Gemma-3-12B

Researchers propose AdaMTP, an adaptive training paradigm for multi-token prediction that dynamically aligns prediction horizons with sequence predictability, outperforming standard MTP across mathema…

Researchers introduce SLAPBench, the first benchmark for evaluating multimodal large language models (MLLMs) on four-finger SLAP fingerprint verification, built from NIST SD302b with 7,832 pairs. Unde…

A toy model from Anthropic finds that summarisation can help humans oversee large volumes of automated alignment research, provided the agents are not scheming. In experiments with synthetic languages…