😰 Fear of SQL Coding? NEW LLM beats ChatGPT 3.5 in SQL!!!

Defog's SQLCoder is a state-of-the-art LLM for converting natural language questions to SQL queries.

❤️ If you want to support the channel ❤️
Support here:

[ewldcyd]

Awesome video! I love learning about these specialized LLMs, thank you!

[Nick_With_A_Stick]

What a legend that license is what I’m gonna start using for all my projects.

[yusefess]

LLMs working in a specialized field have a lot of room for improvement. For example, I would really like to see LLMs in the coming months that, for example, work only for JavaScript backend.

[moondevonyt]

dope breakdown of the SQL coder and its capabilities!

but honestly, i feel like relying too heavily on AI for SQL might make us lose touch with the nuances of crafting efficient queries

it's dope for simplification, but ain't nothing like getting hands-on

still, major kudos to the creator for presenting it in such a clear manner

respect!

[Crash_Nexus]

Каждый раз, когда я смотрю это видео, я получаю удовольствие!

[tyjjvru]

Видео потрясающее! Мне очень понравилось!

[kshitizkhandelwal]

The context length supported is 2048. How do we fit larger schemas?

[DILIPKUMAR-dnnu]

Should we provide the schema everytime or we can store it like embeddings and reuse it with new schema?

[tyjjvru]

Автору респект за такой оригинальный контент!

[tyjjvru]

Видео замечательное! Мне очень понравилось!

[tyjjvru]

Автору признание за такой качественный контент!

[gw]

Does SQLCoder-70B work locally on the computer or does it have to communicate with an online server and send your chats to that server?

[TommyJefferson]

There is a problem with the evals. In real life, the sql queries are much more complex than simple order by and joins

[suren]

Does it reduce dependency of SQL coding developers

[tyjjvru]

Автору признание за такой продуманный контент!

[thexdev]

I’ve been testing these use cases recently and wizard coder vastly outperforms sqlcoder

[Xeronimo]

Why compare it to v3.5?? v4 is way better

[karthickr]

Hi, I tried to run the same in Google colab, but it got crashed due to ram issue

[Crash_Nexus]

The context length supported is 2048. How do we fit larger schemas?

[Crash_Nexus]

LLMs working in a specialized field have a lot of room for improvement. For example, I would really like to see LLMs in the coming months that, for example, work only for JavaScript backend.