This article has been translated • Hello, I'm Toshiki, and I'm promoting AI utilization at a company.
I feel that many companies are struggling with AI utilization in their businesses. As the first post, I would like to share the basics.
Since the literacy gap in AI is widening, it's important to get people to understand the basics from the start.
The first session will focus on LLMs (Large Language Models)
The biggest reason AI has evolved is thanks to LLMs, NVIDIA, and scale AI. The founder of scale AI recently said at Meta's Muse launch event Her outfit was so casual that it became a hot topic.
I think many of you have various images of AI, but LLMs are the true brain.
To sum up LLMs (Large Language Models) in one word:
This is a system where you keep guessing the following words.
For example, as shown in the image below, if you input 'Today is a very nice day,' select the next most likely word and generate the word.
This probability changes every time even with the same input. Therefore, LLMs may not generate exactly the same words.
Wait,😱 is that really possible?! Some of you might be thinking, but this system is designed to make AI instantly usable in real life.
Finally, I will explain representative LLMs and conclude.
◾️Claude Series Anthology Claude Mythos 、Claude Fable 、Claude Opus 、Claude Sonnet 、Claude Haiku
◾️GPT-based OpenAI GPT- Astra 、GPT- Sol、GPT- Terra、GPT- Luna
◾️Gemini is Google. Gemini Pro、Gemini Flash
◾️Llama series Meta Llama Scout、Llama Maverick、Meta Muse Spark
◾️Grok-series xAI Grok 、Grok Fast
◾️Kimi Series Moonshot AI Kimi
◾️Qwen Group Alibaba Qwen
◾️Mistral AI Series Mistral 、Codestral
◾️DeepSeek System
DeepSeek