Models & Labs
Google Enhances On-Device AI with Multi-Token Prediction
Google Research Bloghigh confidence
Why it matters
- →MTP enhances on-device AI efficiency, crucial for mobile applications.
- →It reduces memory usage and energy consumption, improving user experience.
- →The approach allows faster AI features without compromising model capabilities.
Google has unveiled a new architecture to improve the efficiency of on-device AI models, focusing on the Gemini Nano models used in Pixel devices. The method involves retrofitting Multi-Token Prediction (MTP) onto existing models, enhancing speed and reducing energy consumption without requiring separate drafting models. This advancement allows features like AI Notification Summaries to function more efficiently on mobile devices. The integration of MTP reduces memory usage and improves battery life, representing a significant advancement in mobile AI technology.
Read originalMore from Google Research Blog
More in Models & Labs
Models & Labsmodels
llama.cpp b10955 release addresses heap corruption
The latest release of llama.cpp, b10955, tackles a critical issue of heap corruption by disabling the ggml-cpu precompiled header and fixing CACHE_LINE_SIZE ambiguity. This update ensures consistent CACHE_LINE_SIZE values across C++ kernels and C work-buffer sizing code, preventing heap-buffer-overflow and subsequent crashes. By restoring the natural include order and removing the std::hardware_destructive_interference_size branch, the update makes the value deterministic and include-order independent. This release is a technical fix that stabilizes the runtime environment for developers using llama.cpp.
llama.cpp Releases
Models & Labsmodels
llama.cpp b10956 release enhances SYCL backend
The latest llama.cpp release, b10956, introduces significant improvements to the SYCL backend, particularly for handling large k values in TOP_K operations. By implementing a radix select method, the update allows for efficient GPU-resident processing, avoiding previous limitations that forced operations to fall back to the CPU. This change enhances performance, especially in scenarios requiring large k values, such as qwen4exp's sparse-attention indexer. The update ensures that operations are more efficient and scalable, providing a notable boost in processing speed without regressing any measured shapes.