torrmal

what is the goal of these tools, having a hard timeunderstanding the value vs just running them on terminal

Open-source, very smart and It will make you think that Glean, ThoughtSpot, and practically any tool you've ever used to "answer questions from data to get your job done" look like fax-machines at an antique shop.

we have been trying to make it so that people dont have to reinvent the wheel, over and over and over again, and have a very straight forward all batteries included that can scale to many millions of documents, combining the best of RAG with traditional search and parametric search, https://docs.mindsdb.com/mindsdb_sql/knowledge_bases/overvie... Would love your feedback.

1000% we think it can be very valuable, and also it can add a layer of security, have you been able to try it?

Please share what you think about this! Your feedback means the world to us

At Mindsdb we have been using it, and I truly believe it’s the future of open source code management.

Hey HN,

So, casually trying to make LLaMA achieve consciousness (as one does on a Tuesday), when I stumbled upon something hilarious. Turns out, you can make these language models "reason" with about as much code as it takes to write a "Hello World" program. No, really!

https://colab.research.google.com/drive/1jfsG0_XP8a5mME76F5a...

Here's the philosophical crisis I'm having now: When you ask an LLM to code something vs. asking it to reason about something... are we basically watching the same neural spaghetti being twirled around?

The real question is: If one can make an AI model "think" with 5 lines of code, does this mean: a) An LLM should be able to write its own reasoning code b) We've been overthinking AI? c) The simulation is running low on RAM d) All of the above

Would love to hear your thoughts, preferably in the form of recursive functions or philosophical paradoxes.

P.S. No LLAMAs were harmed in the making of this experiment, though several did ask for a raise.

great question, i think rather than smaller inputs, you should be able to have as much context as needed in. Its more aboue smaller outputs, that are well typed and allow you to control workflow via standard coding mechanics. Therefore getting the best of both worlds, llms power of understanding unstructured info (which python sucks at) and being able to control things which python is great at.

Hey — We built a text-to-SQL skill, connected it to an agent, and in the end, you can publish a Chatbot to Slack.

This is useful because it lets non-technical people ask questions of a database and get the answers they need.

Please check out our demo above and let us know what you think.

A MindsDB engineers demonstrates how to create a Text to SQL skill, connect it to an agent, and publish a chatbot to Slack.

Learn how to make your own Midjourney bot with a few simple SQL commands. This tutorial uses a Twitter bot, but you can use texts, Slack, Rocket.chat or email.

Are you excited for Hacktoberfest 2023? While this year’s event may offer only digital rewards, we’ve got something extraordinary in store for you! Join us at MindsDB for the most thrilling open-source contribution journey, where you can unlock a world of exciting prizes.

actually, yes they do have support for Llama2 70B, you need 39GB of ram, compared to 130GB that it originally takes.

well, I am not sure how small you can turn your 70b model into a small one, but its a great question for the Ollama creators: https://lu.ma/yourdatayourmodel

this is pretty cool, can I connect to it with any Postgres client?

yes!! we were thinking about this too! we have made some massive progress on timeseries models and lowering latency, lets chat soon, I'll connect over email!

Hi Omneity, thanks for asking. You can deploy Mindsdb on a container, no application that you build querying the Mindsdb server has licensing dependencies, as its essentially no different from using most opensource databases out there.

absolutely lets connect!! jorge at mindsdb

These are amazing questions Streetcat, We do some hyperparameter search using Optuna, we may be moving to Ray Tune because it can be highly parallelized. If MindsDB fails, it depends on how various DBs manage federated storage, but essentially you will get a query error. Funny that you mention automatic retraining, people have been asking for this recently, we will be supporting a retrain_frequency parameter in the coming releases, would you like to give it a test drive?

this is a great question, I actually think that if your language is also R, MSsql r integration is a great option, what we bring to the table for MSSQL users in particular is more options as well as better performance for some types of problems like high cardinality on time-series, for example: predicting inventory for all products in a database taking into account all previous inventory as well as say marketing data, building this in the R bindings would be quite a challenge, with mindsdb its a simple SQL statement

Thanks for asking! We develop based on community requests, if you want we can setup a call, and we can see if we can help you with your data on DynamoDB/SCylia, its becoming pretty fast now to support new databases. send us an email to adam jorge at mindsdb.com

you are right!! the main thing is to offer it for other databases (mysql, mariadb, postgres, clickhouse, clickhouse, timescale, mongodb) as well as to support more powerful machine learning capabilities than the vanila classical models supported by oracle, for instance great timeseries support

thank you!! lets chat, we would love to show you the timeseries cool stuff we have done for clickhouse!

Also worth mentioning, Mindsdb can take input columns of any of the following (numerical, categorical, text, images) and it's getting pretty good at Timeseries problems (for which we support a variety of techniques, including novel approaches to sequential data such as (RNNs, Transformers, CNN tiling, ...). Given the nature of data in databases where there is often a chronological order of transactions we put allot of focus on offer capabilities to make the models time aware.