Search papers, labs, and topics across Lattice.
Shanghai Jiao Tong University
6
0
10
Stale rollouts can introduce a significant bias in learning rates, fundamentally altering the stability landscape of asynchronous RLHF systems.
OmniCoT recalibrates the challenges of panoramic reasoning, enabling MLLMs to leverage global evidence for complex multi-step inference.
BabelTele achieves 99.5% semantic fidelity while reducing text volume to just 27.9%, challenging the necessity of human-readable language for LLMs.
Coding agents struggle to create complete and engaging games, with top performers barely reaching 41.46% success in end-to-end game generation.
The new REO framework reveals that the true challenge in differential equation discovery lies not just in recovering equations, but in leveraging them to reshape scientific understanding.
Leveraging historical solving traces transforms software engineering agents into self-evolving entities, achieving a 50.40% success rate on SWE-bench Verified after just three iterations.
