Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
4
0
6
0
Training-free LQEC recovers 84% of the FP16 accuracy gap at 2-bit quantization by decoupling calibration and compensation, outperforming baselines by over 20%.
Long-horizon reasoning traces repeatedly backtrack to early planning steps via predictable query clusters, revealing that a handful of representative "beacon" vectors can guide 5.8× KV cache compression with near-zero quality loss.
Diminishing returns in inference-time scaling reveal that more computation doesn't always equate to better performance in local computer-use agents.
Tangram transforms multi-turn LLM serving by achieving up to 2.6 times throughput improvement without sacrificing accuracy, all while sidestepping the pitfalls of memory fragmentation.