distributed

AlpinDale reports running GLM-5.2-FP8 on 4 nodes of RTX 4090 (48GB) achieving ~28 tok/s decode over 10Gbit ethernet, with plans to optimize using DSpark.

This paper presents a unified algorithmic framework for distributed online submodular maximization under partition matroid constraints, achieving sublinear (1-1/e)-regret guarantees for both full-information and bandit feedback. It also introduces a bounded stochastic pipage rounding scheme to ensure cumulative sampling violations remain sublinear.

LangChain announces SmithDB, a purpose-built distributed database for agent observability that powers LangSmith, offering up to 12x performance improvements and support for complex agent trace queries.

ActiveMem introduces a distributed active memory system that decouples agent memory from the core LLM reasoning process, achieving state-of-the-art accuracy on long-horizon tasks with significantly reduced overhead.

A programmer used the caches of 3.9 million open DNS resolvers to store file fragments, creating an absurd open-source project called dnsfs. The files disappear as the caches expire.

This paper presents a distributed approach for constrained multi-agent reinforcement learning that uses state-augmented policy learning and neighbor-to-neighbor consensus over dual variables to satisfy global resource constraints while scaling linearly with the number of agents. Experiments on smart grid demand response demonstrate that consensus coordination is essential for feasibility, scaling to thousands of agents unlike centralized training approaches.

MLX's RDMA-over-Thunderbolt implementation for macOS is now available as a standalone library, enabling high-speed Mac clusters for local AI workloads.

Apache Cassandra is a highly-scalable partitioned row store database that automatically distributes data across machines using its Cassandra Query Language (CQL).