Eureka Labs - The Future of Education?
More topics: Mistral releases two new open-source models, and OpenAI launches a small, cost-efficient model
Hi AI Enthusiasts,
Welcome to this week’s Magic AI News, where we present you the most exciting AI news of the week. A week full of news is behind us, and we have organized the updates for you!
This week’s Magic AI tool is perfect for bloggers and online writers. Imagine transforming your texts into lifelike speech in seconds! Stay curious! 😎
Let’s explore this week’s AI news together. 👇🏽
Top AI news of the week
📚 Eureka Labs: The Future of Education?
Andrej Karpathy, formerly Tesla Director of AI and OpenAI founding member, announced the launch of an AI+Education company called Eureka Labs.
The details:
- Its first AI course, LLM101n, is an undergraduate-level course teaching students to train their own AI.
- They plan digital and physical cohorts of students.
- The platform combines human teachers with an AI teaching assistant to guide the students through the learning materials.
⚡️ Excited to share that I am starting an AI+Education company called Eureka Labs.
— Andrej Karpathy (@karpathy) July 16, 2024
The announcement:
---
We are Eureka Labs and we are building a new kind of school that is AI native.
How can we approach an ideal experience for learning something new? For example, in the case… pic.twitter.com/RHPkqdjB8R
Our thoughts
The future of education is AI-powered personalized learning. The latest breakthroughs in Generative AI are paving the way for this. AI tutors can guide the students through the learning material and answer individual questions about the learning materials.
In addition, the upscaling of the education system gives more people access to education. In our opinion, AI tutors + human teachers are the future of the education system. Great move, Andrej Karpathy!
✍🏽 Would you use such an education platform?
More information
- Eureka Labs Website
👩🏽💻 Mistral releases Codestral Mamba and Mathstral
Mistral launched two new open-source models, called Codestral Mamba and Mathstral. These two new small language models achieve state-of-the-art performance on various benchmarks.
The details of Codestral Mamba:
- 7B parameter model with advanced coding capabilities, ideal for complex coding tasks and local development
- It is based on a Mamba architecture, not a transformer model.
- The model has the advantage of linear time inference and the theoretical ability to model sequences of infinite length.
The details of Mathstral:
- 7B parameter model for math tasks
- State-of-the-art performance on benchmarks MATH (56.6%) and MMLU (63.47%)
- Produced in the context of the collaboration with Project Numina
The models were released under the Apache 2.0 license and are available for free use, modification, and distribution via Mistral’s API and HuggingFace.
Our thoughts
Mistral is a European AI company and can keep up with the big tech companies. As Europeans, we are naturally pleased about this. Mistral is also shaking up the AI landscape with innovative approaches and demonstrating with these specialized models that small models can also be very powerful. We especially like Mistral’s open-source approach!
More information
- Codestral Mamba - Mistral AI Blog
- Mamba: Linear-Time Sequence Modeling with Selective State Spaces - arXiv
- MathΣtral - Mistral AI Blog
🤖 OpenAI launches a small, cost-efficient model
OpenAI launched GPT-4o mini, a small and cost-efficient version of GPT-4o. According to OpenAI, “GPT-4o mini will significantly expand the range of applications built with AI by making intelligence much more affordable”.
The details:
- GPT-4o mini costs: 15 cents per million input tokens and 60 cents per million output tokens, 60% cheaper than GPT-3.5 Turbo
- GPT-4o mini (MMLU: 82%) outperforms Google’s Gemini Flash (MMLU: 77.9%) and Anthropic’s Claude Haiku (MMLU: 73.8%)
- GPT-4o mini replaces GPT-3.5 Turbo in ChatGPT
Our thoughts
The models are becoming cheaper, faster, and more intelligent. But it is irritating that there is still no open-source model! We think this is a mistake, as the open-source community is responsible for many innovations. See Meta’s Llama 3 family!
More information
- GPT-4o mini: advancing cost-efficient intelligence - OpenAI Blog
Magic AI tool of the week
This week’s Magic AI tool is ElevenLabs*. With this tool, you can create realistic speech from texts in seconds!
ElevenLabs is a platform that uses advanced AI to generate realistic speech. And yes, it sounds really realistic! As a blogger, you can turn your texts into audio tracks! You can also offer the audio in different languages. With ElevenLabs, language barriers are a thing of the past!
Step-by-Step Guide:
- Sign up for free at ElevenLabs.com* (10,000 characters per month (~10 min) for free)
- Click on the “Speech” tab to navigate to the Speech Synthesis tool
- Enter your text and select a voice of your choice
- Optional: You can adjust the voice in the settings menu.
- Click “Generate speech” to create your audio file. That’s it! 🎉🎉
Articles of the week
- Mistral’s Codestral: Create a local AI Coding Assistant for VSCode
- Our Journey to Become a Data Scientist
- Build a Local Chatbot in Minutes with Chainlit
- Responsible Development of an LLM Application + Best Practices
💡 Do you enjoy our content and want to read super-detailed articles about AI? If so, subscribe to our blog and get our popular data science cheat sheets for FREE.
Thanks for reading, and see you next time.
- Tinz Twins
P.S. Have a nice weekend! 😉😉
Leave a comment