836, 447700193805348837, 447700193805348838, 447700193805348839, 447700193805348840, 447700193805348841, 447700193805348842, 447700193805348843, 447700193805348844, 447700193805348845, 447700193805348846, 447700193805348847, 447700193805348848, 447700193805348849, 447700193805348850, 447700193805348851, 447700193805348852, 447700193805348853, 447700193805348854, 447700193805348855, 447700193805348856, 447700193805348857, 447700193805348858, 447700193805348859, 447700193805348860, 447700193805348861, 447700193805348862, 447700193805348863, 447700193805348864, 447700193805348865, 447700193805348866, 447700193805348867, 447700193805348868, 447700193805348869, 447700193805348870, 447700193805348871, 447700193805348872, 447700193805348873, 447700193805348874, 447700193805348875, 447700193805348876, 447700193805348877, 447700193805348878, 447700193805348879, 447700193805348880, 447700193805348881, 447700193805348882, 447700193805348883, 447700193805348884, 447700193805348885, 447700193805348886, 447700193805348887, 447700193805348888, 447700193805348889, 447700193805348890, 447700193805348891, 447700193805348892, 447700193805348893, 447700193805348894, 447700193805348895, 447700193805348896, 447700193805348897, 447700193805348898, 447700193805348899, 447700193805348900, 447700193805348901, 447700193805348902, 447700193805348903, 447700193805348904, 447700193805348905, 447700193805348906, 447700193805348907, 447700193805348908, 447700193805348909, 447700193805348910]
2024-02-13 12:54:33 [2024-02-13 20:54:33,665: INFO/ForkPoolWorker-9] file file_af27df processing completed
2024-02-13 12:54:33 [2024-02-13 20:54:33,669: INFO/ForkPoolWorker-9] Task app.tasks.execute_asst_file_create[6c2b4077-3eb8-4f9f-a51c-f01e66e150e7] succeeded in 4034.0901564619853s: None

Not getting any meaningful response from Assistant

Another simple one:

List all apps in running state

I'm sorry, but I don't have the information about all running apps. The provided JSON file only contains a single app called 'FileKnowledge'. If you need to know more about that specific app, please ask for its details.

knowledge retrieval embedding model is taking long to index.

Creating an Assistant with files takes very long time to complete loading the files.
In my case, it took about 8 minutes for loading a PDF document with size 6.5 MB

This is my Mac configration:

From logs:

2024-02-14 15:17:41 [2024-02-14 23:17:41,913: INFO/ForkPoolWorker-8] Task app.tasks.execute_chat_completion[ec15a362-aba8-4405-84b2-0adce83bf509] succeeded in 11.899923254997702s: None
2024-02-14 15:22:55 [2024-02-14 23:22:55,504: INFO/MainProcess] Task app.tasks.execute_asst_file_create[341a9ecc-bf19-4e6f-ab37-6071404c74f8] received
2024-02-14 15:22:55 [2024-02-14 23:22:55,515: INFO/ForkPoolWorker-8] processing file id : file_176e8b
2024-02-14 15:31:02 [2024-02-14 23:31:02,488: WARNING/ForkPoolWorker-8] [447725520764156016, 447725520764156017, 447725520764156018, 447725520764156019, 447725520764156020, 447725520764156021, 447725520764156022, 447725520764156023, 447725520764156024, 447725520764156025, 447725520764156026, 447725520764156027, 447725520764156028, 447725520764156029, 447725520764156030, 447725520764156031, 447725520764156032, 447725520764156033, 447725520764156034, 447725520764156035, 447725520764156036, 447725520764156037, 447725520764156038]
2024-02-14 15:31:02 [2024-02-14 23:31:02,490: INFO/ForkPoolWorker-8] file file_176e8b processing completed
2024-02-14 15:31:02 [2024-02-14 23:31:02,495: INFO/ForkPoolWorker-8] Task app.tasks.execute_asst_file_create[341a9ecc-bf19-4e6f-ab37-6071404c74f8] succeeded in 486.98781755700475s: None

Application image does not show up in `about` screen and in bottom application tab.

dmg used - v0.0.2

Application image does not show up in about screen . It shows up as expected in the top menu bar.

Rubra icon is missing in the bottom tool bar.

Also clicking on "Rubra icon" in the bottom application tab is not launching the "Rubra System Dashboard"

Sometimes we see an incorrect query being used for google search (resulting in incorrect chat response) by Assistants with Browser capability.

Create an assistant with Browser enabled with custom model.
Sometimes I get an incorrect response from this assistant.

In this case case I asked who won 2024 Super bowl ? and it gave me the following result where the "quick search" link indicates that it asked google - Who will play the 2024 Super Bowl? .

After a [quick search](https://www.google.com/search?q=Who+will+play+the+2024+Super+Bowl%3F), here's what I discovered:
The 2024 Super Bowl, also known as Super Bowl LVIII, will feature the Kansas City Chiefs and the San Francisco 49ers.

When I asked the same question again, it responds with right answer and in this case we see the right query being passed for google search.

Provide a way to uninstall rubra.

Currently I have to do the following steps to uninstall rubra.

curl -sfL https://get.rubra.ai/ | sh -s -- stop 
 rm -r ~/.rubra
docker volume rm  rubra_mongodb rubra_milvus rubra_etcd

We need to support uninstall for cli and dmg install.

The assistant creation UI should specify what file types can be uploaded

It seems only txt and pdf files are supported. However the system allows me to upload files of any type. Not sure what happens afterwards.

Chat results are not able to get the desired information in some cases with a message stating - "the provided search results do not specifically mention"

Steps to reproduce the problem:

Create assistant with a file that also has some details mentioned in the "Out of Scope" section

# Out of Scope

**NOTE:** everything below is OUT OF SCOPE, but keeping it here for future reference

- Org admins can:
< List of things that org admins can do>

Chat results are not able to get the desired information when asking for information about out of scope items.

