4 papers
The Last Mile to the Real World: Agent-Native Primitives as Reusable Building Blocks for Solving Real-World Tasks
Haibo Jin, Peng Kuang, Xiaopeng Yuan +11
How can an agent handle a real-world task end to end? Frontier models with their own harnesses still fall short on real-world benchmarks: on Terminal-Bench 4.0, GPT-5.6 Sol with Co…
How Much Does Random Rotation Help GPTQ? Sharp Worst-Case Rate Gaps in the High-Rate Regime
Runguo Li
GPTQ is Babai's nearest-plane algorithm on the lattice of the Cholesky factor of the layer's input second moment . In the high-rate, entropy-coded regime it needs $\tfrac12\log_…
Recency Is the Wrong Axis: A Negative Result on Windowed Deepest-Plane KV Caching in Looped Language Models
Dezhen Lu, Runguo Li, Zixiang Ni +4
Looped language models reuse one block of layers over recurrent steps but store a key-value (KV) plane for each step. Keeping only the deepest plane shrinks the cache up to …
ScaleRLT: Exact Log-Probabilities and Fast Deferred Rollouts for Reinforcement Learning on Looped Transformers
Dezhen Lu, Zixiang Ni, Runguo Li
Looped language models such as Ouro reuse one block of layers several times per token. A rollout engine can batch the deeper loops of several response positions and run them only w…