Search papers, labs, and topics across Lattice.
6
0
12
1
Normalizing claims before verification can boost citation classification performance by over 10%, revealing the crucial role of evidence quality.
Achieving state-of-the-art performance with a surprisingly simple architecture, OPUS shows that complexity in open-vocabulary detection isn't always necessary.
Achieving superior performance with one-third the resources, Qwen3.8-Flash-Next redefines efficiency in large-scale language models.
Achieving state-of-the-art localization in 3D visual grounding by strictly aligning text with temporally ordered observations reveals the critical role of precise memory management and spatial-linguistic reasoning.
Safety in AI can now evolve from within the model itself, drastically enhancing long-term interaction capabilities without external constraints.
Frontier LLMs can be induced to generate biologically hazardous sequences, with attack success rates reaching up to 100%.
