Martin Zlocha

Gheorghe Comanici, Eric Bieber, Mike Schaekermann et al. · amazon-science, baidu

In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal understanding and it is now able to process up to 3 hours of video content. Its unique combination of long context, multimodal and reasoning capabilities can be combined to unlock new agentic workflows. Gemini 2.5 Flash provides excellent reasoning abilities at a fraction of the compute and latency requirements and Gemini 2.0 Flash and Flash-Lite provide high performance at low latency and cost. Taken together, the Gemini 2.X model generation spans the full Pareto frontier of model capability vs cost, allowing users to explore the boundaries of what is possible with complex agentic problem solving.

Martin Zlocha, Qi Dou, Ben Glocker

Accurate, automated lesion detection in Computed Tomography (CT) is an important yet challenging task due to the large variation of lesion types, sizes, locations and appearances. Recent work on CT lesion detection employs two-stage region proposal based methods trained with centroid or bounding-box annotations. We propose a highly accurate and efficient one-stage lesion detector, by re-designing a RetinaNet to meet the particular challenges in medical imaging. Specifically, we optimize the anchor configurations using a differential evolution search algorithm. For training, we leverage the response evaluation criteria in solid tumors (RECIST) annotation which are measured in clinical routine. We incorporate dense masks from weak RECIST labels, obtained automatically using GrabCut, into the training objective, which in combination with other advancements yields new state-of-the-art performance. We evaluate our method on the public DeepLesion benchmark, consisting of 32,735 lesions across the body. Our one-stage detector achieves a sensitivity of 90.77% at 4 false positives per image, significantly outperforming the best reported methods by over 5%.

Harrison Uglow, Martin Zlocha, Szymon Zmyślony

We provide a comprehensive investigation of different custom and off-the-shelf architectures as well as different approaches to generating feature vectors for offensive language detection. We also show that these approaches work well on small and noisy datasets such as on the Offensive Language Identification Dataset (OLID), so it should be possible to use them for other applications.