Even though Apple seems like a latecomer to the Large Language Model (LLM) arena, the popular technology company has released MM1 to compete with some of the early AI adopters like OpenAI and Google. 

Apple MM1 Capabilities

Apple MM1 was recently introduced in a preprint research paper titled “MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training” by the Apple developer team. The new AI tool offers a system of utilizing AI-generated data and labels to boost the training of other LLMs. Apple researchers believe that this is a way to retrieve state-of-the-art results. 

Precisely, the research was aimed at demonstrating “how carefully combining different types of training data and model architectures can lead to state-of-the-art performance on a range of AI benchmarks.” 

Furthermore, the research team is convinced that Apple MM1 can speed up performance as well as reduce the volume of follow-up prompts that are required to obtain the desired results. Some of the touted functionalities of Apple MM1 are enhanced in-context learning, multi-image reasoning, and enabling few-shot chain-of-thought prompting.

A few of these capabilities make this LLM a close rival of ChatGPT and Google’s Gemini. Even the researchers confirmed this brings competition by stating that Apple MM1 produces performance that is quite competitive on several benchmarks “while enabling multi-image reasoning and few-shot prompting.”

Due to its large-scale multimodal pretraining, the Apple MM1 can perform in-context predictions conveniently. MM1 looks like a family of AI models with around 30 billion parameters. Compared to OpenAI’s GPT-4 and Claude 3 Opus which have a trillion parameters, this figure is only a small percentage. 

Expansion in the AI Landscape

Notably, Apple MM1 is one of the many AI models that have been released in recent times. Many earlier applications that equally competed with ChatGPT have been launched. 

Newly launched applied AI firm Cognition Labs debuted the first AI software engineer dubbed Devin a few days ago. Devin has been screened by several top AI companies and is said to possess the capacity to perform autonomous coding and also fine-tune its AI models.

There is also Google Genie which was launched by DeepMind. The new AI tool is capable of creating playable virtual worlds from a single image. It is clear that the AI landscape is gradually expanding and more models are bound to show up in the coming months.

The post AI News: Apple Quietly Unveils MM1, New ChatGPT and Gemini Killer appeared first on CoinGape.

---

Filed under: News - @ January 1, 1970 12:00 am