Models & Labs
NVIDIA Partners with Ineffable for Reinforcement Learning Infrastructure
NVIDIA Bloghigh confidence
Why it matters
- →This collaboration could redefine AI by enabling systems to learn autonomously from experience.
- →It addresses the challenge of creating AI that discovers new knowledge independently.
- →The partnership leverages cutting-edge NVIDIA technology to support large-scale reinforcement learning.
NVIDIA has announced a collaboration with Ineffable Intelligence, an AI lab founded by AlphaGo architect David Silver, to develop infrastructure for large-scale reinforcement learning. The partnership focuses on creating systems that learn continuously from experience, a step beyond traditional AI models. Utilizing NVIDIA's Grace Blackwell and the upcoming Vera Rubin platform, the project aims to build a pipeline that supports the unique demands of reinforcement learning. This effort could enable AI systems to autonomously discover new knowledge, potentially leading to significant advancements in AI capabilities.
Read originalMore from NVIDIA Blog
© NVIDIA BlogAgentsagents
Hermes Agent Revolutionizes AI with Self-Improvement
Hermes Agent, developed by Nous Research, is making waves in the AI community with its self-improving capabilities and robust performance on NVIDIA hardware. Unlike traditional agents, Hermes can autonomously refine its skills, making it a standout in the field of agentic AI. Its design allows for seamless integration with local systems, leveraging NVIDIA RTX and DGX Spark for optimal performance. This development signifies a shift towards more reliable and efficient AI agents that can operate continuously and improve over time, offering a new level of autonomy and efficiency for developers.
NVIDIA Blog
© NVIDIA BlogMore in Models & Labs
Models & Labsmodels
Llama.cpp b9133 Release Enhances Reasoning Models
The latest b9133 release of llama.cpp introduces significant improvements for reasoning models, particularly in server and web UI environments. By removing the blocking assistant prefill and orchestrating thinking tags, the update ensures smoother continuation of generation tasks. This release also drops the reasoning guard on the Continue button, allowing for persistent reasoning content even after reloads. While the update focuses on templates with simple thinking tags, it sets the stage for future enhancements in reasoning model capabilities.
llama.cpp Releases
Models & Labsmodels
Llama.cpp b9142 Release Expands OpenCL Support
The latest b9142 release of llama.cpp introduces significant updates for OpenCL, particularly enhancing support for Adreno GPUs with the addition of q5_0 and q5_1 Mixture of Experts (MoE) models. This update also addresses potential memory leaks and suppresses warnings for unused variables when building for non-Adreno platforms. These improvements make llama.cpp more robust and versatile, especially for developers working with diverse hardware configurations. The release continues to solidify llama.cpp's position as a flexible inference runtime across multiple operating systems and architectures.
