Weekly Magic AI - Apple Wonderlust Event, DeciLM, and Stable Audio
Top AI news of the week and Magic AI book of the week
Hi AI Enthusiasts,
Welcome to our weekly email newsletter, where we bring you the latest updates on Artificial Intelligence (AI) in an accessible way.
Top AI news of the week
🤖 Apple Wonderlust Event
On Sept. 12, 2023, the Wonderlust event took place. Apple did not mention the word “AI” in its announcements. Unlike other tech giants, Apple has been quietly using the technology to improve the fundamentals of its new devices. Apple thus set itself apart from other tech companies that have presented their grand AI visions. However, no one should underestimate Apple. It is known that Apple is developing a ChatGPT competitor. Furthermore, Apple has a lot of market power with billions of customers.
The iPhone 15 has some AI-related improvements in noise cancellation, blurred portraits, and predictive text. In the Apple Watch Series 9, a “Neural Engine” with four cores provides Machine Learning. The Neural Engine increases Siri’s accuracy by 25%.
More information:
🚀 Deci introduces DeciLM-6B
The deep learning community continues to push the limits of large language models (LLMs), exponentially increasing the computational demands of the models for both training and inference. Achieving a balance between model performance, computational efficiency, and latency has become a focus of recent LLM development.
For this reason, Deci has introduced the DeciLM 6B model and DeciLM 6B-Instruct (fine-tuned DeciLM 6B for instruction-based use cases). With 5.7 billion parameters, DeciLM 6B provides 15 times the throughput of Llama 2 7B with comparable quality. Impressively, DeciLM 6B and DeciLM 6B-Instruct are consistently among the best-performing open-source LLMs in the 7 billion parameter category, despite the significantly lower number of parameters in various LLM evaluation tasks. That is a new benchmark for inference efficiency and speed. For more details about DeciLM, we recommend reading Deci’s blog article.
More information:
- 15 times Faster than Llama 2: Introducing DeciLM - Deci Blog
- X Post about DeciLM by thesherrycode - Social Media Platform X
🌄 Stability AI launches Stable Audio
Stable Audio can generate music tracks and other audio content based on prompts and is Stability AI’s first text-to-audio AI. Stability AI is best known for its image-generating AI Stable Diffusion.
Technically, Stable Audio uses very similar techniques to those used for image AI, as it also uses a diffusion model. The AI was trained on audio content instead of images. For this, Stability AI used its own library, AudioSparks. You can find more technical details in the related blog post.
Stable Audio is available for testing on its own website. There is currently a free version available.
More information:
- Stable Audio: Fast Timing-Conditioned Latent Audio Diffusion - Stability AI
Magic AI book of the week
The book “Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems” gives you an intuitive understanding of the concepts and tools through concrete examples, minimal theory and production-ready Python frameworks (Scikit-Learn, Keras and TensorFlow). You will learn how to apply machine learning algorithms. The book supports you in implementing everything from simple linear regression to deep neural networks.
We can recommend this book to newcomers and experts in Machine Learning. The authors explain the individual concepts intuitively, and the practical exercises motivate us to continue learning.
💡 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