Search “Google”

Google Research releases the SLEEP offline memory consolidation method

Google Research releases the SLEEP offline memory consolidation method. According to a report by IT Home · AI, Google's research team has released a continual learning study called the "Sleep Framework," which divides model operation into two phases, awake and sleep, to address the problem of large models having fixed knowledge and difficulty absorbing new information; during the sleep phase, short-term context is written into long-term parameters through parameter expansion, mixture-of-experts memory blocks, and knowledge distillation, and synthetic samples generated through "dreaming" are used for reinforcement learning. Experiments show that the framework can handle contexts of up to about 10 million tokens, that performance degradation is close to zero in multi-stage continual learning on unseen languages, and that it achieves higher accuracy on math benchmarks such as AIME 24/25. Phoenix Technology reported that the study was made public on October 27, and that the paper is titled "Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories." There are currently no subsequent developments.

Google paper proposes engineering task benchmark PLE

Google has published a new paper proposing the PLE benchmark for entity engineering tasks. The benchmark contains 85 electrical, civil, structural, and mechanical engineering tasks written or verified by practicing engineers. According to a report by Rohan Paul, the strongest agent scores only 50.6% on PLE: Claude Opus 5.5 in Claude Code ranks first among 14 models at 50.6%, consuming 27M tokens and 63 minutes per task, requiring reading about 61 project files and delivering schedules, reports, or 2D/3D drawings. The report also states that in design tasks its output satisfies 91% of constraints, but only 41.3% satisfy all hard constraints. There are currently no follow-up developments.

Google Research releases SLEEP offline memory consolidation method

Google Research releases the SLEEP offline memory consolidation method. According to a report by IT Home · AI, Google's research team has released a continual learning study called the "Sleep Framework," which divides model operation into two phases, awake and sleep, to address the problem of large models having fixed knowledge and difficulty absorbing new information; during the sleep phase, short-term context is written into long-term parameters through parameter expansion, mixture-of-experts memory blocks, and knowledge distillation, and synthetic samples generated through "dreaming" are used for reinforcement learning. Experiments show that the framework can handle contexts of up to about 10 million tokens, that performance degradation is close to zero in multi-stage continual learning on unseen languages, and that it achieves higher accuracy on math benchmarks such as AIME 24/25. Phoenix Technology reported that the study was made public on October 27, and that the paper is titled "Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories." There are currently no subsequent developments.

View story:Google Research releases the SLEEP offline memory consolidation method