As demand for artificial intelligence technology boosts construction and proposed construction of data centers around the world, those computers require not just electricity and land, but also a significant amount of water. Data centers use water directly, with cooling water pumped through pipes in and around the computer equipment. They also use water indirectly, through the water required to produce the electricity to power the facility. The amount of water used to produce electricity increases dramatically when the source is fossil fuels compared with solar or wind. A 2024 report from the Lawrence Berkeley National Laboratory estimated that in 2023, U.S. data centers consumed 17 billion gallons (64 billion liters) of water directly through cooling, and projects that by 2028, those figures could double - or even quadruple.

We present OLMo 2, the next generation of our fully open language models. OLMo 2 includes dense autoregressive models with improved architecture and training recipe, pretraining data mixtures, and instruction tuning recipes. Our modified model architecture and training recipe achieve both better training stability and improved per-token efficiency. Our updated pretraining data mixture introduces a new, specialized data mix called Dolmino Mix 1124, which significantly improves model capabilities across many downstream task benchmarks when introduced via late-stage curriculum training (i.e. specialized data during the annealing phase of pretraining). Finally, we incorporate best practices from T\"ulu 3 to develop OLMo 2-Instruct, focusing on permissive data and extending our final-stage reinforcement learning with verifiable rewards (RLVR). Our OLMo 2 base models sit at the Pareto frontier of performance to compute, often matching or outperforming open-weight only models like Llama 3.1 and Qwen 2.5 while using fewer FLOPs and with fully transparent training data, code, and recipe. Our fully open OLMo 2-Instruct models are competitive with or surpassing open-weight only models of comparable size, including Qwen 2.5, Llama 3.1 and Gemma 2. We release all OLMo 2 artifacts openly -- models at 7B and 13B scales, both pretrained and post-trained, including their full training data, training code and recipes, training logs and thousands of intermediate checkpoints. The final instruction model is available on the Ai2 Playground as a free research demo.

Building height is an important indicator for scientific research and practical application. However, building height products with a high spatial resolution (10m) are still very scarce. To meet the needs of high-resolution building height estimation models, this study established a set of spatial-spectral-temporal feature databases, combining SAR data provided by Sentinel-1, optical data provided by Sentinel-2, and shape data provided by building footprints. The statistical indicators on the time scale are extracted to form a rich database of 160 features. This study combined with permutation feature importance, Shapley Additive Explanations, and Random Forest variable importance, and the final stable features are obtained through an expert scoring system. This study took 12 large, medium, and small cities in the United States as the training data. It used moving windows to aggregate the pixels to solve the impact of SAR image displacement and building shadows. This study built a building height model based on a random forest model and compared three model ensemble methods of bagging, boosting, and stacking. To evaluate the accuracy of the prediction results, this study collected Lidar data in the test area, and the evaluation results showed that its R-Square reached 0.78, which can prove that the building height can be obtained effectively. The fast production of high-resolution building height data can support large-scale scientific research and application in many fields.

Reports this week of Russian intrusions into U.S. election systems have startled many voters, but computer experts are not surprised. They have long warned that Americans vote in a way that's so insecure that hackers could change the outcome of races at the local, state and even national level. Multibillion-dollar investments in better election technology after the troubled 2000 presidential election count prompted widespread abandonment of flawed paper-based systems, such as punch ballots. But the rush to embrace electronic voting technology -- and leave old-fashioned paper tallies behind -- created new sets of vulnerabilities that have taken years to fix. "There are computers used in all points of the election process, and they can all be hacked," said Princeton computer scientist Andrew Appel, an expert in voting technologies.