2 papers
cs.LG2026
GDN Tree-Scan: Served Tree Verification for Recurrent-Hybrid Language Models
Zhiyuan Ma
Tree speculative decoding verifies multiple candidate continuations in one target forward pass. For attention-only transformers, the verifier mainly needs an ancestry mask. Recurre…
cs.AI2026
CADWorld: Computer-Use Benchmark for Long-Horizon Computer-Aided Design
Zihan Dong, Yuanzhe Liu, Zhiyuan Ma +4
Computer-use agents are increasingly evaluated in realistic desktop environments, but existing benchmarks provide limited coverage of professional engineering workflows whose outpu…