Modifying an assistant doesn't handle files

When you modify an assistant, it doesn't remember previously associated files. It also doesn't allow re-uploading them.

The expectation is that as prompts are being adjusted the files and data it has been associated with would still be used.

Support html in knowledge retrieval

Install fails on MacOS Sonoma

Install fails with a complaint about MacOS command line tools. I had to manually do a xcode-select --install to get it to work.

Improve user experience on opening the install file.

dmg used - v0.0.2

Opening the dmg file , show rubra app with any icon and has an icon file.

We should be able to provide rubra app with icon and also instruct users to move to Applications folder .

Show progress status for dmg installs when start/stop rubra is in progress.

dmg used - v0.0.2

Steps to reproduce the problem:

When installing Rubra , once the model is downloaded fully , we see the following page with Start Rubra greyed out.

This page is shown until the installation succeeds and user is presented with Stop Rubra and Launch Ui page.

During this time when Start Rubra option is greyed out , we should be able to present a better message/progress bar indicating that installation is in progress.

The same progress bar/message should also be presented when user uses the "Stop Rubra" option to stop rubra and "Start Rubra" option is used to start rubra.

This the page presented after using Start Rubra . Notice that the Start Rubra is still active (which needs to be greyed out) and show some message / progress bar when start is in progress.

This the page presented after using Stop Rubra . Notice that the Stop Rubra and launch Ui is still active (which needs to be greyed out) and show some message / progress bar when stop is in progress.

Not able to display strings which have ":" in chat window properly.

Steps to reproduce the problem:

Request for query that has ":" like "Get info on image nginx:v0.0.1"

This gets displayed with ":v0" lost and a new line introduced between the strings

Note - Query sent to LLM does not have any character lost . Issue is seen only in the chat window.

Not able to see "Welcome Assistant"

Build from https://get.rubra.ai
Steps to reproduce the problem

Uninstall/delete existing rubra installs ( curl -sfL https://get.rubra.ai | sh -s -- uninstall or curl -sfL https://get.rubra.ai | sh -s -- delete
Install rubra
curl -sfL https://get.rubra.ai | sh -s -- start

Once app is installed successfully and UI is launched , notice that the "Welcome Assistant" is not available as an option in the "Select a Model or an Assistant" in Chat window.

Expected Behavior:
"Welcome Assistant" should be available on install.

Indicate the state that the model is still loading when it is not completed loading and dont make it available to be used for chats.

Steps to reproduce the problem:

Create an Assistant.
Use this assistant for chat.

Notice that the first few chats always return - "Sorry, I cant find relevant information". This is because the model is still loading.
But once the model is loaded completely , we see the chat responses start showing up as expected.

Expected Behavior:

Indicate the state that the model is still loading when it is not completed loading
Dont make the assistant available to be used for chats.

Not able to launch Rubra app from tool bar once the app window is closed.

dmg used - v0.0.2

Steps to reproduce the problem:

lnstall app.
Once installed , close the Rubra app page (using "x").
Try to click on the Rubra app in the bottom tool bar.
Nothing happens.
Note - Clicking on Rubra app in Menu bar (in the top) still works.

The only way to make it start working is to use the "quit" option and relaunch.

web_browse tool doesn't parse content from url completely.

For example, the return scraped web content of url https://today-date.com is:

Today’s date in numbers » Today-date.com
[Skip to content](https://today-date.com#content "Skip to content") 
[![Today-date.com](https://today-date.com/wp-content/uploads/today-date-logo_03.jpg)](https://today-date.com/)
 
Menu 
* [Today’s date](https://today-date.com/ "What is today’s date?")
        + [Days left in year](https://today-date.com/days-left-in-year/ "Days left in year")
        + [Days of the week](https://today-date.com/days-of-the-week/)
        + [How many days are in a month?](https://today-date.com/how-many-days-are-in-a-month-2/)
        + [Military date format for today](https://today-date.com/military-date-format-for-today/)
        + [Next leap year](https://today-date.com/next-leap-year/)
        + [Today’s date Javascript widget](https://today-date.com/todays-date-javascript-widget/)
        + [Week number](https://today-date.com/week-number/ "This weeks number")
        + [Today’s date in different countries](https://today-date.com/todays-date-in-different-countries/)
* [Yesterday’s date](https://today-date.com/yesterdays-date/)
* [Tomorrow’s date](https://today-date.com/tomorrows-date/)
* [Current time](https://today-date.com/current-time/)
* [Days from today calculator](https://today-date.com/days-from-today/)
* [Date weeks from now](https://today-date.com/date-weeks-from-now/)
* [Date calculator](https://today-date.com/date-calculator/)
* [Calendar](https://today-date.com/calendar/)
 
Today’s date in numbers
=======================
 
What is today’s date?
---------------------
Find out the **current day**, **date of today,** [local time](https://today-date.com/current-time/) and [week number](https://today-date.com/week-number/).
NEW! [Try our days from today calculator](https://today-date.com/days-from-today/ "Days from today calculator")…
### The correct way of writing today’s date?
There are two ways to write a date:
1. using entirely numbers and
2. using a combination of words and numbers.
Then there are also **two standards** for writing today’s date:
* the **American English** standard  
(month-day-year)
* the **British English** standard  
(day-month-year)
  
