A report on Apple’s big model has also been released, revealing a large number of technical details. The report shows that Apple’s cloud model has achieved better results than GPT-4 in tasks such as instruction compliance and text summary.
Apple conducted an IFEval test, and the results showed that at both command and prompt levels, the cloud-side AFM surpassed GPT-4 and became the new SOTA. The performance of the end-side model also exceeds that of similar scale models such as Llama 3-8B and Mistral-7B. Among AlpacaEval, both end-side and cloud-side AFM also achieved second place.
The training of the model is carried out through its own JAX-based AXLearn framework, and strategies such as tensor parallelism and pipeline parallelism are adopted.
The data mainly comes from web pages crawled through Applebot, as well as publicly licensed code and mathematical datasets. It is worth mentioning that none of the datasets selected by Apple uses the GPL.
If you want to learn more, you can click on the link below the video.
Thank you for watching this video. If you like it, please subscribe and like it. thank
Original text:https://machinelearning.apple.com/research/apple-intelligence-foundation-language-models
Oil tubing: