Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
Huawei Technologies Co, University of Groningen
5
0
7
Scientific agent swarms can achieve state-of-the-art performance and generalize to new tasks by evolving through feedback-guided mutation and complementary recombination of hypotheses and code.
Training-free feature clustering beats sophisticated token-pruning heuristics in Video LLMs, maintaining competitive downstream accuracy even when discarding 99% of visual tokens.
Task-level natural-language priors can transform low-resource LLM training, boosting performance even with minimal data.
Achieving 94.36% accuracy in gesture recognition with a lightweight model opens new avenues for intuitive human-machine interfaces on portable devices.