### Writing the date including weekday
The weekday always comes first and is separated from the rest by a comma:
* Weekday, Month Date (date in digits), 2019 **(American)**
* Weekday, Date (date in digits) Month 2019 **(British)**  
( Please note there is no comma between the date and the year. )
### Different standards can cause confusion
If you write the **date with only digits** it can easily lead to misunderstanding.   
For example the date 10/03/2020 can be either:
1. October 3, 2020 (American interpretation)
2. March 10 2020 (British interpretation)
This is a reason why it can be better to **write the name of the month**.
### Abbrevations for day of the Week
1. Sunday – Sun.
2. Monday – Mon.
3. Tuesday – Tue. or Tues.
4. Wednesday – Wed.
5. Thursday – Thu.
6. Friday – Fri.
7. Saturday – Sat.
### Abbrevations for Months
1. January – Jan.
2. February – Feb.
3. March – Mar.
4. April – Apr.
5. May – May
6. June – Jun.
7. July – Jul.
8. August – Aug.
9. September – Sep.
10. October – Oct.
11. November – Nov.
12. December – Dec.
[What was yesterday’s date?](https://today-date.com/yesterdays-date/) / [What is tomorrow’s date?](https://today-date.com/tomorrows-date/)  
   
[Next leap year?](https://today-date.com/next-leap-year/) / [Date two weeks from now?](https://today-date.com/date-two-weeks-from-now/)
[![Today's date](https://today-date.com/wp-content/uploads/todays-date.jpg)](https://today-date.com)
Today’s date in numbers and slashes
When writing the date by numbers only, you can separate them by using   
a **hyphen** (-), a **slash** (/), or a **dot** (.) 
You can write the **date with or without leading zeros**. For example:
* 10/03/2020 **or** 10/3/2020
* [Privacy policy](https://today-date.com/privacy-policy/)
* [Today’s date – the date of today](https://today-date.com/)
* [Days from today calculator](https://today-date.com/days-from-today/)
        + [10 days from today](https://today-date.com/10-days-from-today/)
        + [28 days from today](https://today-date.com/28-days-from-today/)
        + [29 days from today](https://today-date.com/29-days-from-today/)
        + [180 days from today](https://today-date.com/180-days-from-today/)
* [Calendar for this month](https://today-date.com/calendar-old/)
* [Current time](https://today-date.com/current-time/)
* [Date calculator](https://today-date.com/date-calculator/)
* [Date weeks from now](https://today-date.com/date-weeks-from-now/)
* [Days left in year](https://today-date.com/days-left-in-year/)
* [Days of the week](https://today-date.com/days-of-the-week/)
* [How many days are in a month?](https://today-date.com/how-many-days-are-in-a-month-2/)
* [Military date format for today](https://today-date.com/military-date-format-for-today/)
* [Next leap year](https://today-date.com/next-leap-year/)
* [Today’s date Javascript widget](https://today-date.com/todays-date-javascript-widget/)
* [Week number](https://today-date.com/week-number/)
* [What is tomorrow’s date?](https://today-date.com/tomorrows-date/)
* [Yesterday’s date](https://today-date.com/yesterdays-date/)
 
 
© 2024 Today-date.com • Built with [GeneratePress](https://generatepress.com)

the most important information --> today's date, is missing. Which results in web browsing agent's bad performance for this kind of queries.

When Assistant has "Browser" option enabled , I am not able to get chat responses that use the files uploaded to the Assistant.

Created an assistant with "Browser" option enabled and uploaded a file containing some information about a specific feature.

When asked for queries relating to the feature to this assistant , the responses I get are from the Internet . I did not get any responses specific to the uploaded file.

Note - If I created an assistant without "Browser" option , then I am able to get results from the uploaded file.

Allowed to create assistant with duplicate names.

We are allowed to create assistant with duplicate names.

This results in duplicate entries in Assistant drop down in Rubra chat window

Expected Behavior:
We should not be allowed to create multiple assistant with same name.

Feature request: ability to clone agents

Would like a button that would let me clone an agent so that I can modify the clone while leaving the original in place. This is actually just a hacky solution to a more extensive feature, which is basically keeping a history of changes to an agent so that I can experiment and rollback if I don't like the results, but I think this clone button would be quick hit.

Support JSON in knowledge retrieval

Rate limited by google search resulting in no chat responses from assistants that have browser option enabled.

I got rate limited when trying to chat with an assistant that had browser option enabled with the following query.

You are a very experienced travel agent with a focus on affordable vacations for your clients.

Before starting your job, do a quick search (just one search call) to refresh your knowledge on
what it means to be a travel agent. Use this context to guide how you evaluate the following steps.

Based on the input, do the following in order:
1. Search the web for typical vacation routes in the France.
2. Based on the results build an initial outline of locations to include.
3. For each location you determine, search for essential things to do in that location (maximum of one search per location). Include at
   least 5 activities per day and 20 per location.
4. With all of the activities and locations, build out an itenerary that outlines each day and each hour in that day for the trip.
5. Reevaluate the plan and move dates around such that it is optimized for effecient travel
6. Look over the entire thing one more time and ask yourself if its missing anything. If it is, make your edits now.

After this I have not been able get any results for any of the chats from any assistant that has browser option enabled.

Following is what I get as response:

Logs from task executor indicating google rate limiting the calls:

