Search papers, labs, and topics across Lattice.
4
0
7
4
A simple triadic outer product expansion of linear attention's state dramatically boosts long-context modeling capacity without a proportional parameter increase.
Nonuniform width allocation in transformers can lead to a 22% reduction in FLOPs while enhancing language modeling performance.
Even the most advanced models fall short, with only 5.6% accuracy on new benchmarks, highlighting a critical gap in semantic understanding in code generation.
Forget RLHF alchemy - this study shows that *what* you teach your LLM *before* RLHF is the real secret to unlocking reasoning abilities.
