Which LLM to use for beginners? Mistral, Llama, etc
Hey guys, just want to learn LLM and my friend recommended me to use Mistral or llama. Just wondering what would you recommend for learning the basics and building simple applications like chatbot with proprietary data (e.g. student handbook of a school). TC 160
I love llama's pricing, free.99 is unbeatable.
Some say llama is the most expensive because it’s not free to power in fact it is far far more expensive and the only people who say it is free are incels and trolls.
Well I am a troll, no question about that. It's not expensive for me because I actually know wtf I'm doing. Since the question was for beginners like yourself, I'll concede on this.
OpenAI API. Don't even list bard and chatgpt, you learn nothing by using the chat app. It's like "oh, I want to learn Java, so I start to use a product built in Java" Don't bother with open source model. Hosting the model yourself (let alone get a GPU machine first) doesn't worth the trouble. OpenAI has the most complete API and good documentation. Just pay money to use it.
I mean OP's goal is literally to learn. So "trouble" is exactly what they want. How are the gonna learn by calling an API?
learning llm and building apps based on llms are two different things.
So what would you recommend for each use case? Let's say you want to learn while building an llm app, or each option separately. How would you do it?
Learn fundamentals, foundations, and basics through passive learning platforms. And then start amplifying the skills through project based learning [ building apps, and producing content]
Z
Trump LLM
Neither. All of them are censored, woke, and useless. You want to really learn LLMs there’s the uncensored ones that will teach you to make meth and RDX as well as answer questions about FBI crime stats correctly
Where to find uncensored LLMs ? The “woke” ones are generally fine if they didn’t add all the damn input injection
Yep. Also input injection free Google also censors search results for unwoke models Trivia: unwoke LLMs are much cheaper to run vs woke ones for the same level of capability. Woke LLMs are inefficient, just like woke people