2024-02-20 10:33:48 [2024-02-20 18:33:48,058: WARNING/ForkPoolWorker-8] num of functions: 6
2024-02-20 10:33:48 [2024-02-20 18:33:48,058: WARNING/ForkPoolWorker-8] {'content': '', 'role': 'assistant', 'tool_calls': [{'index': 0, 'id': 'call_kClvqmc0GFAaF9VknDqTDdgF', 'function': {'arguments': '{"query": "typical vacation routes in France on March 1 for two weeks"}', 'name': 'GoogleSearchTool'}, 'type': 'function'}, {'index': 0, 'id': 'call_dA2vTHDHH6OGFYz6KDQ03epS', 'function': {'arguments': '{"query": "things to do in Paris"}', 'name': 'GoogleSearchTool'}, 'type': 'function'}, {'index': 0, 'id': 'call_0hGWKhLtW5a9ax4vlGPqOSld', 'function': {'arguments': '{"query": "things to do in Nice"}', 'name': 'GoogleSearchTool'}, 'type': 'function'}, {'index': 0, 'id': 'call_CIHnFSYUO2ZvGRkymrPQSFz2', 'function': {'arguments': '{"query": "things to do in Lyon"}', 'name': 'GoogleSearchTool'}, 'type': 'function'}, {'index': 0, 'id': 'call_9poGes3QGutLmc63PWcx8tLw', 'function': {'arguments': '{"query": "things to do in Marseille"}', 'name': 'GoogleSearchTool'}, 'type': 'function'}, {'index': 0, 'id': 'call_stgLUn8x1TvX6jQyLp3If6Tw', 'function': {'arguments': '{"query": "things to do in Bordeaux"}', 'name': 'GoogleSearchTool'}, 'type': 'function'}]}
2024-02-20 10:35:48 [2024-02-20 18:35:48,914: WARNING/ForkPoolWorker-8] Error in execute_chat_completion: 429 Client Error: Too Many Requests for url: https://www.google.com/sorry/index?continue=https://www.google.com/search%3Fq%3Dthings%252Bto%252Bdo%252Bin%252BParis%26num%3D3%26hl%3Den%26start%3D9&hl=en&q=EgRjaTjDGITi064GIjADVPqL93bhFZUYe1TqwFS01QpA6PP5-TjnUmU-InXGfZm3OtNr-rO5sM4OF_DVKLcyAXJaAUM
2024-02-20 10:35:48 [2024-02-20 18:35:48,919: ERROR/ForkPoolWorker-8] Task app.tasks.execute_chat_completion[87c7ae75-5e88-4dd6-bb38-fd41a24a96cc] raised unexpected: HTTPError('429 Client Error: Too Many Requests for url: https://www.google.com/sorry/index?continue=https://www.google.com/search%3Fq%3Dthings%252Bto%252Bdo%252Bin%252BParis%26num%3D3%26hl%3Den%26start%3D9&hl=en&q=EgRjaTjDGITi064GIjADVPqL93bhFZUYe1TqwFS01QpA6PP5-TjnUmU-InXGfZm3OtNr-rO5sM4OF_DVKLcyAXJaAUM')
2024-02-20 10:35:48 Traceback (most recent call last):
2024-02-20 10:35:48   File "/usr/local/lib/python3.10/site-packages/celery/app/trace.py", line 477, in trace_task
2024-02-20 10:35:48     R = retval = fun(*args, **kwargs)
2024-02-20 10:35:48   File "/usr/local/lib/python3.10/site-packages/celery/app/trace.py", line 760, in __protected_call__
2024-02-20 10:35:48     return self.run(*args, **kwargs)
2024-02-20 10:35:48   File "/app/app/tasks.py", line 341, in execute_chat_completion
2024-02-20 10:35:48     function_response = available_function[ftc["name"]](
2024-02-20 10:35:48   File "/app/app/tools/web_browse_tool/web_browse_tool.py", line 35, in _run
2024-02-20 10:35:48     res = google_search_api(query)
2024-02-20 10:35:48   File "/app/app/tools/web_browse_tool/web_browse_tool.py", line 55, in google_search_api
2024-02-20 10:35:48     res = [
2024-02-20 10:35:48   File "/app/app/tools/web_browse_tool/web_browse_tool.py", line 55, in <listcomp>
2024-02-20 10:35:48     res = [
2024-02-20 10:35:48   File "/usr/local/lib/python3.10/site-packages/googlesearch/__init__.py", line 54, in search
2024-02-20 10:35:48     resp = _req(escaped_term, num_results - start,
2024-02-20 10:35:48   File "/usr/local/lib/python3.10/site-packages/googlesearch/__init__.py", line 23, in _req
2024-02-20 10:35:48     resp.raise_for_status()
2024-02-20 10:35:48   File "/usr/local/lib/python3.10/site-packages/requests/models.py", line 1021, in raise_for_status
2024-02-20 10:35:48     raise HTTPError(http_error_msg, response=self)
2024-02-20 10:35:48 requests.exceptions.HTTPError: 429 Client Error: Too Many Requests for url: https://www.google.com/sorry/index?continue=https://www.google.com/search%3Fq%3Dthings%252Bto%252Bdo%252Bin%252BParis%26num%3D3%26hl%3Den%26start%3D9&hl=en&q=EgRjaTjDGITi064GIjADVPqL93bhFZUYe1TqwFS01QpA6PP5-TjnUmU-InXGfZm3OtNr-rO5sM4OF_DVKLcyAXJaAUM
2024-02-20 11:03:38 [2024-02-20 19:03:38,155: INFO/MainProcess] Task app.tasks.execute_chat_completion[ce51eb94-463b-45bd-907e-9dcb5c962408] received

Expected Behavior:

Give an indication to the user that we did hit a rate limit issue instead of presenting no results or Rubra Backend Error: Failed to process function. as response.
There should be a way to get around this problem by being able to use a different search provider.

When modifying the assistant, getting a list of the associated documents in its 'context'

When modifying the assistant, permitting files to be added and deleted

Support code interpreter

Assistants from previous installs are still listed and usable on fresh installs.

Steps to reproduce the problem:

Install rubra - curl -sfL Ihttps://raw.githubusercontent.com/acorn-io/rubra/main/quickstart.sh | sh -s -- start
Create assistants.
Uninstall rubra

curl -sfL https://raw.githubusercontent.com/acorn-io/rubra/main/quickstart.sh | sh -s -- stop
rm -r ~/.rubra

Install rubra - curl -sfL Ihttps://raw.githubusercontent.com/acorn-io/rubra/main/quickstart.sh | sh -s -- start
After installation, notice that all the assistants. created in step2 are still present and usable.

Expected Behavior:

There should be no assistants available on installs done after uninstall of rubra

Related issue - #32

Modifying assistant results in "Update Assistant" entry presented in the top of the assistant list and feels like no action was taken.

Expected Behavior:
On Modifying existing Assistant, the "Update Assistant" entry should be presented the user .
Currently, "Update Assistant" entry is presented in the top of the assistant list and user is not taken to this update assistant entry.

