AI Revolution: Apple Opens Source OpenELM for Devices
Apple has introduced its own family of open artificial intelligence language models, OpenELM, designed to run on local devices without cloud connections.
OpenELM from Apple
Introducing OpenELM
OpenELM is a series of compact language models from Apple aimed at efficiently creating texts. They are distinguished by their ability to function directly on ordinary gadgets such as laptops and smartphones without the need to access remote servers in the cloud. The models were pre-trained on public data from Wikipedia, Reddit, arXiv.org and other sources, amounting to 1.8 trillion tokens.
Technical characteristics
The OpenELM line includes 8 models, divided into pre-trained and customized variations instructions. Their sizes vary from 270 million to 3 billion parameters. OpenELM testing was carried out on a workstation with a powerful Intel processor, 64 GB of RAM and an NVIDIA RTX 4090 graphics card, as well as on an Apple MacBook Pro laptop with an M2 Max chip and 64 GB of RAM running macOS 14.4.1.
Open model without guarantees
Apple distributes its OpenELM models under a code sample license, allowing their commercial use and modification, provided the original notice and text are preserved. Training benchmarks, performance data, and parameter tuning instructions are provided with the models. The company warns that the models are not guaranteed to be secure and may generate inaccurate, harmful or biased results.
Glossary
- Apple is an American corporation, manufacturer of personal and tablet computers, audio players, smartphones, and software.
- OpenELM is a family of open source language models from Apple designed to run locally on devices.
- LLM (Large Language Model) is a large artificial intelligence language model trained on massive text data.
- Intel is an American company, the largest manufacturer of processors and other electronics.
- NVIDIA is an American company, a developer of graphics processors and systems on a chip.
Links
- Hugging Face hub with OpenELM models
- OpenELM release tweet
- OpenELM model collection on Hugging Face
- OpenELM release article on VentureBeat
FAQ
What is OpenELM from Apple?
What parameter sizes do OpenELM models have?
What devices can OpenELM models run on?
What licensing terms does Apple offer for OpenELM models?
Are there any warnings from Apple regarding the use of OpenELM models?
Hashtags
Save a link to this article
Discussion of the topic – AI Revolution: Apple Opens Source OpenELM for Devices
For the first time, Apple is open-sourcing its OpenELM artificial intelligence models, designed to run on devices without connecting to cloud servers. This family of large, open-source language models aims to empower and strengthen open AI research.
Latest comments
8 comments
Write a comment
Your email address will not be published. Required fields are checked *
Григорий
This is exciting news from Apple! They've never released an open source AI model before. I'm curious to see how effective their OpenELM models will be on devices. 🤔
Франческа
Yes, this is a great initiative! Finally, we will have high-quality AI models to run on smartphones and laptops without the need to connect to the cloud. It is very convenient and energy efficient. 💻✨
Райнер
I don't think Apple's models will be as productive as cloud solutions. They have limited resources on their devices. But for simple tasks like writing text, it can work. 🧐
Марыся
And I like that Apple provides open access to its models and instructions. This will allow the community to experiment and develop the technology further. A good step towards open science! 🙌
Джованни
Phew, why is all this necessary? Artificial intelligence will ruin our lives! 😠 Better to go back to the good old days without all this fancy stuff.
Анна
Giovanni, don't be such a grump! 😄 Technologies open up new opportunities. Imagine how convenient it will be to dictate messages or letters to your phone using AI.
Франческа
I agree with Anna. AI models on the device can also help with photo and video editing, text translation and much more without transferring data to the cloud. This is good for privacy. 🔒
Григорий
By the way, I recently tried one of Apple's models on my MacBook and was quite impressed. She was able to turn my draft article into an interesting and coherent text in a couple of seconds. I was surprised by such speed of work. 😎