Search papers, labs, and topics across Lattice.
4
0
7
Mainstream video models consistently falter in fine-grained understanding, revealing critical vulnerabilities in their hallucination capabilities.
Riemannian self-attention can revolutionize EEG decoding by overcoming the limitations of traditional metrics, leading to more accurate brain-computer interfaces.
YOCO++ proves you can halve the KV cache size in LLMs and still beat a standard Transformer, thanks to a clever residual connection trick.
Generative recommendation models can amplify popularity bias due to imbalanced tokenization, but a simple codebook rebalancing strategy can significantly improve performance.