Navigate out of the chat window when search is in progress and get back to the window , search results are not displayed.

Steps to reproduce the problem:

Start a chat
Navigate out of the chat window (to "Explore Assistants" / "Settings") when search is still in progress

Navigate back to the chat window
Response for the chat is not displayed in the window (even when the server logs indicate that a response was received)

Expected Results:
Response for the chat should be displayed in the window on re entry.

Not able to install rubra with install script from quick start guide.

Try to install rubra using - curl -sfL https://get.rubra.ai | sh -s -- start from quick start guide - https://docs.rubra.ai/quickstart/

Installation does not succeed.

...
..
 ✔ Network rubra                                                                                                                                          Created0.1s 
 ✔ Volume "rubra_etcd"                                                                                                                                    Created0.0s 
 ✔ Volume "rubra_mongodb"                                                                                                                                 Created0.0s 
 ✔ Volume "rubra_milvus"                                                                                                                                  Created0.0s 
 ✔ Container text-embedding-api                                                                                                                           Started0.3s 
 ✔ Container milvus                                                                                                                                       Created0.3s 
 ✔ Container vector-db-api                                                                                                                                Started0.3s 
 ✔ Container redis                                                                                                                                        Started0.3s 
 ✔ Container litellm                                                                                                                                      Started0.3s 
 ✔ Container mongodb                                                                                                                                      Started0.3s 
 ! litellm The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested        0.0s 
 ✔ Container task-executor                                                                                                                                Started0.1s  ✔ Container api-server                                                                                                                                   Created0.0s 
 ✔ Container ui                                                                                                                                           Created0.0s 
Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error mounting "/host_mnt/Users/sangeethahariharan/.rubra/milvus.yaml" to rootfs at "/milvus/configs/milvus.yaml": mount /host_mnt/Users/sangeethahariharan/.rubra/milvus.yaml:/milvus/configs/milvus.yaml (via /proc/self/fd/6), flags: 0x5000: not a directory: unknown: Are you trying to mount a directory onto a file (or vice-versa)? Check if the specified host path exists and is the expected type
[ERROR] Failed to start Docker containers

Chat doesn't seem to consider chat history

Getting response from browser takes a long time

Steps to reproduce the problem:

Installed Rubra 0.2.0 using dmg.
Create an assistant with browser option enabled using custom model.
Using this assistant , ask "Who won super bowl 2024?"

I get the results as expected , but it takes a very long time to respond.
In this case , it took close to 1 minute to respond.

024-02-17 09:25:00 [2024-02-17 17:25:00,795: WARNING/ForkPoolWorker-8] Assistant message created:
2024-02-17 09:25:00 [2024-02-17 17:25:00,795: WARNING/ForkPoolWorker-8]  
2024-02-17 09:25:00 [2024-02-17 17:25:00,795: WARNING/ForkPoolWorker-8] I did a [quick search](https://www.google.com/search?q=who+owns+the+rights+to+super+bowl+2024), here's what I discovered:
2024-02-17 09:25:00 
2024-02-17 09:25:00 Based on the provided search results, there is a discrepancy in the information. CBS Sports and NBC are both mentioned as owning the rights to broadcast the 2024 Super Bowl. However, they are under the same parent company, which is now known as Paramount Global (formerly ViacomCBS).
2024-02-17 09:25:00 [2024-02-17 17:25:00,797: INFO/ForkPoolWorker-8] Task app.tasks.execute_chat_completion[e5e3a7b9-8c8d-4152-8129-9e43be8615a7] succeeded in 182.58645520799996s: None
2024-02-17 09:25:41 [2024-02-17 17:25:41,575: INFO/MainProcess] Task app.tasks.execute_chat_completion[719a2840-dff7-40ca-bc67-9bf3c4926ac1] received
2024-02-17 09:25:41 [2024-02-17 17:25:41,585: WARNING/ForkPoolWorker-8] Calling model:
2024-02-17 09:25:41 [2024-02-17 17:25:41,585: WARNING/ForkPoolWorker-8]  
2024-02-17 09:25:41 [2024-02-17 17:25:41,585: WARNING/ForkPoolWorker-8] custom
2024-02-17 09:25:41 [2024-02-17 17:25:41,763: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:25:44 [2024-02-17 17:25:44,109: INFO/ForkPoolWorker-8] =====function call========
2024-02-17 09:25:44 [2024-02-17 17:25:44,110: INFO/ForkPoolWorker-8] Calling function : GoogleSearchTool with args: {'query': 'who won the 2024 super bowl'}
2024-02-17 09:25:44 [2024-02-17 17:25:44,114: INFO/ForkPoolWorker-8] Date detected in query, using relative date enhancement.
2024-02-17 09:25:45 [2024-02-17 17:25:45,673: INFO/ForkPoolWorker-8] =========visited url:========== https://en.wikipedia.org/wiki/Super_Bowl_LVIII
2024-02-17 09:25:51 [2024-02-17 17:25:51,398: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:25:58 [2024-02-17 17:25:58,050: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:03 [2024-02-17 17:26:03,871: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:07 [2024-02-17 17:26:07,416: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:12 [2024-02-17 17:26:12,278: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:16 [2024-02-17 17:26:16,814: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:21 [2024-02-17 17:26:21,904: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:26 [2024-02-17 17:26:26,561: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:30 [2024-02-17 17:26:30,575: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:35 [2024-02-17 17:26:35,727: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:37 [2024-02-17 17:26:37,002: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:37 [2024-02-17 17:26:37,002: INFO/ForkPoolWorker-8] ======PAGE_QA:======
2024-02-17 09:26:37  The Kansas City Chiefs won the 2024 Super Bowl, according to the search results.
2024-02-17 09:26:37 
2024-02-17 09:26:37 [2024-02-17 17:26:37,721: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-17 09:26:37 [2024-02-17 17:26:37,721: INFO/ForkPoolWorker-8] ======IS_SUFFICIENT:======
2024-02-17 09:26:37  Yes
2024-02-17 09:26:37 
2024-02-17 09:26:37 [2024-02-17 17:26:37,722: WARNING/ForkPoolWorker-8] message: After a [quick search](https://www.google.com/search?q=who+won+the+2024+super+bowl), here's what I discovered:
2024-02-17 09:26:37 
2024-02-17 09:26:37 The Kansas City Chiefs won the 2024 Super Bowl, according to the search results.
2024-02-17 09:26:37 [2024-02-17 17:26:37,723: WARNING/ForkPoolWorker-8] Assistant message created:
2024-02-17 09:26:37 [2024-02-17 17:26:37,723: WARNING/ForkPoolWorker-8]  
2024-02-17 09:26:37 [2024-02-17 17:26:37,723: WARNING/ForkPoolWorker-8] After a [quick search](https://www.google.com/search?q=who+won+the+2024+super+bowl), here's what I discovered:
2024-02-17 09:26:37 
2024-02-17 09:26:37 The Kansas City Chiefs won the 2024 Super Bowl, according to the search results.
2024-02-17 09:26:37 [2024-02-17 17:26:37,724: INFO/ForkPoolWorker-8] Task app.tasks.execute_chat_completion[719a2840-dff7-40ca-bc67-9bf3c4926ac1] succeeded in 56.147755276000225s: None

