13 papers
GDPevo: Evaluating Agent Self-Evolution on Real Business Tasks
Leijun Zhou, Zhihao Liu, Xiang Qu +9
Agent self-evolution updates an agent's persistent state from prior experience and reuses it to solve related tasks more effectively. Evaluating self-evolution is difficult: existi…
When Does Recurrence Become an Algorithm? Convergence Selection in Weight-Tied Looped Transformers
Tong Zhang, Junhao Hu, Yun Peng +1
When does a weight-tied looped transformer -- one block applied T times -- implement an actual algorithm? We answer with four findings from controlled populations on group word pro…
HYPIC: Accelerating Hybrid-Attention LLM Serving with Position-Independent Caching
Yifei Liu, Juntong Wu, Yang Liu +4
In retrieval-augmented generation and agentic LLM serving, prompts are assembled from independent segments into long contexts, making the prefill stage dominate per-request cost. T…
Akashic: A Low-Overhead LLM Inference Service with MemAttention
Yang Liu, Zhaokai Luo, Huayi Jin +7
Recent LLM-based agent systems continuously accumulate context across multi-turn interactions, tool invocations, and cross-session workflows. Replaying the full history for every r…
RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention
Yang Liu, ZhaoKai Luo, Zhaokai Luo +11
As the input length of large language model (LLM) serving continues to grow, the KV cache has become a dominant bottleneck in AI infrastructure. It limits GPU memory capacity, serv…
DynaTrain: Fast Online Parallelism Switching for Elastic LLM Training
Yuanqing Wang, Yuchen Zhang, Hao Lin +9
Modern large language model (LLM) training is inherently dynamic: resource fluctuations, RLHF phase shifts, and cluster elasticity continually reshape the optimal parallelism layou…