Researchers Discover Powerful Tool for Local Machine Learning Workflows with pmetal Framework
apple fine-tuning inference meta training
| Source: Mastodon | Original article
Researchers unveil pmetal, a local machine learning framework. It enables LLM fine-tuning on terminals.
Researchers have discovered a terminal-based user interface (TUI) for local machine learning workflows, dubbed **pmetal**. This framework allows for LLM fine-tuning, leveraging Metal kernels, Apple Neural Engine (ANE) support, LoRA training, inference, and quantization. Written in Rust and built with @ratatui_rs, **pmetal** offers a comprehensive machine learning platform for Apple Silicon devices.
This breakthrough matters because it enables developers to work with machine learning models locally, without relying on cloud services or external providers. As we reported earlier, the shutdown of Anthropic's OAuth service highlighted the risks of dependence on external LLM providers. **pmetal**'s local approach mitigates such risks, providing a more sustainable and efficient solution for machine learning workflows.
As **pmetal** continues to evolve, it will be interesting to watch how it integrates with other local machine learning tools, such as LM Studio, which recently shipped with Apple MLX support. The combination of these technologies could revolutionize local ML workflows, enabling faster and more efficient development of on-device LLMs. With **pmetal**'s open-source nature and active development, we can expect to see significant advancements in the field of local machine learning.
Sources
Back to AIPULSEN