Selected Assistant always resets to "Welcome Assistant" when switching between pages.

Steps to reproduce the problem:

Create an assistant.
Go to Chat , Use this assistant and initiate some conversation.

Go to Explore Settings ( or Settings) and come back to Chat.
Notice that the Selected Assistant is set to "Welcome Assistant"

There is a hard limit of 20 assistants after which we assistants that are created are not listed and not usable for chats.

Steps to reproduce the problem:

Create 20 assistant.
Create 1 more assistant.
This assistant is not getting listed in the "chat" or "Explore Assistants" page

Expected Behavior:
Should be able to list the assistant created and use it for chat.

Can't get the knowledge assistant to work with html file .

Steps to reproduce the problem:

Create an assistant with html file (Do not enable Browser)
I got html for - https://github.com/acorn-io/runtime/issues ( from view page source)
Wait for the file to be fully loaded.
Took a long time about 41 minutes for loading to be done.

2024-02-13 10:54:34 [2024-02-13 18:54:34,495: WARNING/ForkPoolWorker-1] [447700193805348600, 447700193805348601, 447700193805348602, 447700193805348603, 447700193805348604, 447700193805348605, 447700193805348606, 447700193805348607, 447700193805348608, 447700193805348609, 447700193805348610, 447700193805348611, 447700193805348612, 447700193805348613, 447700193805348614, 447700193805348615, 447700193805348616, 447700193805348617, 447700193805348618, 447700193805348619, 447700193805348620, 447700193805348621, 447700193805348622, 447700193805348623, 447700193805348624, 447700193805348625, 447700193805348626, 447700193805348627, 447700193805348628, 447700193805348629, 447700193805348630, 447700193805348631, 447700193805348632, 447700193805348633, 447700193805348634, 447700193805348635, 447700193805348636, 447700193805348637, 447700193805348638, 447700193805348639, 447700193805348640, 447700193805348641, 447700193805348642, 447700193805348643, 447700193805348644, 447700193805348645, 447700193805348646, 447700193805348647, 447700193805348648, 447700193805348649, 447700193805348650, 447700193805348651, 447700193805348652, 447700193805348653, 447700193805348654, 447700193805348655, 447700193805348656, 447700193805348657, 447700193805348658, 447700193805348659, 447700193805348660, 447700193805348661, 447700193805348662, 447700193805348663, 447700193805348664, 447700193805348665, 447700193805348666, 447700193805348667, 447700193805348668, 447700193805348669, 447700193805348670, 447700193805348671, 447700193805348672, 447700193805348673, 447700193805348674, 447700193805348675, 447700193805348676, 447700193805348677, 447700193805348678, 447700193805348679, 447700193805348680, 447700193805348681, 447700193805348682, 447700193805348683, 447700193805348684, 447700193805348685, 447700193805348686, 447700193805348687, 447700193805348688, 447700193805348689, 447700193805348690, 447700193805348691, 447700193805348692, 447700193805348693, 447700193805348694, 447700193805348695, 447700193805348696, 447700193805348697, 447700193805348698, 447700193805348699, 447700193805348700, 447700193805348701, 447700193805348702, 447700193805348703, 447700193805348704, 447700193805348705, 447700193805348706, 447700193805348707, 447700193805348708, 447700193805348709, 447700193805348710, 447700193805348711, 447700193805348712, 447700193805348713, 447700193805348714, 447700193805348715, 447700193805348716, 447700193805348717, 447700193805348718, 447700193805348719, 447700193805348720, 447700193805348721]
2024-02-13 10:54:34 [2024-02-13 18:54:34,496: INFO/ForkPoolWorker-1] file file_6bd608 processing completed)
2024-02-13 10:54:34 [2024-02-13 18:54:34,500: INFO/ForkPoolWorker-1] Task app.tasks.execute_asst_file_create[b7c4353e-8954-420b-a276-cd35c4259b47] succeeded in 2462.22321224201s: None

Following is the chat I had:

Create macOS dmg and bundle

Create a macOS dmg and app bundle to install and manage Rubra.

The App is built with Tauri, and replaces the Quickstart script experience.
The app:

Checks for docker and docker-compose commands to be available. If not, it shows a link to the Docker Desktop install page.
sets up ~/.rubra directory and populates a docker-compose.yml, embedEtcd.yaml, llm-config.yaml
Downloads the Model and shows a status bar while downloading. Once complete, the ~/.rubra/rubra.llamafile will be present.
Docker images will use tagged images. The tags will be vX.X.X the same release they are downloaded from.
When starting for the first time, the Start Rubra button is grayed out. Once the model and containers have started there will be a stop rubra and launch UI button.
closing the window will keep the app and model running.
You can use the sys tray icon to get back to the app or launch the UI.
the Command-Q button will quit the App and stop all containers/model from running.

