Weihang Wang

MemeMind recovers from failed rollouts by leveraging offline reference answers, leading to a remarkable 22% boost in performance on complex meme interpretation tasks.

WasmMend achieves a remarkable 70% fix rate for discrepancies between WebAssembly and native binaries, showcasing the power of divergence-guided reasoning in automated repair.

LLMs are poised to flip the script on personalization, giving users unprecedented control over their data and how it's used across platforms.

Code search models still lean heavily on lexical features, failing to grasp code semantics, as revealed by a new C/C++ benchmark that anonymizes identifiers and compiles code to low-level languages.