Launching MM1, Apple plunges into AI scene
The company’s researchers hinted at the MM1 launch to compete with ChatGPT.
Several AI applications have just stirred up the scene with their recent release.
Apple may have been slow to enter the Large Language Model (LLM) game, but the prominent tech corporation has launched MM1 to take on OpenAI and Google, two of the first companies to use AI.
Features of the Apple MM1
The Apple developer team recently published a preprint research article called “MM1: Methods, Analysis & Insights from Multimodal LLM Pretraining,” in which they unveiled Apple MM1. The new AI tool provides a way to train other LLMs more effectively using data and labels supplied by AI. Scientists at Apple think this is a great approach to get cutting-edge data.
Finding out “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” was the exact goal of the study work.
In addition, the study’s authors are confident that Apple MM1 can boost performance and reduce the number of reminders needed to achieve the target outcomes. Apple MM1 is said to have the following capabilities:
- Better in-context learning.
- Multi-image reasoning.
- The ability to provide few-shot chain-of-thought prompting.
With some of these features, this LLM can compete with ChatGPT and Gemini from Google. “While enabling multi-image reasoning and few-shot prompting,” Apple MM1 achieves competitive performance on multiple benchmarks, as confirmed by the researchers themselves.
The Apple Machinist’s mate First Class can easily make in-context predictions because of its large-scale multimodal pretraining. MM1 is a set of artificial intelligence models with about 30 billion parameters. This fraction pales compared to OpenAI’s GPT-4 and Claude 3 Opus, which include a trillion parameters.
The AI Landscape is Growing
Several artificial intelligence models have been released recently, with Apple MM1 being one among them. Many more apps were out there before ChatGPT that were just as competitive.
Cognition Labs, a recently formed artificial intelligence (AI) company, introduced its first AI software engineer, Devin, just recently. Supposedly capable of autonomous coding and AI model fine-tuning, Devin has been vetted by multiple leading AI businesses.
Google Genie, developed by DeepMind, is another option. The new AI program can generate fully playable virtual environments using a picture as a starting point. As the AI landscape grows, more models will undoubtedly be released in the next few months.