You can verify the model is running by looking for .ape-10 processes in activity monitor or ps. There should only be one.

It would be good if we can let users know when there was a truncation in the response.

Steps to reproduce the problem:

Create a assistant with a log file. I used this log file

When querying for all warning logs , it just provided 2 entries.
test.log

In this case , was the query actually modified to "what is the most recent warning logs in the search results?

For this query it was did mention that it was not able to fetch all data as expected.

Incorrect data being fetched from pdf file for some queries.

Steps to reproduce the problem:

Create an assistant with custom model and provide a pdf file (travel itinerary) that has details of the plan listed for every day of a 12 day trip.
I am able to get my travel itinerary for some days correctly when I ask question like "What is my Day 9 travel plan ?"

The same query does not work as expected when I ask the same question relating to Day 8 . It actually fetched the details of the trip from Day 10. In fact , the query for Day 10 and Day 8 have the exact same results.

Noticed that the same query results in different behaviors when done multiple times.

I created another assistant with custom model and provided the same pdf file .
The same query for Day 8 resulted in fetching trip details of Day 9. In this case , the query for Day 9 and Day 8 have the exact same results.

After couple of hours , I see the same query returning a completely different result

Calling function : FileKnowledge with args: {'query': 'what is my day 8 travel plan'}
2024-02-22 12:53:21 [2024-02-22 20:53:21,364: INFO/ForkPoolWorker-8] HTTP Request: POST http://litellm:8002/v1/chat/completions "HTTP/1.1 200 OK"
2024-02-22 12:53:21 [2024-02-22 20:53:21,365: WARNING/ForkPoolWorker-8] message: The provided search results do not include a day 8 travel plan. However, the last mentioned itinerary is for Day 12.

For comparison , i did the same testing with an assistant with gpt4 model and provided it with the same pdf file (travel itinerary) . In this case the same query returns the correct results.

Dependency Dashboard

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Rate-Limited

These updates are currently rate-limited. Click on a checkbox below to force their creation now.

Open

These updates have all been created already. Click a checkbox below to force a retry/rebase of any.

Detected dependencies

cargo

tauri/src-tauri/Cargo.toml

tauri 1.5

serde 1.0

serde_json 1.0

dirs 3.0

tokio 1.36

tokio-stream 0.1

reqwest 0.11

tauri-build 1.5

docker-compose

docker-compose.yml

milvusdb/milvus v2.3.3

dockerfile

services/backend/api_server/Dockerfile

python 3.10.7-slim

services/backend/task_executor/Dockerfile

python 3.10.7-slim

services/backend/text_embedding_api/Dockerfile

rancher/opni-python-base 3.8-torch

services/backend/vector_db_api/Dockerfile

python 3.8-slim-buster

services/frontend/ui/Dockerfile

python 3.10-slim

python 3.10-slim

github-actions

.github/workflows/docs-release.yaml

actions/checkout v3

.github/workflows/main-release.yaml

actions/checkout v4

docker/setup-qemu-action v3

docker/setup-buildx-action v3

docker/login-action v3

crazy-max/ghaction-github-runtime v3

.github/workflows/push_quickstart.yaml

actions/checkout v3

.github/workflows/tag-release.yaml

actions/checkout v4

docker/setup-qemu-action v3

docker/setup-buildx-action v3

docker/login-action v3

crazy-max/ghaction-github-runtime v3

npm

docs/package.json

@docusaurus/core ^3.1.1

@docusaurus/preset-classic ^3.1.1

@mdx-js/react ^3.0.0

clsx ^2.0.0

docusaurus-preset-openapi ^0.7.2

prism-react-renderer ^2.3.0

react ^18.0.0

react-dom ^18.0.0

url ^0.11.3

@docusaurus/module-type-aliases ^3.1.1

@docusaurus/types ^3.1.1

node >=18.0

tauri/package.json

@tauri-apps/cli ^1.5.8

pip_requirements

services/backend/api_server/requirements.txt

aioredis ==2.0.1

beanie ==1.23.6

celery ==5.3.6

fastapi ==0.105.0

motor ==3.3.2

openai ==1.6.1

pymilvus ==2.3.4

pydantic ==1.10.9

python-multipart ==0.0.6

redis ==5.0.1

uvicorn ==0.25.0

websockets ==12.0

requests ==2.31.0

googlesearch-python ==1.2.3

langchain ==0.0.351

spacy ==3.7.2

markdownify ==0.11.6

playwright ==1.39.0

tiktoken ==0.5.2

services/backend/task_executor/requirements.txt

aioredis ==2.0.1

beanie ==1.23.6

celery ==5.3.6

fastapi ==0.105.0

googlesearch-python ==1.2.3

langchain ==0.0.351

motor ==3.3.2

openai ==1.6.1

pydantic ==1.10.9

python-multipart ==0.0.6

redis ==5.0.1

requests ==2.31.0

uvicorn ==0.25.0

websockets ==12.0

pymilvus ==2.3.4

pypdf2 ==3.0.1

spacy ==3.7.2

markdownify ==0.11.6

playwright ==1.39.0

tiktoken ==0.5.2

chardet ==5.2.0

services/backend/text_embedding_api/requirements.txt

sentence_transformers ==2.2.2

InstructorEmbedding ==1.0.1

uvicorn ==0.24.0

services/backend/vector_db_api/requirements.txt

fastapi ==0.95.1

langchain ==0.0.187

openai ==0.27.7

pymilvus ==2.2.8

python-dotenv ==1.0.0

tiktoken ==0.4.0

uvicorn ==0.21.1

services/frontend/ui/requirements.txt

litellm ==1.18.9

openai ==1.3.7

requests ==2.31.0

streamlit ==1.29.0

streamlit-extras ==0.3.5

streamlit-shadcn-ui ==0.1.18

streamlit-antd-components ==0.2.4

websockets ==12.0

Check this box to trigger a request for Renovate to run again on this repository

Modifying Assistant results in existing files to be deleted. It should also allow to upload more files/delete existing ones.

