
MLCommons launches MLPerf Endpoints v1.0 for testing live AI systems
MLCommons has launched MLPerf Endpoints v1.0, a benchmark that measures complete AI systems through live API endpoints under production-like loads.
AI news, topic Research. Short posts, checked against the source, each with a link to it.

MLCommons has launched MLPerf Endpoints v1.0, a benchmark that measures complete AI systems through live API endpoints under production-like loads.
Showing 1 to 6 of 6

At CoRL in Austin, Texas, NVIDIA and its partners presented open robotics tools for training robots on precise tasks in simulation.

Researchers Alex M Tseng and Luca Zancato introduced HOPE, a pruning method for Mixture-of-Experts language models that accounts for interactions between experts.

Researchers at MIT, UBC and elsewhere introduced WorldParticle, one AI system that simulates many physical materials, from sand to proteins.

HRL Laboratories, an IBM company, has developed a MEMS-based inertial measurement unit offering an alternative to GPS, with production expected to begin in 2027.

Apple researchers describe small neural networks that learn to speed up the vector search step used in RAG.

Salesforce AI Research says an updated memory policy lets Agentforce Memory recall more customer facts on the live system.