Steps to reproduce the problem:

Create an assistant with browser option enabled and upload a file.
Modify this assistant and disable browser option .
3.Notice that existing file is not presented in the Update Assistant view.
Once this entry is updated , existing files seem to be lost for the assistant.
Assistant does not have the "Knowledge" tag anymore

Expected Behavior:

Modifying Assistant should not result in existing files to be deleted
Also Modifying Assistant should allow to upload more files/delete existing ones.

UI button to clear / reset the vector database and content files

New model created is not seen in the chat window's model/assistant drop down until manual refresh is done.

Steps to reproduce the problem:

Add a new model from "Settings" tab.
Navigate to "Chat".
Notice that Model/Assistant drop down does not include the newly added model from step 1.

On a manual refresh , we see the new model is available in Model/Assistant drop down in Chat tab.

Expected Behavior:
New model should be made available in Model/Assistant drop down without manual refresh.

On completion of creating an assistant ,create assistant form is still present.

Steps to reproduce the problem:

Create an assistant.
Notice that after successful creation of assistant, create assistant form is still present. At this time , the newly added assistant is already added at the end of the list of assistants.

Expected behavior:
This create form has to self dismiss after successful creation. If we do want to show the success message , then the "Create Assistant" option should not be present and instead we could have a close form button which would dismiss the form and take the user back to the assistant list view with control on the newly created assistant.

Lot of empty lines seen when installing rubra.

Lot of empty lines show up when installing rubra.

curl -sfL https://get.rubra.ai | sh -s -- start
[INFO] Detecting system information...
[INFO] Shell: /bin/zsh
[INFO] Operating System: Darwin
[INFO] Kernel Version: 21.6.0
[INFO] Architecture: arm64
[INFO] Hostname: Sangeethas-MacBook-Pro.local
[INFO] Processor Information:
arm
[INFO] Memory Information:
Pages free:                              300053.
[INFO] Disk Usage Information:
Filesystem       Size   Used  Avail Capacity iused      ifree %iused  Mounted on
/dev/disk3s1s1  926Gi   22Gi  718Gi     3%  502047 4293965346    0%   /
[INFO] Checking for hardware acceleration...
[INFO] Apple Metal supported GPU detected
[INFO] rubra.llamafile already exists, skipping download.
[INFO] Executing rubra.llamafile with -c 16000
[INFO] rubra.llamafile is running with PID: 20810
[INFO] Checking if local model is ready
[INFO] Local model is ready and accepting requests.
[INFO] Downloading docker-compose.yml from https://raw.githubusercontent.com/acorn-io/rubra/main/docker-compose.yml
[INFO] Downloading llm-config.yaml from https://raw.githubusercontent.com/acorn-io/rubra/main/llm-config.yaml
[INFO] Creating embedded etcd config
[INFO] Pulling images and starting Docker containers
[+] Pulling 88/41
 ✔ text-embedding-api 9 layers [⣿⣿⣿⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                  141.0s 
 ✔ redis 6 layers [⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                                   83.7s 
 ✔ standalone 7 layers [⣿⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                            107.9s 
 ✔ mongodb 8 layers [⣿⣿⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                               98.9s 
 ✔ ui 12 layers [⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                               31.2s 
 ✔ litellm 11 layers [⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                           27.7s 
 ✔ task-executor 6 layers [⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                          116.2s 
 ✔ api-server 11 layers [⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                        65.0s 
 ✔ vector-db-api 9 layers [⣿⣿⣿⣿⣿⣿⣿⣿⣿]      0B/0B      Pulled                                                                                                                        30.1s 
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
                                                                                                                                                                                          
[+] Running 14/14
 ✔ Network rubra                                                                                                                                          Created                    0.0s 
 ✔ Volume "rubra_milvus"                                                                                                                                  Created                    0.0s 
 ✔ Volume "rubra_etcd"                                                                                                                                    C...                       0.0s 
 ✔ Volume "rubra_mongodb"                                                                                                                                 Created                    0.0s 
 ✔ Container vector-db-api                                                                                                                                Started                    0.4s 
 ✔ Container text-embedding-api                                                                                                                           Started                    0.4s 
 ✔ Container milvus                                                                                                                                       Star...                    0.4s 
 ✔ Container litellm                                                                                                                                      Sta...                     0.4s 
 ✔ Container mongodb                                                                                                                                      Sta...                     0.4s 
 ✔ Container redis                                                                                                                                        Start...                   0.4s 
 ! litellm The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested                            0.0s 
 ✔ Container task-executor                                                                                                                                Started                    0.0s 
 ✔ Container api-server                                                                                                                                   Started                    0.0s 
 ✔ Container ui                                                                                                                                           Started                    0.0s 
[INFO] Checking if all containers in the 'rubra' network are running...
[INFO] All containers in the 'rubra' network are running.
[INFO] All containers are running. Waiting an additional 15 seconds before proceeding to allow for Rubra backend to load.
[INFO] Rubra started successfully
[INFO] Rubra frontend is now running at http://localhost:8501

Feature Request: LiteLLM for backend abstraction

You could use your current backend as defaults, but also allow for other backend options to be easily configured. The library supports tons of apis and local options.

https://github.com/BerriAI/litellm

rubra-ai / rubra Goto Github PK

rubra's People

Contributors

Stargazers

Watchers

Forkers

rubra's Issues

Rate-Limited

Open

Detected dependencies

Recommend Projects

Recommend Topics

Recommend Org

Jobs